diff options
Diffstat (limited to 'test/Analysis/CostModel')
| -rw-r--r-- | test/Analysis/CostModel/AArch64/free-widening-casts.ll | 622 | ||||
| -rw-r--r-- | test/Analysis/CostModel/AMDGPU/extractelement.ll | 74 | ||||
| -rw-r--r-- | test/Analysis/CostModel/AMDGPU/insertelement.ll | 43 | ||||
| -rw-r--r-- | test/Analysis/CostModel/AMDGPU/shufflevector.ll | 43 | ||||
| -rw-r--r-- | test/Analysis/CostModel/X86/div.ll | 32 | ||||
| -rw-r--r-- | test/Analysis/CostModel/X86/vshift-ashr-cost.ll | 138 | ||||
| -rw-r--r-- | test/Analysis/CostModel/X86/vshift-lshr-cost.ll | 128 | ||||
| -rw-r--r-- | test/Analysis/CostModel/X86/vshift-shl-cost.ll | 134 |
8 files changed, 957 insertions, 257 deletions
diff --git a/test/Analysis/CostModel/AArch64/free-widening-casts.ll b/test/Analysis/CostModel/AArch64/free-widening-casts.ll new file mode 100644 index 0000000000000..07f32d1d8ba21 --- /dev/null +++ b/test/Analysis/CostModel/AArch64/free-widening-casts.ll @@ -0,0 +1,622 @@ +; RUN: opt < %s -mtriple=aarch64--linux-gnu -cost-model -analyze | FileCheck %s --check-prefix=COST +; RUN: llc < %s -mtriple=aarch64--linux-gnu | FileCheck %s --check-prefix=CODE + +; COST-LABEL: uaddl_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i8> %a to <8 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <8 x i8> %b to <8 x i16> +; CODE-LABEL: uaddl_8h +; CODE: uaddl v0.8h, v0.8b, v1.8b +define <8 x i16> @uaddl_8h(<8 x i8> %a, <8 x i8> %b) { + %tmp0 = zext <8 x i8> %a to <8 x i16> + %tmp1 = zext <8 x i8> %b to <8 x i16> + %tmp2 = add <8 x i16> %tmp0, %tmp1 + ret <8 x i16> %tmp2 +} + +; COST-LABEL: uaddl_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i16> %a to <4 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <4 x i16> %b to <4 x i32> +; CODE-LABEL: uaddl_4s +; CODE: uaddl v0.4s, v0.4h, v1.4h +define <4 x i32> @uaddl_4s(<4 x i16> %a, <4 x i16> %b) { + %tmp0 = zext <4 x i16> %a to <4 x i32> + %tmp1 = zext <4 x i16> %b to <4 x i32> + %tmp2 = add <4 x i32> %tmp0, %tmp1 + ret <4 x i32> %tmp2 +} + +; COST-LABEL: uaddl_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <2 x i32> %a to <2 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <2 x i32> %b to <2 x i64> +; CODE-LABEL: uaddl_2d +; CODE: uaddl v0.2d, v0.2s, v1.2s +define <2 x i64> @uaddl_2d(<2 x i32> %a, <2 x i32> %b) { + %tmp0 = zext <2 x i32> %a to <2 x i64> + %tmp1 = zext <2 x i32> %b to <2 x i64> + %tmp2 = add <2 x i64> %tmp0, %tmp1 + ret <2 x i64> %tmp2 +} + +; COST-LABEL: uaddl2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <16 x i8> %a to <16 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <16 x i8> %b to <16 x i16> +; CODE-LABEL: uaddl2_8h +; CODE: uaddl2 v2.8h, v0.16b, v1.16b +; CODE-NEXT: uaddl v0.8h, v0.8b, v1.8b +define <16 x i16> @uaddl2_8h(<16 x i8> %a, <16 x i8> %b) { + %tmp0 = zext <16 x i8> %a to <16 x i16> + %tmp1 = zext <16 x i8> %b to <16 x i16> + %tmp2 = add <16 x i16> %tmp0, %tmp1 + ret <16 x i16> %tmp2 +} + +; COST-LABEL: uaddl2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i16> %a to <8 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <8 x i16> %b to <8 x i32> +; CODE-LABEL: uaddl2_4s +; CODE: uaddl2 v2.4s, v0.8h, v1.8h +; CODE-NEXT: uaddl v0.4s, v0.4h, v1.4h +define <8 x i32> @uaddl2_4s(<8 x i16> %a, <8 x i16> %b) { + %tmp0 = zext <8 x i16> %a to <8 x i32> + %tmp1 = zext <8 x i16> %b to <8 x i32> + %tmp2 = add <8 x i32> %tmp0, %tmp1 + ret <8 x i32> %tmp2 +} + +; COST-LABEL: uaddl2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i32> %a to <4 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <4 x i32> %b to <4 x i64> +; CODE-LABEL: uaddl2_2d +; CODE: uaddl2 v2.2d, v0.4s, v1.4s +; CODE-NEXT: uaddl v0.2d, v0.2s, v1.2s +define <4 x i64> @uaddl2_2d(<4 x i32> %a, <4 x i32> %b) { + %tmp0 = zext <4 x i32> %a to <4 x i64> + %tmp1 = zext <4 x i32> %b to <4 x i64> + %tmp2 = add <4 x i64> %tmp0, %tmp1 + ret <4 x i64> %tmp2 +} + +; COST-LABEL: saddl_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i8> %a to <8 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <8 x i8> %b to <8 x i16> +; CODE-LABEL: saddl_8h +; CODE: saddl v0.8h, v0.8b, v1.8b +define <8 x i16> @saddl_8h(<8 x i8> %a, <8 x i8> %b) { + %tmp0 = sext <8 x i8> %a to <8 x i16> + %tmp1 = sext <8 x i8> %b to <8 x i16> + %tmp2 = add <8 x i16> %tmp0, %tmp1 + ret <8 x i16> %tmp2 +} + +; COST-LABEL: saddl_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i16> %a to <4 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <4 x i16> %b to <4 x i32> +; CODE-LABEL: saddl_4s +; CODE: saddl v0.4s, v0.4h, v1.4h +define <4 x i32> @saddl_4s(<4 x i16> %a, <4 x i16> %b) { + %tmp0 = sext <4 x i16> %a to <4 x i32> + %tmp1 = sext <4 x i16> %b to <4 x i32> + %tmp2 = add <4 x i32> %tmp0, %tmp1 + ret <4 x i32> %tmp2 +} + +; COST-LABEL: saddl_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <2 x i32> %a to <2 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <2 x i32> %b to <2 x i64> +; CODE-LABEL: saddl_2d +; CODE: saddl v0.2d, v0.2s, v1.2s +define <2 x i64> @saddl_2d(<2 x i32> %a, <2 x i32> %b) { + %tmp0 = sext <2 x i32> %a to <2 x i64> + %tmp1 = sext <2 x i32> %b to <2 x i64> + %tmp2 = add <2 x i64> %tmp0, %tmp1 + ret <2 x i64> %tmp2 +} + +; COST-LABEL: saddl2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <16 x i8> %a to <16 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <16 x i8> %b to <16 x i16> +; CODE-LABEL: saddl2_8h +; CODE: saddl2 v2.8h, v0.16b, v1.16b +; CODE-NEXT: saddl v0.8h, v0.8b, v1.8b +define <16 x i16> @saddl2_8h(<16 x i8> %a, <16 x i8> %b) { + %tmp0 = sext <16 x i8> %a to <16 x i16> + %tmp1 = sext <16 x i8> %b to <16 x i16> + %tmp2 = add <16 x i16> %tmp0, %tmp1 + ret <16 x i16> %tmp2 +} + +; COST-LABEL: saddl2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i16> %a to <8 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <8 x i16> %b to <8 x i32> +; CODE-LABEL: saddl2_4s +; CODE: saddl2 v2.4s, v0.8h, v1.8h +; CODE-NEXT: saddl v0.4s, v0.4h, v1.4h +define <8 x i32> @saddl2_4s(<8 x i16> %a, <8 x i16> %b) { + %tmp0 = sext <8 x i16> %a to <8 x i32> + %tmp1 = sext <8 x i16> %b to <8 x i32> + %tmp2 = add <8 x i32> %tmp0, %tmp1 + ret <8 x i32> %tmp2 +} + +; COST-LABEL: saddl2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i32> %a to <4 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <4 x i32> %b to <4 x i64> +; CODE-LABEL: saddl2_2d +; CODE: saddl2 v2.2d, v0.4s, v1.4s +; CODE-NEXT: saddl v0.2d, v0.2s, v1.2s +define <4 x i64> @saddl2_2d(<4 x i32> %a, <4 x i32> %b) { + %tmp0 = sext <4 x i32> %a to <4 x i64> + %tmp1 = sext <4 x i32> %b to <4 x i64> + %tmp2 = add <4 x i64> %tmp0, %tmp1 + ret <4 x i64> %tmp2 +} + +; COST-LABEL: usubl_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i8> %a to <8 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <8 x i8> %b to <8 x i16> +; CODE-LABEL: usubl_8h +; CODE: usubl v0.8h, v0.8b, v1.8b +define <8 x i16> @usubl_8h(<8 x i8> %a, <8 x i8> %b) { + %tmp0 = zext <8 x i8> %a to <8 x i16> + %tmp1 = zext <8 x i8> %b to <8 x i16> + %tmp2 = sub <8 x i16> %tmp0, %tmp1 + ret <8 x i16> %tmp2 +} + +; COST-LABEL: usubl_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i16> %a to <4 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <4 x i16> %b to <4 x i32> +; CODE-LABEL: usubl_4s +; CODE: usubl v0.4s, v0.4h, v1.4h +define <4 x i32> @usubl_4s(<4 x i16> %a, <4 x i16> %b) { + %tmp0 = zext <4 x i16> %a to <4 x i32> + %tmp1 = zext <4 x i16> %b to <4 x i32> + %tmp2 = sub <4 x i32> %tmp0, %tmp1 + ret <4 x i32> %tmp2 +} + +; COST-LABEL: usubl_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <2 x i32> %a to <2 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <2 x i32> %b to <2 x i64> +; CODE-LABEL: usubl_2d +; CODE: usubl v0.2d, v0.2s, v1.2s +define <2 x i64> @usubl_2d(<2 x i32> %a, <2 x i32> %b) { + %tmp0 = zext <2 x i32> %a to <2 x i64> + %tmp1 = zext <2 x i32> %b to <2 x i64> + %tmp2 = sub <2 x i64> %tmp0, %tmp1 + ret <2 x i64> %tmp2 +} + +; COST-LABEL: usubl2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <16 x i8> %a to <16 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <16 x i8> %b to <16 x i16> +; CODE-LABEL: usubl2_8h +; CODE: usubl2 v2.8h, v0.16b, v1.16b +; CODE-NEXT: usubl v0.8h, v0.8b, v1.8b +define <16 x i16> @usubl2_8h(<16 x i8> %a, <16 x i8> %b) { + %tmp0 = zext <16 x i8> %a to <16 x i16> + %tmp1 = zext <16 x i8> %b to <16 x i16> + %tmp2 = sub <16 x i16> %tmp0, %tmp1 + ret <16 x i16> %tmp2 +} + +; COST-LABEL: usubl2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i16> %a to <8 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <8 x i16> %b to <8 x i32> +; CODE-LABEL: usubl2_4s +; CODE: usubl2 v2.4s, v0.8h, v1.8h +; CODE-NEXT: usubl v0.4s, v0.4h, v1.4h +define <8 x i32> @usubl2_4s(<8 x i16> %a, <8 x i16> %b) { + %tmp0 = zext <8 x i16> %a to <8 x i32> + %tmp1 = zext <8 x i16> %b to <8 x i32> + %tmp2 = sub <8 x i32> %tmp0, %tmp1 + ret <8 x i32> %tmp2 +} + +; COST-LABEL: usubl2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i32> %a to <4 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <4 x i32> %b to <4 x i64> +; CODE-LABEL: usubl2_2d +; CODE: usubl2 v2.2d, v0.4s, v1.4s +; CODE-NEXT: usubl v0.2d, v0.2s, v1.2s +define <4 x i64> @usubl2_2d(<4 x i32> %a, <4 x i32> %b) { + %tmp0 = zext <4 x i32> %a to <4 x i64> + %tmp1 = zext <4 x i32> %b to <4 x i64> + %tmp2 = sub <4 x i64> %tmp0, %tmp1 + ret <4 x i64> %tmp2 +} + +; COST-LABEL: ssubl_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i8> %a to <8 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <8 x i8> %b to <8 x i16> +; CODE-LABEL: ssubl_8h +; CODE: ssubl v0.8h, v0.8b, v1.8b +define <8 x i16> @ssubl_8h(<8 x i8> %a, <8 x i8> %b) { + %tmp0 = sext <8 x i8> %a to <8 x i16> + %tmp1 = sext <8 x i8> %b to <8 x i16> + %tmp2 = sub <8 x i16> %tmp0, %tmp1 + ret <8 x i16> %tmp2 +} + +; COST-LABEL: ssubl_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i16> %a to <4 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <4 x i16> %b to <4 x i32> +; CODE-LABEL: ssubl_4s +; CODE: ssubl v0.4s, v0.4h, v1.4h +define <4 x i32> @ssubl_4s(<4 x i16> %a, <4 x i16> %b) { + %tmp0 = sext <4 x i16> %a to <4 x i32> + %tmp1 = sext <4 x i16> %b to <4 x i32> + %tmp2 = sub <4 x i32> %tmp0, %tmp1 + ret <4 x i32> %tmp2 +} + +; COST-LABEL: ssubl_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <2 x i32> %a to <2 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <2 x i32> %b to <2 x i64> +; CODE-LABEL: ssubl_2d +; CODE: ssubl v0.2d, v0.2s, v1.2s +define <2 x i64> @ssubl_2d(<2 x i32> %a, <2 x i32> %b) { + %tmp0 = sext <2 x i32> %a to <2 x i64> + %tmp1 = sext <2 x i32> %b to <2 x i64> + %tmp2 = sub <2 x i64> %tmp0, %tmp1 + ret <2 x i64> %tmp2 +} + +; COST-LABEL: ssubl2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <16 x i8> %a to <16 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <16 x i8> %b to <16 x i16> +; CODE-LABEL: ssubl2_8h +; CODE: ssubl2 v2.8h, v0.16b, v1.16b +; CODE-NEXT: ssubl v0.8h, v0.8b, v1.8b +define <16 x i16> @ssubl2_8h(<16 x i8> %a, <16 x i8> %b) { + %tmp0 = sext <16 x i8> %a to <16 x i16> + %tmp1 = sext <16 x i8> %b to <16 x i16> + %tmp2 = sub <16 x i16> %tmp0, %tmp1 + ret <16 x i16> %tmp2 +} + +; COST-LABEL: ssubl2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i16> %a to <8 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <8 x i16> %b to <8 x i32> +; CODE-LABEL: ssubl2_4s +; CODE: ssubl2 v2.4s, v0.8h, v1.8h +; CODE-NEXT: ssubl v0.4s, v0.4h, v1.4h +define <8 x i32> @ssubl2_4s(<8 x i16> %a, <8 x i16> %b) { + %tmp0 = sext <8 x i16> %a to <8 x i32> + %tmp1 = sext <8 x i16> %b to <8 x i32> + %tmp2 = sub <8 x i32> %tmp0, %tmp1 + ret <8 x i32> %tmp2 +} + +; COST-LABEL: ssubl2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i32> %a to <4 x i64> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = sext <4 x i32> %b to <4 x i64> +; CODE-LABEL: ssubl2_2d +; CODE: ssubl2 v2.2d, v0.4s, v1.4s +; CODE-NEXT: ssubl v0.2d, v0.2s, v1.2s +define <4 x i64> @ssubl2_2d(<4 x i32> %a, <4 x i32> %b) { + %tmp0 = sext <4 x i32> %a to <4 x i64> + %tmp1 = sext <4 x i32> %b to <4 x i64> + %tmp2 = sub <4 x i64> %tmp0, %tmp1 + ret <4 x i64> %tmp2 +} + +; COST-LABEL: uaddw_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i8> %a to <8 x i16> +; CODE-LABEL: uaddw_8h +; CODE: uaddw v0.8h, v1.8h, v0.8b +define <8 x i16> @uaddw_8h(<8 x i8> %a, <8 x i16> %b) { + %tmp0 = zext <8 x i8> %a to <8 x i16> + %tmp1 = add <8 x i16> %b, %tmp0 + ret <8 x i16> %tmp1 +} + +; COST-LABEL: uaddw_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i16> %a to <4 x i32> +; CODE-LABEL: uaddw_4s +; CODE: uaddw v0.4s, v1.4s, v0.4h +define <4 x i32> @uaddw_4s(<4 x i16> %a, <4 x i32> %b) { + %tmp0 = zext <4 x i16> %a to <4 x i32> + %tmp1 = add <4 x i32> %b, %tmp0 + ret <4 x i32> %tmp1 +} + +; COST-LABEL: uaddw_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <2 x i32> %a to <2 x i64> +; CODE-LABEL: uaddw_2d +; CODE: uaddw v0.2d, v1.2d, v0.2s +define <2 x i64> @uaddw_2d(<2 x i32> %a, <2 x i64> %b) { + %tmp0 = zext <2 x i32> %a to <2 x i64> + %tmp1 = add <2 x i64> %b, %tmp0 + ret <2 x i64> %tmp1 +} + +; COST-LABEL: uaddw2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <16 x i8> %a to <16 x i16> +; CODE-LABEL: uaddw2_8h +; CODE: uaddw2 v2.8h, v2.8h, v0.16b +; CODE-NEXT: uaddw v0.8h, v1.8h, v0.8b +define <16 x i16> @uaddw2_8h(<16 x i8> %a, <16 x i16> %b) { + %tmp0 = zext <16 x i8> %a to <16 x i16> + %tmp1 = add <16 x i16> %b, %tmp0 + ret <16 x i16> %tmp1 +} + +; COST-LABEL: uaddw2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i16> %a to <8 x i32> +; CODE-LABEL: uaddw2_4s +; CODE: uaddw2 v2.4s, v2.4s, v0.8h +; CODE-NEXT: uaddw v0.4s, v1.4s, v0.4h +define <8 x i32> @uaddw2_4s(<8 x i16> %a, <8 x i32> %b) { + %tmp0 = zext <8 x i16> %a to <8 x i32> + %tmp1 = add <8 x i32> %b, %tmp0 + ret <8 x i32> %tmp1 +} + +; COST-LABEL: uaddw2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i32> %a to <4 x i64> +; CODE-LABEL: uaddw2_2d +; CODE: uaddw2 v2.2d, v2.2d, v0.4s +; CODE-NEXT: uaddw v0.2d, v1.2d, v0.2s +define <4 x i64> @uaddw2_2d(<4 x i32> %a, <4 x i64> %b) { + %tmp0 = zext <4 x i32> %a to <4 x i64> + %tmp1 = add <4 x i64> %b, %tmp0 + ret <4 x i64> %tmp1 +} + +; COST-LABEL: saddw_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i8> %a to <8 x i16> +; CODE-LABEL: saddw_8h +; CODE: saddw v0.8h, v1.8h, v0.8b +define <8 x i16> @saddw_8h(<8 x i8> %a, <8 x i16> %b) { + %tmp0 = sext <8 x i8> %a to <8 x i16> + %tmp1 = add <8 x i16> %b, %tmp0 + ret <8 x i16> %tmp1 +} + +; COST-LABEL: saddw_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i16> %a to <4 x i32> +; CODE-LABEL: saddw_4s +; CODE: saddw v0.4s, v1.4s, v0.4h +define <4 x i32> @saddw_4s(<4 x i16> %a, <4 x i32> %b) { + %tmp0 = sext <4 x i16> %a to <4 x i32> + %tmp1 = add <4 x i32> %b, %tmp0 + ret <4 x i32> %tmp1 +} + +; COST-LABEL: saddw_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <2 x i32> %a to <2 x i64> +; CODE-LABEL: saddw_2d +; CODE: saddw v0.2d, v1.2d, v0.2s +define <2 x i64> @saddw_2d(<2 x i32> %a, <2 x i64> %b) { + %tmp0 = sext <2 x i32> %a to <2 x i64> + %tmp1 = add <2 x i64> %b, %tmp0 + ret <2 x i64> %tmp1 +} + +; COST-LABEL: saddw2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <16 x i8> %a to <16 x i16> +; CODE-LABEL: saddw2_8h +; CODE: saddw2 v2.8h, v2.8h, v0.16b +; CODE-NEXT: saddw v0.8h, v1.8h, v0.8b +define <16 x i16> @saddw2_8h(<16 x i8> %a, <16 x i16> %b) { + %tmp0 = sext <16 x i8> %a to <16 x i16> + %tmp1 = add <16 x i16> %b, %tmp0 + ret <16 x i16> %tmp1 +} + +; COST-LABEL: saddw2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i16> %a to <8 x i32> +; CODE-LABEL: saddw2_4s +; CODE: saddw2 v2.4s, v2.4s, v0.8h +; CODE-NEXT: saddw v0.4s, v1.4s, v0.4h +define <8 x i32> @saddw2_4s(<8 x i16> %a, <8 x i32> %b) { + %tmp0 = sext <8 x i16> %a to <8 x i32> + %tmp1 = add <8 x i32> %b, %tmp0 + ret <8 x i32> %tmp1 +} + +; COST-LABEL: saddw2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i32> %a to <4 x i64> +; CODE-LABEL: saddw2_2d +; CODE: saddw2 v2.2d, v2.2d, v0.4s +; CODE-NEXT: saddw v0.2d, v1.2d, v0.2s +define <4 x i64> @saddw2_2d(<4 x i32> %a, <4 x i64> %b) { + %tmp0 = sext <4 x i32> %a to <4 x i64> + %tmp1 = add <4 x i64> %b, %tmp0 + ret <4 x i64> %tmp1 +} + +; COST-LABEL: usubw_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i8> %a to <8 x i16> +; CODE-LABEL: usubw_8h +; CODE: usubw v0.8h, v1.8h, v0.8b +define <8 x i16> @usubw_8h(<8 x i8> %a, <8 x i16> %b) { + %tmp0 = zext <8 x i8> %a to <8 x i16> + %tmp1 = sub <8 x i16> %b, %tmp0 + ret <8 x i16> %tmp1 +} + +; COST-LABEL: usubw_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i16> %a to <4 x i32> +; CODE-LABEL: usubw_4s +; CODE: usubw v0.4s, v1.4s, v0.4h +define <4 x i32> @usubw_4s(<4 x i16> %a, <4 x i32> %b) { + %tmp0 = zext <4 x i16> %a to <4 x i32> + %tmp1 = sub <4 x i32> %b, %tmp0 + ret <4 x i32> %tmp1 +} + +; COST-LABEL: usubw_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <2 x i32> %a to <2 x i64> +; CODE-LABEL: usubw_2d +; CODE: usubw v0.2d, v1.2d, v0.2s +define <2 x i64> @usubw_2d(<2 x i32> %a, <2 x i64> %b) { + %tmp0 = zext <2 x i32> %a to <2 x i64> + %tmp1 = sub <2 x i64> %b, %tmp0 + ret <2 x i64> %tmp1 +} + +; COST-LABEL: usubw2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <16 x i8> %a to <16 x i16> +; CODE-LABEL: usubw2_8h +; CODE: usubw2 v2.8h, v2.8h, v0.16b +; CODE-NEXT: usubw v0.8h, v1.8h, v0.8b +define <16 x i16> @usubw2_8h(<16 x i8> %a, <16 x i16> %b) { + %tmp0 = zext <16 x i8> %a to <16 x i16> + %tmp1 = sub <16 x i16> %b, %tmp0 + ret <16 x i16> %tmp1 +} + +; COST-LABEL: usubw2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <8 x i16> %a to <8 x i32> +; CODE-LABEL: usubw2_4s +; CODE: usubw2 v2.4s, v2.4s, v0.8h +; CODE-NEXT: usubw v0.4s, v1.4s, v0.4h +define <8 x i32> @usubw2_4s(<8 x i16> %a, <8 x i32> %b) { + %tmp0 = zext <8 x i16> %a to <8 x i32> + %tmp1 = sub <8 x i32> %b, %tmp0 + ret <8 x i32> %tmp1 +} + +; COST-LABEL: usubw2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = zext <4 x i32> %a to <4 x i64> +; CODE-LABEL: usubw2_2d +; CODE: usubw2 v2.2d, v2.2d, v0.4s +; CODE-NEXT: usubw v0.2d, v1.2d, v0.2s +define <4 x i64> @usubw2_2d(<4 x i32> %a, <4 x i64> %b) { + %tmp0 = zext <4 x i32> %a to <4 x i64> + %tmp1 = sub <4 x i64> %b, %tmp0 + ret <4 x i64> %tmp1 +} + +; COST-LABEL: ssubw_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i8> %a to <8 x i16> +; CODE-LABEL: ssubw_8h +; CODE: ssubw v0.8h, v1.8h, v0.8b +define <8 x i16> @ssubw_8h(<8 x i8> %a, <8 x i16> %b) { + %tmp0 = sext <8 x i8> %a to <8 x i16> + %tmp1 = sub <8 x i16> %b, %tmp0 + ret <8 x i16> %tmp1 +} + +; COST-LABEL: ssubw_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i16> %a to <4 x i32> +; CODE-LABEL: ssubw_4s +; CODE: ssubw v0.4s, v1.4s, v0.4h +define <4 x i32> @ssubw_4s(<4 x i16> %a, <4 x i32> %b) { + %tmp0 = sext <4 x i16> %a to <4 x i32> + %tmp1 = sub <4 x i32> %b, %tmp0 + ret <4 x i32> %tmp1 +} + +; COST-LABEL: ssubw_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <2 x i32> %a to <2 x i64> +; CODE-LABEL: ssubw_2d +; CODE: ssubw v0.2d, v1.2d, v0.2s +define <2 x i64> @ssubw_2d(<2 x i32> %a, <2 x i64> %b) { + %tmp0 = sext <2 x i32> %a to <2 x i64> + %tmp1 = sub <2 x i64> %b, %tmp0 + ret <2 x i64> %tmp1 +} + +; COST-LABEL: ssubw2_8h +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <16 x i8> %a to <16 x i16> +; CODE-LABEL: ssubw2_8h +; CODE: ssubw2 v2.8h, v2.8h, v0.16b +; CODE-NEXT: ssubw v0.8h, v1.8h, v0.8b +define <16 x i16> @ssubw2_8h(<16 x i8> %a, <16 x i16> %b) { + %tmp0 = sext <16 x i8> %a to <16 x i16> + %tmp1 = sub <16 x i16> %b, %tmp0 + ret <16 x i16> %tmp1 +} + +; COST-LABEL: ssubw2_4s +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <8 x i16> %a to <8 x i32> +; CODE-LABEL: ssubw2_4s +; CODE: ssubw2 v2.4s, v2.4s, v0.8h +; CODE-NEXT: ssubw v0.4s, v1.4s, v0.4h +define <8 x i32> @ssubw2_4s(<8 x i16> %a, <8 x i32> %b) { + %tmp0 = sext <8 x i16> %a to <8 x i32> + %tmp1 = sub <8 x i32> %b, %tmp0 + ret <8 x i32> %tmp1 +} + +; COST-LABEL: ssubw2_2d +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp0 = sext <4 x i32> %a to <4 x i64> +; CODE-LABEL: ssubw2_2d +; CODE: ssubw2 v2.2d, v2.2d, v0.4s +; CODE-NEXT: ssubw v0.2d, v1.2d, v0.2s +define <4 x i64> @ssubw2_2d(<4 x i32> %a, <4 x i64> %b) { + %tmp0 = sext <4 x i32> %a to <4 x i64> + %tmp1 = sub <4 x i64> %b, %tmp0 + ret <4 x i64> %tmp1 +} + +; COST-LABEL: neg_wrong_operand_order +; COST-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %tmp0 = zext <8 x i8> %a to <8 x i16> +define <8 x i16> @neg_wrong_operand_order(<8 x i8> %a, <8 x i16> %b) { + %tmp0 = zext <8 x i8> %a to <8 x i16> + %tmp1 = sub <8 x i16> %tmp0, %b + ret <8 x i16> %tmp1 +} + +; COST-LABEL: neg_non_widening_op +; COST-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %tmp0 = zext <8 x i8> %a to <8 x i16> +define <8 x i16> @neg_non_widening_op(<8 x i8> %a, <8 x i16> %b) { + %tmp0 = zext <8 x i8> %a to <8 x i16> + %tmp1 = udiv <8 x i16> %b, %tmp0 + ret <8 x i16> %tmp1 +} + +; COST-LABEL: neg_dissimilar_operand_kind_0 +; COST-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %tmp0 = sext <8 x i8> %a to <8 x i16> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <8 x i8> %b to <8 x i16> +define <8 x i16> @neg_dissimilar_operand_kind_0(<8 x i8> %a, <8 x i8> %b) { + %tmp0 = sext <8 x i8> %a to <8 x i16> + %tmp1 = zext <8 x i8> %b to <8 x i16> + %tmp2 = add <8 x i16> %tmp0, %tmp1 + ret <8 x i16> %tmp2 +} + +; COST-LABEL: neg_dissimilar_operand_kind_1 +; COST-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %tmp0 = zext <4 x i8> %a to <4 x i32> +; COST-NEXT: Cost Model: Found an estimated cost of 0 for instruction: %tmp1 = zext <4 x i16> %b to <4 x i32> +define <4 x i32> @neg_dissimilar_operand_kind_1(<4 x i8> %a, <4 x i16> %b) { + %tmp0 = zext <4 x i8> %a to <4 x i32> + %tmp1 = zext <4 x i16> %b to <4 x i32> + %tmp2 = add <4 x i32> %tmp0, %tmp1 + ret <4 x i32> %tmp2 +} + +; COST-LABEL: neg_illegal_vector_type_0 +; COST-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %tmp0 = zext <16 x i4> %a to <16 x i8> +define <16 x i8> @neg_illegal_vector_type_0(<16 x i4> %a, <16 x i8> %b) { + %tmp0 = zext <16 x i4> %a to <16 x i8> + %tmp1 = sub <16 x i8> %b, %tmp0 + ret <16 x i8> %tmp1 +} + +; COST-LABEL: neg_llegal_vector_type_1 +; COST-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %tmp0 = zext <1 x i16> %a to <1 x i32> +define <1 x i32> @neg_llegal_vector_type_1(<1 x i16> %a, <1 x i32> %b) { + %tmp0 = zext <1 x i16> %a to <1 x i32> + %tmp1 = add <1 x i32> %b, %tmp0 + ret <1 x i32> %tmp1 +} + +; COST-LABEL: neg_llegal_vector_type_2 +; COST-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %tmp0 = zext <4 x i16> %a to <4 x i64> +define <4 x i64> @neg_llegal_vector_type_2(<4 x i16> %a, <4 x i64> %b) { + %tmp0 = zext <4 x i16> %a to <4 x i64> + %tmp1 = add <4 x i64> %b, %tmp0 + ret <4 x i64> %tmp1 +} + +; COST-LABEL: neg_llegal_vector_type_3 +; COST-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %tmp0 = zext <3 x i34> %a to <3 x i68> +define <3 x i68> @neg_llegal_vector_type_3(<3 x i34> %a, <3 x i68> %b) { + %tmp0 = zext <3 x i34> %a to <3 x i68> + %tmp1 = add <3 x i68> %b, %tmp0 + ret <3 x i68> %tmp1 +} diff --git a/test/Analysis/CostModel/AMDGPU/extractelement.ll b/test/Analysis/CostModel/AMDGPU/extractelement.ll index 1efbb5873acb2..54c8b6c52365c 100644 --- a/test/Analysis/CostModel/AMDGPU/extractelement.ll +++ b/test/Analysis/CostModel/AMDGPU/extractelement.ll @@ -1,7 +1,9 @@ -; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa < %s | FileCheck %s +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa %s | FileCheck -check-prefixes=GCN,CI %s +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa -mcpu=fiji %s | FileCheck -check-prefixes=GCN,VI %s +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx900 %s | FileCheck -check-prefixes=GCN,GFX9 %s -; CHECK: 'extractelement_v2i32' -; CHECK: estimated cost of 0 for {{.*}} extractelement <2 x i32> +; GCN: 'extractelement_v2i32' +; GCN: estimated cost of 0 for {{.*}} extractelement <2 x i32> define amdgpu_kernel void @extractelement_v2i32(i32 addrspace(1)* %out, <2 x i32> addrspace(1)* %vaddr) { %vec = load <2 x i32>, <2 x i32> addrspace(1)* %vaddr %elt = extractelement <2 x i32> %vec, i32 1 @@ -9,8 +11,8 @@ define amdgpu_kernel void @extractelement_v2i32(i32 addrspace(1)* %out, <2 x i32 ret void } -; CHECK: 'extractelement_v2f32' -; CHECK: estimated cost of 0 for {{.*}} extractelement <2 x float> +; GCN: 'extractelement_v2f32' +; GCN: estimated cost of 0 for {{.*}} extractelement <2 x float> define amdgpu_kernel void @extractelement_v2f32(float addrspace(1)* %out, <2 x float> addrspace(1)* %vaddr) { %vec = load <2 x float>, <2 x float> addrspace(1)* %vaddr %elt = extractelement <2 x float> %vec, i32 1 @@ -18,8 +20,8 @@ define amdgpu_kernel void @extractelement_v2f32(float addrspace(1)* %out, <2 x f ret void } -; CHECK: 'extractelement_v3i32' -; CHECK: estimated cost of 0 for {{.*}} extractelement <3 x i32> +; GCN: 'extractelement_v3i32' +; GCN: estimated cost of 0 for {{.*}} extractelement <3 x i32> define amdgpu_kernel void @extractelement_v3i32(i32 addrspace(1)* %out, <3 x i32> addrspace(1)* %vaddr) { %vec = load <3 x i32>, <3 x i32> addrspace(1)* %vaddr %elt = extractelement <3 x i32> %vec, i32 1 @@ -27,8 +29,8 @@ define amdgpu_kernel void @extractelement_v3i32(i32 addrspace(1)* %out, <3 x i32 ret void } -; CHECK: 'extractelement_v4i32' -; CHECK: estimated cost of 0 for {{.*}} extractelement <4 x i32> +; GCN: 'extractelement_v4i32' +; GCN: estimated cost of 0 for {{.*}} extractelement <4 x i32> define amdgpu_kernel void @extractelement_v4i32(i32 addrspace(1)* %out, <4 x i32> addrspace(1)* %vaddr) { %vec = load <4 x i32>, <4 x i32> addrspace(1)* %vaddr %elt = extractelement <4 x i32> %vec, i32 1 @@ -36,8 +38,8 @@ define amdgpu_kernel void @extractelement_v4i32(i32 addrspace(1)* %out, <4 x i32 ret void } -; CHECK: 'extractelement_v8i32' -; CHECK: estimated cost of 0 for {{.*}} extractelement <8 x i32> +; GCN: 'extractelement_v8i32' +; GCN: estimated cost of 0 for {{.*}} extractelement <8 x i32> define amdgpu_kernel void @extractelement_v8i32(i32 addrspace(1)* %out, <8 x i32> addrspace(1)* %vaddr) { %vec = load <8 x i32>, <8 x i32> addrspace(1)* %vaddr %elt = extractelement <8 x i32> %vec, i32 1 @@ -46,8 +48,8 @@ define amdgpu_kernel void @extractelement_v8i32(i32 addrspace(1)* %out, <8 x i32 } ; FIXME: Should be non-0 -; CHECK: 'extractelement_v8i32_dynindex' -; CHECK: estimated cost of 2 for {{.*}} extractelement <8 x i32> +; GCN: 'extractelement_v8i32_dynindex' +; GCN: estimated cost of 2 for {{.*}} extractelement <8 x i32> define amdgpu_kernel void @extractelement_v8i32_dynindex(i32 addrspace(1)* %out, <8 x i32> addrspace(1)* %vaddr, i32 %idx) { %vec = load <8 x i32>, <8 x i32> addrspace(1)* %vaddr %elt = extractelement <8 x i32> %vec, i32 %idx @@ -55,8 +57,8 @@ define amdgpu_kernel void @extractelement_v8i32_dynindex(i32 addrspace(1)* %out, ret void } -; CHECK: 'extractelement_v2i64' -; CHECK: estimated cost of 0 for {{.*}} extractelement <2 x i64> +; GCN: 'extractelement_v2i64' +; GCN: estimated cost of 0 for {{.*}} extractelement <2 x i64> define amdgpu_kernel void @extractelement_v2i64(i64 addrspace(1)* %out, <2 x i64> addrspace(1)* %vaddr) { %vec = load <2 x i64>, <2 x i64> addrspace(1)* %vaddr %elt = extractelement <2 x i64> %vec, i64 1 @@ -64,8 +66,8 @@ define amdgpu_kernel void @extractelement_v2i64(i64 addrspace(1)* %out, <2 x i64 ret void } -; CHECK: 'extractelement_v3i64' -; CHECK: estimated cost of 0 for {{.*}} extractelement <3 x i64> +; GCN: 'extractelement_v3i64' +; GCN: estimated cost of 0 for {{.*}} extractelement <3 x i64> define amdgpu_kernel void @extractelement_v3i64(i64 addrspace(1)* %out, <3 x i64> addrspace(1)* %vaddr) { %vec = load <3 x i64>, <3 x i64> addrspace(1)* %vaddr %elt = extractelement <3 x i64> %vec, i64 1 @@ -73,8 +75,8 @@ define amdgpu_kernel void @extractelement_v3i64(i64 addrspace(1)* %out, <3 x i64 ret void } -; CHECK: 'extractelement_v4i64' -; CHECK: estimated cost of 0 for {{.*}} extractelement <4 x i64> +; GCN: 'extractelement_v4i64' +; GCN: estimated cost of 0 for {{.*}} extractelement <4 x i64> define amdgpu_kernel void @extractelement_v4i64(i64 addrspace(1)* %out, <4 x i64> addrspace(1)* %vaddr) { %vec = load <4 x i64>, <4 x i64> addrspace(1)* %vaddr %elt = extractelement <4 x i64> %vec, i64 1 @@ -82,8 +84,8 @@ define amdgpu_kernel void @extractelement_v4i64(i64 addrspace(1)* %out, <4 x i64 ret void } -; CHECK: 'extractelement_v8i64' -; CHECK: estimated cost of 0 for {{.*}} extractelement <8 x i64> +; GCN: 'extractelement_v8i64' +; GCN: estimated cost of 0 for {{.*}} extractelement <8 x i64> define amdgpu_kernel void @extractelement_v8i64(i64 addrspace(1)* %out, <8 x i64> addrspace(1)* %vaddr) { %vec = load <8 x i64>, <8 x i64> addrspace(1)* %vaddr %elt = extractelement <8 x i64> %vec, i64 1 @@ -91,8 +93,8 @@ define amdgpu_kernel void @extractelement_v8i64(i64 addrspace(1)* %out, <8 x i64 ret void } -; CHECK: 'extractelement_v4i8' -; CHECK: estimated cost of 0 for {{.*}} extractelement <4 x i8> +; GCN: 'extractelement_v4i8' +; GCN: estimated cost of 1 for {{.*}} extractelement <4 x i8> define amdgpu_kernel void @extractelement_v4i8(i8 addrspace(1)* %out, <4 x i8> addrspace(1)* %vaddr) { %vec = load <4 x i8>, <4 x i8> addrspace(1)* %vaddr %elt = extractelement <4 x i8> %vec, i8 1 @@ -100,11 +102,31 @@ define amdgpu_kernel void @extractelement_v4i8(i8 addrspace(1)* %out, <4 x i8> a ret void } -; CHECK: 'extractelement_v2i16' -; CHECK: estimated cost of 0 for {{.*}} extractelement <2 x i16> -define amdgpu_kernel void @extractelement_v2i16(i16 addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { +; GCN: 'extractelement_0_v2i16': +; CI: estimated cost of 1 for {{.*}} extractelement <2 x i16> %vec, i16 0 +; VI: estimated cost of 0 for {{.*}} extractelement <2 x i16> +; GFX9: estimated cost of 0 for {{.*}} extractelement <2 x i16> +define amdgpu_kernel void @extractelement_0_v2i16(i16 addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { + %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr + %elt = extractelement <2 x i16> %vec, i16 0 + store i16 %elt, i16 addrspace(1)* %out + ret void +} + +; GCN: 'extractelement_1_v2i16': +; GCN: estimated cost of 1 for {{.*}} extractelement <2 x i16> +define amdgpu_kernel void @extractelement_1_v2i16(i16 addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr %elt = extractelement <2 x i16> %vec, i16 1 store i16 %elt, i16 addrspace(1)* %out ret void } + +; GCN: 'extractelement_var_v2i16' +; GCN: estimated cost of 1 for {{.*}} extractelement <2 x i16> +define amdgpu_kernel void @extractelement_var_v2i16(i16 addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr, i32 %idx) { + %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr + %elt = extractelement <2 x i16> %vec, i32 %idx + store i16 %elt, i16 addrspace(1)* %out + ret void +} diff --git a/test/Analysis/CostModel/AMDGPU/insertelement.ll b/test/Analysis/CostModel/AMDGPU/insertelement.ll index 6f296a3e7a345..67ab2607acd5f 100644 --- a/test/Analysis/CostModel/AMDGPU/insertelement.ll +++ b/test/Analysis/CostModel/AMDGPU/insertelement.ll @@ -1,37 +1,50 @@ -; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa < %s | FileCheck %s +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa %s | FileCheck -check-prefixes=GCN,CI %s +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa -mcpu=fiji %s | FileCheck -check-prefixes=GCN,VI %s +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx900 %s | FileCheck -check-prefixes=GCN,GFX9 %s -; CHECK: 'insertelement_v2i32' -; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i32> +; GCN-LABEL: 'insertelement_v2i32' +; GCN: estimated cost of 0 for {{.*}} insertelement <2 x i32> define amdgpu_kernel void @insertelement_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %vaddr) { %vec = load <2 x i32>, <2 x i32> addrspace(1)* %vaddr - %insert = insertelement <2 x i32> %vec, i32 1, i32 123 + %insert = insertelement <2 x i32> %vec, i32 123, i32 1 store <2 x i32> %insert, <2 x i32> addrspace(1)* %out ret void } -; CHECK: 'insertelement_v2i64' -; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i64> +; GCN-LABEL: 'insertelement_v2i64' +; GCN: estimated cost of 0 for {{.*}} insertelement <2 x i64> define amdgpu_kernel void @insertelement_v2i64(<2 x i64> addrspace(1)* %out, <2 x i64> addrspace(1)* %vaddr) { %vec = load <2 x i64>, <2 x i64> addrspace(1)* %vaddr - %insert = insertelement <2 x i64> %vec, i64 1, i64 123 + %insert = insertelement <2 x i64> %vec, i64 123, i64 1 store <2 x i64> %insert, <2 x i64> addrspace(1)* %out ret void } -; CHECK: 'insertelement_v2i16' -; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i16> -define amdgpu_kernel void @insertelement_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { +; GCN-LABEL: 'insertelement_0_v2i16' +; CI: estimated cost of 1 for {{.*}} insertelement <2 x i16> +; VI: estimated cost of 0 for {{.*}} insertelement <2 x i16> +; GFX9: estimated cost of 0 for {{.*}} insertelement <2 x i16> +define amdgpu_kernel void @insertelement_0_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr - %insert = insertelement <2 x i16> %vec, i16 1, i16 123 + %insert = insertelement <2 x i16> %vec, i16 123, i16 0 store <2 x i16> %insert, <2 x i16> addrspace(1)* %out ret void } -; CHECK: 'insertelement_v2i8' -; CHECK: estimated cost of 0 for {{.*}} insertelement <2 x i8> -define amdgpu_kernel void @insertelement_v2i8(<2 x i8> addrspace(1)* %out, <2 x i8> addrspace(1)* %vaddr) { +; GCN-LABEL: 'insertelement_1_v2i16' +; GCN: estimated cost of 1 for {{.*}} insertelement <2 x i16> +define amdgpu_kernel void @insertelement_1_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { + %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr + %insert = insertelement <2 x i16> %vec, i16 123, i16 1 + store <2 x i16> %insert, <2 x i16> addrspace(1)* %out + ret void +} + +; GCN-LABEL: 'insertelement_1_v2i8' +; GCN: estimated cost of 1 for {{.*}} insertelement <2 x i8> +define amdgpu_kernel void @insertelement_1_v2i8(<2 x i8> addrspace(1)* %out, <2 x i8> addrspace(1)* %vaddr) { %vec = load <2 x i8>, <2 x i8> addrspace(1)* %vaddr - %insert = insertelement <2 x i8> %vec, i8 1, i8 123 + %insert = insertelement <2 x i8> %vec, i8 123, i8 1 store <2 x i8> %insert, <2 x i8> addrspace(1)* %out ret void } diff --git a/test/Analysis/CostModel/AMDGPU/shufflevector.ll b/test/Analysis/CostModel/AMDGPU/shufflevector.ll new file mode 100644 index 0000000000000..cc756c82fed32 --- /dev/null +++ b/test/Analysis/CostModel/AMDGPU/shufflevector.ll @@ -0,0 +1,43 @@ +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa -mcpu=gfx900 %s | FileCheck -check-prefixes=GFX9,GCN %s +; RUN: opt -cost-model -analyze -mtriple=amdgcn-unknown-amdhsa -mcpu=fiji %s | FileCheck -check-prefixes=VI,GCN %s + +; GFX9: estimated cost of 0 for {{.*}} shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> zeroinitializer +define amdgpu_kernel void @shufflevector_00_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { + %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr + %shuf = shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> zeroinitializer + store <2 x i16> %shuf, <2 x i16> addrspace(1)* %out + ret void +} + +; GFX9: estimated cost of 0 for {{.*}} shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> <i32 0, i32 1> +define amdgpu_kernel void @shufflevector_01_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { + %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr + %shuf = shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> <i32 0, i32 1> + store <2 x i16> %shuf, <2 x i16> addrspace(1)* %out + ret void +} + +; GFX9: estimated cost of 0 for {{.*}} shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> <i32 1, i32 0> +define amdgpu_kernel void @shufflevector_10_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { + %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr + %shuf = shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> <i32 1, i32 0> + store <2 x i16> %shuf, <2 x i16> addrspace(1)* %out + ret void +} + +; GFX9: estimated cost of 0 for {{.*}} shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> <i32 1, i32 1> +define amdgpu_kernel void @shufflevector_11_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr) { + %vec = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr + %shuf = shufflevector <2 x i16> %vec, <2 x i16> undef, <2 x i32> <i32 1, i32 1> + store <2 x i16> %shuf, <2 x i16> addrspace(1)* %out + ret void +} + +; GCN: estimated cost of 2 for {{.*}} shufflevector <2 x i16> %vec0, <2 x i16> %vec1, <2 x i32> <i32 0, i32 2> +define amdgpu_kernel void @shufflevector_02_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %vaddr0, <2 x i16> addrspace(1)* %vaddr1) { + %vec0 = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr0 + %vec1 = load <2 x i16>, <2 x i16> addrspace(1)* %vaddr1 + %shuf = shufflevector <2 x i16> %vec0, <2 x i16> %vec1, <2 x i32> <i32 0, i32 2> + store <2 x i16> %shuf, <2 x i16> addrspace(1)* %out + ret void +} diff --git a/test/Analysis/CostModel/X86/div.ll b/test/Analysis/CostModel/X86/div.ll index 0ac06ff75ebec..dabaaef3596ab 100644 --- a/test/Analysis/CostModel/X86/div.ll +++ b/test/Analysis/CostModel/X86/div.ll @@ -139,14 +139,14 @@ define i32 @sdiv_uniformconst() { ; SSE2: cost of 38 {{.*}} %V8i32 = sdiv ; SSSE3: cost of 38 {{.*}} %V8i32 = sdiv ; SSE42: cost of 30 {{.*}} %V8i32 = sdiv - ; AVX1: cost of 30 {{.*}} %V8i32 = sdiv + ; AVX1: cost of 32 {{.*}} %V8i32 = sdiv ; AVX2: cost of 15 {{.*}} %V8i32 = sdiv ; AVX512: cost of 15 {{.*}} %V8i32 = sdiv %V8i32 = sdiv <8 x i32> undef, <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7> ; SSE2: cost of 76 {{.*}} %V16i32 = sdiv ; SSSE3: cost of 76 {{.*}} %V16i32 = sdiv ; SSE42: cost of 60 {{.*}} %V16i32 = sdiv - ; AVX1: cost of 60 {{.*}} %V16i32 = sdiv + ; AVX1: cost of 64 {{.*}} %V16i32 = sdiv ; AVX2: cost of 30 {{.*}} %V16i32 = sdiv ; AVX512: cost of 15 {{.*}} %V16i32 = sdiv %V16i32 = sdiv <16 x i32> undef, <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7> @@ -157,12 +157,12 @@ define i32 @sdiv_uniformconst() { ; AVX: cost of 6 {{.*}} %V8i16 = sdiv %V8i16 = sdiv <8 x i16> undef, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7> ; SSE: cost of 12 {{.*}} %V16i16 = sdiv - ; AVX1: cost of 12 {{.*}} %V16i16 = sdiv + ; AVX1: cost of 14 {{.*}} %V16i16 = sdiv ; AVX2: cost of 6 {{.*}} %V16i16 = sdiv ; AVX512: cost of 6 {{.*}} %V16i16 = sdiv %V16i16 = sdiv <16 x i16> undef, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7> ; SSE: cost of 24 {{.*}} %V32i16 = sdiv - ; AVX1: cost of 24 {{.*}} %V32i16 = sdiv + ; AVX1: cost of 28 {{.*}} %V32i16 = sdiv ; AVX2: cost of 12 {{.*}} %V32i16 = sdiv ; AVX512F: cost of 12 {{.*}} %V32i16 = sdiv ; AVX512BW: cost of 6 {{.*}} %V32i16 = sdiv @@ -203,12 +203,12 @@ define i32 @udiv_uniformconst() { ; AVX: cost of 15 {{.*}} %V4i32 = udiv %V4i32 = udiv <4 x i32> undef, <i32 7, i32 7, i32 7, i32 7> ; SSE: cost of 30 {{.*}} %V8i32 = udiv - ; AVX1: cost of 30 {{.*}} %V8i32 = udiv + ; AVX1: cost of 32 {{.*}} %V8i32 = udiv ; AVX2: cost of 15 {{.*}} %V8i32 = udiv ; AVX512: cost of 15 {{.*}} %V8i32 = udiv %V8i32 = udiv <8 x i32> undef, <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7> ; SSE: cost of 60 {{.*}} %V16i32 = udiv - ; AVX1: cost of 60 {{.*}} %V16i32 = udiv + ; AVX1: cost of 64 {{.*}} %V16i32 = udiv ; AVX2: cost of 30 {{.*}} %V16i32 = udiv ; AVX512: cost of 15 {{.*}} %V16i32 = udiv %V16i32 = udiv <16 x i32> undef, <i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7, i32 7> @@ -219,12 +219,12 @@ define i32 @udiv_uniformconst() { ; AVX: cost of 6 {{.*}} %V8i16 = udiv %V8i16 = udiv <8 x i16> undef, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7> ; SSE: cost of 12 {{.*}} %V16i16 = udiv - ; AVX1: cost of 12 {{.*}} %V16i16 = udiv + ; AVX1: cost of 14 {{.*}} %V16i16 = udiv ; AVX2: cost of 6 {{.*}} %V16i16 = udiv ; AVX512: cost of 6 {{.*}} %V16i16 = udiv %V16i16 = udiv <16 x i16> undef, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7> ; SSE: cost of 24 {{.*}} %V32i16 = udiv - ; AVX1: cost of 24 {{.*}} %V32i16 = udiv + ; AVX1: cost of 28 {{.*}} %V32i16 = udiv ; AVX2: cost of 12 {{.*}} %V32i16 = udiv ; AVX512F: cost of 12 {{.*}} %V32i16 = udiv ; AVX512BW: cost of 6 {{.*}} %V32i16 = udiv @@ -269,14 +269,14 @@ define i32 @sdiv_uniformconstpow2() { ; SSE2: cost of 38 {{.*}} %V8i32 = sdiv ; SSSE3: cost of 38 {{.*}} %V8i32 = sdiv ; SSE42: cost of 30 {{.*}} %V8i32 = sdiv - ; AVX1: cost of 30 {{.*}} %V8i32 = sdiv + ; AVX1: cost of 32 {{.*}} %V8i32 = sdiv ; AVX2: cost of 15 {{.*}} %V8i32 = sdiv ; AVX512: cost of 15 {{.*}} %V8i32 = sdiv %V8i32 = sdiv <8 x i32> undef, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16> ; SSE2: cost of 76 {{.*}} %V16i32 = sdiv ; SSSE3: cost of 76 {{.*}} %V16i32 = sdiv ; SSE42: cost of 60 {{.*}} %V16i32 = sdiv - ; AVX1: cost of 60 {{.*}} %V16i32 = sdiv + ; AVX1: cost of 64 {{.*}} %V16i32 = sdiv ; AVX2: cost of 30 {{.*}} %V16i32 = sdiv ; AVX512: cost of 15 {{.*}} %V16i32 = sdiv %V16i32 = sdiv <16 x i32> undef, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16> @@ -287,12 +287,12 @@ define i32 @sdiv_uniformconstpow2() { ; AVX: cost of 6 {{.*}} %V8i16 = sdiv %V8i16 = sdiv <8 x i16> undef, <i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16> ; SSE: cost of 12 {{.*}} %V16i16 = sdiv - ; AVX1: cost of 12 {{.*}} %V16i16 = sdiv + ; AVX1: cost of 14 {{.*}} %V16i16 = sdiv ; AVX2: cost of 6 {{.*}} %V16i16 = sdiv ; AVX512: cost of 6 {{.*}} %V16i16 = sdiv %V16i16 = sdiv <16 x i16> undef, <i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16> ; SSE: cost of 24 {{.*}} %V32i16 = sdiv - ; AVX1: cost of 24 {{.*}} %V32i16 = sdiv + ; AVX1: cost of 28 {{.*}} %V32i16 = sdiv ; AVX2: cost of 12 {{.*}} %V32i16 = sdiv ; AVX512F: cost of 12 {{.*}} %V32i16 = sdiv ; AVX512BW: cost of 6 {{.*}} %V32i16 = sdiv @@ -333,12 +333,12 @@ define i32 @udiv_uniformconstpow2() { ; AVX: cost of 15 {{.*}} %V4i32 = udiv %V4i32 = udiv <4 x i32> undef, <i32 16, i32 16, i32 16, i32 16> ; SSE: cost of 30 {{.*}} %V8i32 = udiv - ; AVX1: cost of 30 {{.*}} %V8i32 = udiv + ; AVX1: cost of 32 {{.*}} %V8i32 = udiv ; AVX2: cost of 15 {{.*}} %V8i32 = udiv ; AVX512: cost of 15 {{.*}} %V8i32 = udiv %V8i32 = udiv <8 x i32> undef, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16> ; SSE: cost of 60 {{.*}} %V16i32 = udiv - ; AVX1: cost of 60 {{.*}} %V16i32 = udiv + ; AVX1: cost of 64 {{.*}} %V16i32 = udiv ; AVX2: cost of 30 {{.*}} %V16i32 = udiv ; AVX512: cost of 15 {{.*}} %V16i32 = udiv %V16i32 = udiv <16 x i32> undef, <i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16, i32 16> @@ -349,12 +349,12 @@ define i32 @udiv_uniformconstpow2() { ; AVX: cost of 6 {{.*}} %V8i16 = udiv %V8i16 = udiv <8 x i16> undef, <i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16> ; SSE: cost of 12 {{.*}} %V16i16 = udiv - ; AVX1: cost of 12 {{.*}} %V16i16 = udiv + ; AVX1: cost of 14 {{.*}} %V16i16 = udiv ; AVX2: cost of 6 {{.*}} %V16i16 = udiv ; AVX512: cost of 6 {{.*}} %V16i16 = udiv %V16i16 = udiv <16 x i16> undef, <i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16, i16 16> ; SSE: cost of 24 {{.*}} %V32i16 = udiv - ; AVX1: cost of 24 {{.*}} %V32i16 = udiv + ; AVX1: cost of 28 {{.*}} %V32i16 = udiv ; AVX2: cost of 12 {{.*}} %V32i16 = udiv ; AVX512F: cost of 12 {{.*}} %V32i16 = udiv ; AVX512BW: cost of 6 {{.*}} %V32i16 = udiv diff --git a/test/Analysis/CostModel/X86/vshift-ashr-cost.ll b/test/Analysis/CostModel/X86/vshift-ashr-cost.ll index a23b13fb2e259..eabc2330ddc66 100644 --- a/test/Analysis/CostModel/X86/vshift-ashr-cost.ll +++ b/test/Analysis/CostModel/X86/vshift-ashr-cost.ll @@ -33,10 +33,10 @@ define <4 x i64> @var_shift_v4i64(<4 x i64> %a, <4 x i64> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v4i64': ; SSE2: Found an estimated cost of 24 for instruction: %shift ; SSE41: Found an estimated cost of 24 for instruction: %shift -; AVX: Found an estimated cost of 24 for instruction: %shift +; AVX: Found an estimated cost of 26 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = ashr <4 x i64> %a, %b ret <4 x i64> %shift } @@ -45,10 +45,10 @@ define <8 x i64> @var_shift_v8i64(<8 x i64> %a, <8 x i64> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v8i64': ; SSE2: Found an estimated cost of 48 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 52 for instruction: %shift ; AVX2: Found an estimated cost of 8 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = ashr <8 x i64> %a, %b ret <8 x i64> %shift } @@ -70,10 +70,10 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v8i32': ; SSE2: Found an estimated cost of 32 for instruction: %shift ; SSE41: Found an estimated cost of 24 for instruction: %shift -; AVX: Found an estimated cost of 24 for instruction: %shift +; AVX: Found an estimated cost of 26 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = ashr <8 x i32> %a, %b ret <8 x i32> %shift @@ -83,10 +83,10 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v16i32': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 52 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = ashr <16 x i32> %a, %b ret <16 x i32> %shift @@ -109,11 +109,11 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v16i16': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 28 for instruction: %shift -; AVX: Found an estimated cost of 28 for instruction: %shift +; AVX: Found an estimated cost of 30 for instruction: %shift ; AVX2: Found an estimated cost of 10 for instruction: %shift ; AVX512F: Found an estimated cost of 10 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = ashr <16 x i16> %a, %b ret <16 x i16> %shift } @@ -122,11 +122,11 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v32i16': ; SSE2: Found an estimated cost of 128 for instruction: %shift ; SSE41: Found an estimated cost of 56 for instruction: %shift -; AVX: Found an estimated cost of 56 for instruction: %shift +; AVX: Found an estimated cost of 60 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = ashr <32 x i16> %a, %b ret <32 x i16> %shift } @@ -147,11 +147,11 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v32i8': ; SSE2: Found an estimated cost of 108 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 50 for instruction: %shift ; AVX2: Found an estimated cost of 24 for instruction: %shift ; AVX512F: Found an estimated cost of 24 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = ashr <32 x i8> %a, %b ret <32 x i8> %shift } @@ -160,11 +160,11 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v64i8': ; SSE2: Found an estimated cost of 216 for instruction: %shift ; SSE41: Found an estimated cost of 96 for instruction: %shift -; AVX: Found an estimated cost of 96 for instruction: %shift +; AVX: Found an estimated cost of 100 for instruction: %shift ; AVX2: Found an estimated cost of 48 for instruction: %shift ; AVX512F: Found an estimated cost of 48 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = ashr <64 x i8> %a, %b ret <64 x i8> %shift } @@ -191,11 +191,10 @@ define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v4i64': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift -; AVX2: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift +; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift -; XOPAVX2: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %insert = insertelement <4 x i64> undef, i64 %b, i32 0 %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer %shift = ashr <4 x i64> %a, %splat @@ -206,11 +205,10 @@ define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v8i64': ; SSE2: Found an estimated cost of 16 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift -; AVX2: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift +; AVX2: Found an estimated cost of 8 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift -; XOPAVX2: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %insert = insertelement <8 x i64> undef, i64 %b, i32 0 %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer %shift = ashr <8 x i64> %a, %splat @@ -235,10 +233,10 @@ define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v8i32': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <8 x i32> undef, i32 %b, i32 0 %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer @@ -250,10 +248,10 @@ define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v16i32': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <16 x i32> undef, i32 %b, i32 0 %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer @@ -279,10 +277,10 @@ define <16 x i16> @splatvar_shift_v16i16(<16 x i16> %a, i16 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v16i16': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <16 x i16> undef, i16 %b, i32 0 %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer @@ -294,11 +292,11 @@ define <32 x i16> @splatvar_shift_v32i16(<32 x i16> %a, i16 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v32i16': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <32 x i16> undef, i16 %b, i32 0 %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer @@ -324,10 +322,10 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v32i8': ; SSE2: Found an estimated cost of 108 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 50 for instruction: %shift ; AVX2: Found an estimated cost of 24 for instruction: %shift ; AVX512: Found an estimated cost of 24 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %insert = insertelement <32 x i8> undef, i8 %b, i32 0 %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer %shift = ashr <32 x i8> %a, %splat @@ -338,11 +336,11 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v64i8': ; SSE2: Found an estimated cost of 216 for instruction: %shift ; SSE41: Found an estimated cost of 96 for instruction: %shift -; AVX: Found an estimated cost of 96 for instruction: %shift +; AVX: Found an estimated cost of 100 for instruction: %shift ; AVX2: Found an estimated cost of 48 for instruction: %shift ; AVX512F: Found an estimated cost of 48 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %insert = insertelement <64 x i8> undef, i8 %b, i32 0 %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer %shift = ashr <64 x i8> %a, %splat @@ -369,10 +367,10 @@ define <4 x i64> @constant_shift_v4i64(<4 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v4i64': ; SSE2: Found an estimated cost of 24 for instruction: %shift ; SSE41: Found an estimated cost of 24 for instruction: %shift -; AVX: Found an estimated cost of 24 for instruction: %shift +; AVX: Found an estimated cost of 26 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = ashr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31> ret <4 x i64> %shift } @@ -381,10 +379,10 @@ define <8 x i64> @constant_shift_v8i64(<8 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v8i64': ; SSE2: Found an estimated cost of 48 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 52 for instruction: %shift ; AVX2: Found an estimated cost of 8 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = ashr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31> ret <8 x i64> %shift } @@ -406,10 +404,10 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v8i32': ; SSE2: Found an estimated cost of 32 for instruction: %shift ; SSE41: Found an estimated cost of 24 for instruction: %shift -; AVX: Found an estimated cost of 24 for instruction: %shift +; AVX: Found an estimated cost of 26 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = ashr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3> ret <8 x i32> %shift @@ -419,10 +417,10 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v16i32': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 52 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = ashr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3> ret <16 x i32> %shift @@ -445,11 +443,11 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v16i16': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 28 for instruction: %shift -; AVX: Found an estimated cost of 28 for instruction: %shift +; AVX: Found an estimated cost of 30 for instruction: %shift ; AVX2: Found an estimated cost of 10 for instruction: %shift ; AVX512F: Found an estimated cost of 10 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = ashr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> ret <16 x i16> %shift } @@ -458,11 +456,11 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v32i16': ; SSE2: Found an estimated cost of 128 for instruction: %shift ; SSE41: Found an estimated cost of 56 for instruction: %shift -; AVX: Found an estimated cost of 56 for instruction: %shift +; AVX: Found an estimated cost of 60 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = ashr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> ret <32 x i16> %shift } @@ -483,10 +481,10 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v32i8': ; SSE2: Found an estimated cost of 108 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 50 for instruction: %shift ; AVX2: Found an estimated cost of 24 for instruction: %shift ; AVX512: Found an estimated cost of 24 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = ashr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> ret <32 x i8> %shift } @@ -495,11 +493,11 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v64i8': ; SSE2: Found an estimated cost of 216 for instruction: %shift ; SSE41: Found an estimated cost of 96 for instruction: %shift -; AVX: Found an estimated cost of 96 for instruction: %shift +; AVX: Found an estimated cost of 100 for instruction: %shift ; AVX2: Found an estimated cost of 48 for instruction: %shift ; AVX512F: Found an estimated cost of 48 for instruction: %shift ; AVX512BW: Found an estimated cost of 24 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = ashr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> ret <64 x i8> %shift } @@ -524,10 +522,11 @@ define <4 x i64> @splatconstant_shift_v4i64(<4 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v4i64': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift +; XOPAVX2: Found an estimated cost of 4 for instruction: %shift %shift = ashr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7> ret <4 x i64> %shift } @@ -536,10 +535,11 @@ define <8 x i64> @splatconstant_shift_v8i64(<8 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v8i64': ; SSE2: Found an estimated cost of 16 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 8 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift +; XOPAVX2: Found an estimated cost of 8 for instruction: %shift %shift = ashr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7> ret <8 x i64> %shift } @@ -560,10 +560,10 @@ define <8 x i32> @splatconstant_shift_v8i32(<8 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v8i32': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = ashr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5> ret <8 x i32> %shift @@ -573,10 +573,10 @@ define <16 x i32> @splatconstant_shift_v16i32(<16 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v16i32': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = ashr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5> ret <16 x i32> %shift @@ -598,10 +598,10 @@ define <16 x i16> @splatconstant_shift_v16i16(<16 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v16i16': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = ashr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> ret <16 x i16> %shift @@ -611,11 +611,11 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v32i16': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = ashr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> ret <32 x i16> %shift @@ -628,7 +628,7 @@ define <16 x i8> @splatconstant_shift_v16i8(<16 x i8> %a) { ; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512: Found an estimated cost of 4 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 2 for instruction: %shift %shift = ashr <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> ret <16 x i8> %shift } @@ -637,10 +637,10 @@ define <32 x i8> @splatconstant_shift_v32i8(<32 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v32i8': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512: Found an estimated cost of 4 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 4 for instruction: %shift %shift = ashr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> ret <32 x i8> %shift @@ -650,11 +650,11 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v64i8': ; SSE2: Found an estimated cost of 16 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 8 for instruction: %shift ; AVX512F: Found an estimated cost of 8 for instruction: %shift ; AVX512BW: Found an estimated cost of 4 for instruction: %shift -; XOPAVX: Found an estimated cost of 16 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 8 for instruction: %shift %shift = ashr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> ret <64 x i8> %shift diff --git a/test/Analysis/CostModel/X86/vshift-lshr-cost.ll b/test/Analysis/CostModel/X86/vshift-lshr-cost.ll index 546b2bb50f26f..6e890369d677e 100644 --- a/test/Analysis/CostModel/X86/vshift-lshr-cost.ll +++ b/test/Analysis/CostModel/X86/vshift-lshr-cost.ll @@ -34,10 +34,10 @@ define <4 x i64> @var_shift_v4i64(<4 x i64> %a, <4 x i64> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v4i64': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = lshr <4 x i64> %a, %b ret <4 x i64> %shift @@ -47,10 +47,10 @@ define <8 x i64> @var_shift_v8i64(<8 x i64> %a, <8 x i64> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v8i64': ; SSE2: Found an estimated cost of 16 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <8 x i64> %a, %b ret <8 x i64> %shift @@ -73,10 +73,10 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v8i32': ; SSE2: Found an estimated cost of 32 for instruction: %shift ; SSE41: Found an estimated cost of 22 for instruction: %shift -; AVX: Found an estimated cost of 22 for instruction: %shift +; AVX: Found an estimated cost of 24 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = lshr <8 x i32> %a, %b ret <8 x i32> %shift @@ -86,10 +86,10 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v16i32': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 44 for instruction: %shift -; AVX: Found an estimated cost of 44 for instruction: %shift +; AVX: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <16 x i32> %a, %b ret <16 x i32> %shift @@ -112,11 +112,11 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v16i16': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 28 for instruction: %shift -; AVX: Found an estimated cost of 28 for instruction: %shift +; AVX: Found an estimated cost of 30 for instruction: %shift ; AVX2: Found an estimated cost of 10 for instruction: %shift ; AVX512F: Found an estimated cost of 10 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = lshr <16 x i16> %a, %b ret <16 x i16> %shift } @@ -125,11 +125,11 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v32i16': ; SSE2: Found an estimated cost of 128 for instruction: %shift ; SSE41: Found an estimated cost of 56 for instruction: %shift -; AVX: Found an estimated cost of 56 for instruction: %shift +; AVX: Found an estimated cost of 60 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = lshr <32 x i16> %a, %b ret <32 x i16> %shift } @@ -150,10 +150,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v32i8': ; SSE2: Found an estimated cost of 52 for instruction: %shift ; SSE41: Found an estimated cost of 24 for instruction: %shift -; AVX: Found an estimated cost of 24 for instruction: %shift +; AVX: Found an estimated cost of 26 for instruction: %shift ; AVX2: Found an estimated cost of 11 for instruction: %shift ; AVX512: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = lshr <32 x i8> %a, %b ret <32 x i8> %shift } @@ -162,11 +162,11 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v64i8': ; SSE2: Found an estimated cost of 104 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 52 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = lshr <64 x i8> %a, %b ret <64 x i8> %shift } @@ -193,10 +193,10 @@ define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v4i64': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <4 x i64> undef, i64 %b, i32 0 %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer @@ -208,10 +208,10 @@ define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v8i64': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <8 x i64> undef, i64 %b, i32 0 %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer @@ -237,10 +237,10 @@ define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v8i32': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <8 x i32> undef, i32 %b, i32 0 %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer @@ -252,10 +252,10 @@ define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v16i32': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <16 x i32> undef, i32 %b, i32 0 %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer @@ -281,10 +281,10 @@ define <16 x i16> @splatvar_shift_v16i16(<16 x i16> %a, i16 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v16i16': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <16 x i16> undef, i16 %b, i32 0 %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer @@ -296,11 +296,11 @@ define <32 x i16> @splatvar_shift_v32i16(<32 x i16> %a, i16 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v32i16': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <32 x i16> undef, i16 %b, i32 0 %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer @@ -326,10 +326,10 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v32i8': ; SSE2: Found an estimated cost of 52 for instruction: %shift ; SSE41: Found an estimated cost of 24 for instruction: %shift -; AVX: Found an estimated cost of 24 for instruction: %shift +; AVX: Found an estimated cost of 26 for instruction: %shift ; AVX2: Found an estimated cost of 11 for instruction: %shift ; AVX512: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %insert = insertelement <32 x i8> undef, i8 %b, i32 0 %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer %shift = lshr <32 x i8> %a, %splat @@ -340,11 +340,11 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v64i8': ; SSE2: Found an estimated cost of 104 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 52 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %insert = insertelement <64 x i8> undef, i8 %b, i32 0 %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer %shift = lshr <64 x i8> %a, %splat @@ -372,10 +372,10 @@ define <4 x i64> @constant_shift_v4i64(<4 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v4i64': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = lshr <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31> ret <4 x i64> %shift @@ -385,10 +385,10 @@ define <8 x i64> @constant_shift_v8i64(<8 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v8i64': ; SSE2: Found an estimated cost of 16 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31> ret <8 x i64> %shift @@ -411,10 +411,10 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v8i32': ; SSE2: Found an estimated cost of 32 for instruction: %shift ; SSE41: Found an estimated cost of 22 for instruction: %shift -; AVX: Found an estimated cost of 22 for instruction: %shift +; AVX: Found an estimated cost of 24 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = lshr <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3> ret <8 x i32> %shift @@ -424,10 +424,10 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v16i32': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 44 for instruction: %shift -; AVX: Found an estimated cost of 44 for instruction: %shift +; AVX: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3> ret <16 x i32> %shift @@ -450,11 +450,11 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v16i16': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 28 for instruction: %shift -; AVX: Found an estimated cost of 28 for instruction: %shift +; AVX: Found an estimated cost of 30 for instruction: %shift ; AVX2: Found an estimated cost of 10 for instruction: %shift ; AVX512F: Found an estimated cost of 10 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = lshr <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> ret <16 x i16> %shift } @@ -463,11 +463,11 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v32i16': ; SSE2: Found an estimated cost of 128 for instruction: %shift ; SSE41: Found an estimated cost of 56 for instruction: %shift -; AVX: Found an estimated cost of 56 for instruction: %shift +; AVX: Found an estimated cost of 60 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = lshr <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> ret <32 x i16> %shift } @@ -488,10 +488,10 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v32i8': ; SSE2: Found an estimated cost of 52 for instruction: %shift ; SSE41: Found an estimated cost of 24 for instruction: %shift -; AVX: Found an estimated cost of 24 for instruction: %shift +; AVX: Found an estimated cost of 26 for instruction: %shift ; AVX2: Found an estimated cost of 11 for instruction: %shift ; AVX512: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 6 for instruction: %shift %shift = lshr <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> ret <32 x i8> %shift } @@ -500,11 +500,11 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v64i8': ; SSE2: Found an estimated cost of 104 for instruction: %shift ; SSE41: Found an estimated cost of 48 for instruction: %shift -; AVX: Found an estimated cost of 48 for instruction: %shift +; AVX: Found an estimated cost of 52 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 8 for instruction: %shift +; XOP: Found an estimated cost of 12 for instruction: %shift %shift = lshr <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> ret <64 x i8> %shift } @@ -529,10 +529,10 @@ define <4 x i64> @splatconstant_shift_v4i64(<4 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v4i64': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = lshr <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7> ret <4 x i64> %shift @@ -542,10 +542,10 @@ define <8 x i64> @splatconstant_shift_v8i64(<8 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v8i64': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7> ret <8 x i64> %shift @@ -567,10 +567,10 @@ define <8 x i32> @splatconstant_shift_v8i32(<8 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v8i32': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = lshr <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5> ret <8 x i32> %shift @@ -580,10 +580,10 @@ define <16 x i32> @splatconstant_shift_v16i32(<16 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v16i32': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5> ret <16 x i32> %shift @@ -605,10 +605,10 @@ define <16 x i16> @splatconstant_shift_v16i16(<16 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v16i16': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = lshr <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> ret <16 x i16> %shift @@ -618,11 +618,11 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v32i16': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> ret <32 x i16> %shift @@ -644,10 +644,10 @@ define <32 x i8> @splatconstant_shift_v32i8(<32 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v32i8': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 6 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 2 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 6 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = lshr <32 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> ret <32 x i8> %shift @@ -657,11 +657,11 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v64i8': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 12 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512F: Found an estimated cost of 4 for instruction: %shift ; AVX512BW: Found an estimated cost of 2 for instruction: %shift -; XOPAVX: Found an estimated cost of 8 for instruction: %shift +; XOPAVX: Found an estimated cost of 12 for instruction: %shift ; XOPAVX2: Found an estimated cost of 4 for instruction: %shift %shift = lshr <64 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> ret <64 x i8> %shift diff --git a/test/Analysis/CostModel/X86/vshift-shl-cost.ll b/test/Analysis/CostModel/X86/vshift-shl-cost.ll index 90356f5ce8be9..5e604bb7983e3 100644 --- a/test/Analysis/CostModel/X86/vshift-shl-cost.ll +++ b/test/Analysis/CostModel/X86/vshift-shl-cost.ll @@ -35,10 +35,10 @@ define <4 x i64> @var_shift_v4i64(<4 x i64> %a, <4 x i64> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v4i64': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <4 x i64> %a, %b ret <4 x i64> %shift @@ -48,10 +48,10 @@ define <8 x i64> @var_shift_v8i64(<8 x i64> %a, <8 x i64> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v8i64': ; SSE2: Found an estimated cost of 16 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <8 x i64> %a, %b ret <8 x i64> %shift @@ -74,10 +74,10 @@ define <8 x i32> @var_shift_v8i32(<8 x i32> %a, <8 x i32> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v8i32': ; SSE2: Found an estimated cost of 20 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <8 x i32> %a, %b ret <8 x i32> %shift @@ -87,10 +87,10 @@ define <16 x i32> @var_shift_v16i32(<16 x i32> %a, <16 x i32> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v16i32': ; SSE2: Found an estimated cost of 40 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <16 x i32> %a, %b ret <16 x i32> %shift @@ -113,11 +113,11 @@ define <16 x i16> @var_shift_v16i16(<16 x i16> %a, <16 x i16> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v16i16': ; SSE2: Found an estimated cost of 64 for instruction: %shift ; SSE41: Found an estimated cost of 28 for instruction: %shift -; AVX: Found an estimated cost of 28 for instruction: %shift +; AVX: Found an estimated cost of 30 for instruction: %shift ; AVX2: Found an estimated cost of 10 for instruction: %shift ; AVX512F: Found an estimated cost of 10 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 2 for instruction: %shift +; XOP: Found an estimated cost of 4 for instruction: %shift %shift = shl <16 x i16> %a, %b ret <16 x i16> %shift } @@ -126,11 +126,11 @@ define <32 x i16> @var_shift_v32i16(<32 x i16> %a, <32 x i16> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v32i16': ; SSE2: Found an estimated cost of 128 for instruction: %shift ; SSE41: Found an estimated cost of 56 for instruction: %shift -; AVX: Found an estimated cost of 56 for instruction: %shift +; AVX: Found an estimated cost of 60 for instruction: %shift ; AVX2: Found an estimated cost of 20 for instruction: %shift ; AVX512F: Found an estimated cost of 20 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 8 for instruction: %shift %shift = shl <32 x i16> %a, %b ret <32 x i16> %shift } @@ -151,10 +151,10 @@ define <32 x i8> @var_shift_v32i8(<32 x i8> %a, <32 x i8> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v32i8': ; SSE2: Found an estimated cost of 52 for instruction: %shift ; SSE41: Found an estimated cost of 22 for instruction: %shift -; AVX: Found an estimated cost of 22 for instruction: %shift +; AVX: Found an estimated cost of 24 for instruction: %shift ; AVX2: Found an estimated cost of 11 for instruction: %shift ; AVX512: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 2 for instruction: %shift +; XOP: Found an estimated cost of 4 for instruction: %shift %shift = shl <32 x i8> %a, %b ret <32 x i8> %shift } @@ -163,11 +163,11 @@ define <64 x i8> @var_shift_v64i8(<64 x i8> %a, <64 x i8> %b) { ; CHECK: 'Cost Model Analysis' for function 'var_shift_v64i8': ; SSE2: Found an estimated cost of 104 for instruction: %shift ; SSE41: Found an estimated cost of 44 for instruction: %shift -; AVX: Found an estimated cost of 44 for instruction: %shift +; AVX: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 8 for instruction: %shift %shift = shl <64 x i8> %a, %b ret <64 x i8> %shift } @@ -194,10 +194,10 @@ define <4 x i64> @splatvar_shift_v4i64(<4 x i64> %a, i64 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v4i64': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <4 x i64> undef, i64 %b, i32 0 %splat = shufflevector <4 x i64> %insert, <4 x i64> undef, <4 x i32> zeroinitializer @@ -209,10 +209,10 @@ define <8 x i64> @splatvar_shift_v8i64(<8 x i64> %a, i64 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v8i64': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <8 x i64> undef, i64 %b, i32 0 %splat = shufflevector <8 x i64> %insert, <8 x i64> undef, <8 x i32> zeroinitializer @@ -238,10 +238,10 @@ define <8 x i32> @splatvar_shift_v8i32(<8 x i32> %a, i32 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v8i32': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <8 x i32> undef, i32 %b, i32 0 %splat = shufflevector <8 x i32> %insert, <8 x i32> undef, <8 x i32> zeroinitializer @@ -253,10 +253,10 @@ define <16 x i32> @splatvar_shift_v16i32(<16 x i32> %a, i32 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v16i32': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <16 x i32> undef, i32 %b, i32 0 %splat = shufflevector <16 x i32> %insert, <16 x i32> undef, <16 x i32> zeroinitializer @@ -282,10 +282,10 @@ define <16 x i16> @splatvar_shift_v16i16(<16 x i16> %a, i16 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v16i16': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %insert = insertelement <16 x i16> undef, i16 %b, i32 0 %splat = shufflevector <16 x i16> %insert, <16 x i16> undef, <16 x i32> zeroinitializer @@ -297,11 +297,11 @@ define <32 x i16> @splatvar_shift_v32i16(<32 x i16> %a, i16 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v32i16': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %insert = insertelement <32 x i16> undef, i16 %b, i32 0 %splat = shufflevector <32 x i16> %insert, <32 x i16> undef, <32 x i32> zeroinitializer @@ -327,10 +327,10 @@ define <32 x i8> @splatvar_shift_v32i8(<32 x i8> %a, i8 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v32i8': ; SSE2: Found an estimated cost of 52 for instruction: %shift ; SSE41: Found an estimated cost of 22 for instruction: %shift -; AVX: Found an estimated cost of 22 for instruction: %shift +; AVX: Found an estimated cost of 24 for instruction: %shift ; AVX2: Found an estimated cost of 11 for instruction: %shift ; AVX512: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 2 for instruction: %shift +; XOP: Found an estimated cost of 4 for instruction: %shift %insert = insertelement <32 x i8> undef, i8 %b, i32 0 %splat = shufflevector <32 x i8> %insert, <32 x i8> undef, <32 x i32> zeroinitializer %shift = shl <32 x i8> %a, %splat @@ -341,11 +341,11 @@ define <64 x i8> @splatvar_shift_v64i8(<64 x i8> %a, i8 %b) { ; CHECK: 'Cost Model Analysis' for function 'splatvar_shift_v64i8': ; SSE2: Found an estimated cost of 104 for instruction: %shift ; SSE41: Found an estimated cost of 44 for instruction: %shift -; AVX: Found an estimated cost of 44 for instruction: %shift +; AVX: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 8 for instruction: %shift %insert = insertelement <64 x i8> undef, i8 %b, i32 0 %splat = shufflevector <64 x i8> %insert, <64 x i8> undef, <64 x i32> zeroinitializer %shift = shl <64 x i8> %a, %splat @@ -373,10 +373,10 @@ define <4 x i64> @constant_shift_v4i64(<4 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v4i64': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 10 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <4 x i64> %a, <i64 1, i64 7, i64 15, i64 31> ret <4 x i64> %shift @@ -386,10 +386,10 @@ define <8 x i64> @constant_shift_v8i64(<8 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v8i64': ; SSE2: Found an estimated cost of 16 for instruction: %shift ; SSE41: Found an estimated cost of 16 for instruction: %shift -; AVX: Found an estimated cost of 16 for instruction: %shift +; AVX: Found an estimated cost of 20 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <8 x i64> %a, <i64 1, i64 7, i64 15, i64 31, i64 1, i64 7, i64 15, i64 31> ret <8 x i64> %shift @@ -415,7 +415,7 @@ define <8 x i32> @constant_shift_v8i32(<8 x i32> %a) { ; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <8 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3> ret <8 x i32> %shift @@ -428,7 +428,7 @@ define <16 x i32> @constant_shift_v16i32(<16 x i32> %a) { ; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <16 x i32> %a, <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3> ret <16 x i32> %shift @@ -453,7 +453,7 @@ define <16 x i16> @constant_shift_v16i16(<16 x i16> %a) { ; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <16 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> ret <16 x i16> %shift @@ -467,7 +467,7 @@ define <32 x i16> @constant_shift_v32i16(<32 x i16> %a) { ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <32 x i16> %a, <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7, i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7> ret <32 x i16> %shift @@ -489,10 +489,10 @@ define <32 x i8> @constant_shift_v32i8(<32 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v32i8': ; SSE2: Found an estimated cost of 52 for instruction: %shift ; SSE41: Found an estimated cost of 22 for instruction: %shift -; AVX: Found an estimated cost of 22 for instruction: %shift +; AVX: Found an estimated cost of 24 for instruction: %shift ; AVX2: Found an estimated cost of 11 for instruction: %shift ; AVX512: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 2 for instruction: %shift +; XOP: Found an estimated cost of 4 for instruction: %shift %shift = shl <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> ret <32 x i8> %shift } @@ -501,11 +501,11 @@ define <64 x i8> @constant_shift_v64i8(<64 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'constant_shift_v64i8': ; SSE2: Found an estimated cost of 104 for instruction: %shift ; SSE41: Found an estimated cost of 44 for instruction: %shift -; AVX: Found an estimated cost of 44 for instruction: %shift +; AVX: Found an estimated cost of 48 for instruction: %shift ; AVX2: Found an estimated cost of 22 for instruction: %shift ; AVX512F: Found an estimated cost of 22 for instruction: %shift ; AVX512BW: Found an estimated cost of 11 for instruction: %shift -; XOP: Found an estimated cost of 4 for instruction: %shift +; XOP: Found an estimated cost of 8 for instruction: %shift %shift = shl <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0> ret <64 x i8> %shift } @@ -531,10 +531,10 @@ define <4 x i64> @splatconstant_shift_v4i64(<4 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v4i64': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <4 x i64> %a, <i64 7, i64 7, i64 7, i64 7> ret <4 x i64> %shift @@ -544,10 +544,10 @@ define <8 x i64> @splatconstant_shift_v8i64(<8 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v8i64': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7> ret <8 x i64> %shift @@ -570,10 +570,10 @@ define <8 x i32> @splatconstant_shift_v8i32(<8 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v8i32': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <8 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5> ret <8 x i32> %shift @@ -583,10 +583,10 @@ define <16 x i32> @splatconstant_shift_v16i32(<16 x i32> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v16i32': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5> ret <16 x i32> %shift @@ -608,10 +608,10 @@ define <16 x i16> @splatconstant_shift_v16i16(<16 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v16i16': ; SSE2: Found an estimated cost of 2 for instruction: %shift ; SSE41: Found an estimated cost of 2 for instruction: %shift -; AVX: Found an estimated cost of 2 for instruction: %shift +; AVX: Found an estimated cost of 4 for instruction: %shift ; AVX2: Found an estimated cost of 1 for instruction: %shift ; AVX512: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 2 for instruction: %shift +; XOPAVX: Found an estimated cost of 4 for instruction: %shift ; XOPAVX2: Found an estimated cost of 1 for instruction: %shift %shift = shl <16 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> ret <16 x i16> %shift @@ -621,11 +621,11 @@ define <32 x i16> @splatconstant_shift_v32i16(<32 x i16> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v32i16': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 8 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512F: Found an estimated cost of 2 for instruction: %shift ; AVX512BW: Found an estimated cost of 1 for instruction: %shift -; XOPAVX: Found an estimated cost of 4 for instruction: %shift +; XOPAVX: Found an estimated cost of 8 for instruction: %shift ; XOPAVX2: Found an estimated cost of 2 for instruction: %shift %shift = shl <32 x i16> %a, <i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3, i16 3> ret <32 x i16> %shift @@ -638,7 +638,7 @@ define <16 x i8> @splatconstant_shift_v16i8(<16 x i8> %a) { ; AVX: Found an estimated cost of 2 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 2 for instruction: %shift -; XOP: Found an estimated cost of 2 for instruction: %shift +; XOP: Found an estimated cost of 1 for instruction: %shift %shift = shl <16 x i8> %a, <i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3, i8 3> ret <16 x i8> %shift } @@ -647,7 +647,7 @@ define <32 x i8> @splatconstant_shift_v32i8(<32 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v32i8': ; SSE2: Found an estimated cost of 4 for instruction: %shift ; SSE41: Found an estimated cost of 4 for instruction: %shift -; AVX: Found an estimated cost of 4 for instruction: %shift +; AVX: Found an estimated cost of 6 for instruction: %shift ; AVX2: Found an estimated cost of 2 for instruction: %shift ; AVX512: Found an estimated cost of 2 for instruction: %shift ; XOPAVX: Found an estimated cost of 4 for instruction: %shift @@ -660,7 +660,7 @@ define <64 x i8> @splatconstant_shift_v64i8(<64 x i8> %a) { ; CHECK: 'Cost Model Analysis' for function 'splatconstant_shift_v64i8': ; SSE2: Found an estimated cost of 8 for instruction: %shift ; SSE41: Found an estimated cost of 8 for instruction: %shift -; AVX: Found an estimated cost of 8 for instruction: %shift +; AVX: Found an estimated cost of 12 for instruction: %shift ; AVX2: Found an estimated cost of 4 for instruction: %shift ; AVX512F: Found an estimated cost of 4 for instruction: %shift ; AVX512BW: Found an estimated cost of 2 for instruction: %shift @@ -761,7 +761,7 @@ define <16 x i16> @test6(<16 x i16> %a) { ; SSE41: Found an estimated cost of 2 for instruction: %shl ; AVX: Found an estimated cost of 4 for instruction: %shl ; AVX2: Found an estimated cost of 1 for instruction: %shl -; XOPAVX: Found an estimated cost of 2 for instruction: %shl +; XOPAVX: Found an estimated cost of 4 for instruction: %shl ; XOPAVX2: Found an estimated cost of 1 for instruction: %shl @@ -778,7 +778,7 @@ define <8 x i32> @test7(<8 x i32> %a) { ; SSE41: Found an estimated cost of 2 for instruction: %shl ; AVX: Found an estimated cost of 4 for instruction: %shl ; AVX2: Found an estimated cost of 1 for instruction: %shl -; XOPAVX: Found an estimated cost of 2 for instruction: %shl +; XOPAVX: Found an estimated cost of 4 for instruction: %shl ; XOPAVX2: Found an estimated cost of 1 for instruction: %shl @@ -794,9 +794,9 @@ define <4 x i64> @test8(<4 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'test8': ; SSE2: Found an estimated cost of 8 for instruction: %shl ; SSE41: Found an estimated cost of 8 for instruction: %shl -; AVX: Found an estimated cost of 8 for instruction: %shl +; AVX: Found an estimated cost of 10 for instruction: %shl ; AVX2: Found an estimated cost of 1 for instruction: %shl -; XOPAVX: Found an estimated cost of 2 for instruction: %shl +; XOPAVX: Found an estimated cost of 4 for instruction: %shl ; XOPAVX2: Found an estimated cost of 1 for instruction: %shl @@ -811,7 +811,7 @@ define <32 x i16> @test9(<32 x i16> %a) { ; SSE41: Found an estimated cost of 4 for instruction: %shl ; AVX: Found an estimated cost of 8 for instruction: %shl ; AVX2: Found an estimated cost of 2 for instruction: %shl -; XOPAVX: Found an estimated cost of 4 for instruction: %shl +; XOPAVX: Found an estimated cost of 8 for instruction: %shl ; XOPAVX2: Found an estimated cost of 2 for instruction: %shl @@ -826,7 +826,7 @@ define <16 x i32> @test10(<16 x i32> %a) { ; SSE41: Found an estimated cost of 4 for instruction: %shl ; AVX: Found an estimated cost of 8 for instruction: %shl ; AVX2: Found an estimated cost of 2 for instruction: %shl -; XOPAVX: Found an estimated cost of 4 for instruction: %shl +; XOPAVX: Found an estimated cost of 8 for instruction: %shl ; XOPAVX2: Found an estimated cost of 2 for instruction: %shl @@ -842,7 +842,7 @@ define <8 x i64> @test11(<8 x i64> %a) { ; CHECK: 'Cost Model Analysis' for function 'test11': ; SSE2: Found an estimated cost of 16 for instruction: %shl ; SSE41: Found an estimated cost of 16 for instruction: %shl -; AVX: Found an estimated cost of 16 for instruction: %shl +; AVX: Found an estimated cost of 20 for instruction: %shl ; AVX2: Found an estimated cost of 2 for instruction: %shl -; XOPAVX: Found an estimated cost of 4 for instruction: %shl +; XOPAVX: Found an estimated cost of 8 for instruction: %shl ; XOPAVX2: Found an estimated cost of 2 for instruction: %shl |