aboutsummaryrefslogtreecommitdiff
path: root/lib/Target/AMDGPU/VOP1Instructions.td
diff options
context:
space:
mode:
Diffstat (limited to 'lib/Target/AMDGPU/VOP1Instructions.td')
-rw-r--r--lib/Target/AMDGPU/VOP1Instructions.td65
1 files changed, 34 insertions, 31 deletions
diff --git a/lib/Target/AMDGPU/VOP1Instructions.td b/lib/Target/AMDGPU/VOP1Instructions.td
index 6bc416ed7d4b..f1cdc3097dc0 100644
--- a/lib/Target/AMDGPU/VOP1Instructions.td
+++ b/lib/Target/AMDGPU/VOP1Instructions.td
@@ -104,9 +104,21 @@ multiclass VOP1Inst <string opName, VOPProfile P,
SDPatternOperator node = null_frag> {
def _e32 : VOP1_Pseudo <opName, P>;
def _e64 : VOP3_Pseudo <opName, P, getVOP1Pat64<node, P>.ret>;
- def _sdwa : VOP1_SDWA_Pseudo <opName, P>;
+
+ foreach _ = BoolToList<P.HasExtSDWA>.ret in
+ def _sdwa : VOP1_SDWA_Pseudo <opName, P>;
+
foreach _ = BoolToList<P.HasExtDPP>.ret in
def _dpp : VOP1_DPP_Pseudo <opName, P>;
+
+ def : MnemonicAlias<opName#"_e32", opName>, LetDummies;
+ def : MnemonicAlias<opName#"_e64", opName>, LetDummies;
+
+ foreach _ = BoolToList<P.HasExtSDWA>.ret in
+ def : MnemonicAlias<opName#"_sdwa", opName>, LetDummies;
+
+ foreach _ = BoolToList<P.HasExtDPP>.ret in
+ def : MnemonicAlias<opName#"_dpp", opName>, LetDummies;
}
// Special profile for instructions which have clamp
@@ -227,10 +239,10 @@ defm V_COS_F32 : VOP1Inst <"v_cos_f32", VOP_F32_F32, AMDGPUcos>;
} // End SchedRW = [WriteQuarterRate32]
defm V_NOT_B32 : VOP1Inst <"v_not_b32", VOP_I32_I32>;
-defm V_BFREV_B32 : VOP1Inst <"v_bfrev_b32", VOP_I32_I32>;
-defm V_FFBH_U32 : VOP1Inst <"v_ffbh_u32", VOP_I32_I32>;
+defm V_BFREV_B32 : VOP1Inst <"v_bfrev_b32", VOP_I32_I32, bitreverse>;
+defm V_FFBH_U32 : VOP1Inst <"v_ffbh_u32", VOP_I32_I32, AMDGPUffbh_u32>;
defm V_FFBL_B32 : VOP1Inst <"v_ffbl_b32", VOP_I32_I32>;
-defm V_FFBH_I32 : VOP1Inst <"v_ffbh_i32", VOP_I32_I32>;
+defm V_FFBH_I32 : VOP1Inst <"v_ffbh_i32", VOP_I32_I32, AMDGPUffbh_i32>;
let SchedRW = [WriteDoubleAdd] in {
defm V_FREXP_EXP_I32_F64 : VOP1Inst <"v_frexp_exp_i32_f64", VOP_I32_F64, int_amdgcn_frexp_exp>;
@@ -434,7 +446,7 @@ let SubtargetPredicate = isGFX10Plus in {
// Target-specific instruction encodings.
//===----------------------------------------------------------------------===//
-class VOP1_DPP<bits<8> op, VOP1_Pseudo ps, VOPProfile p = ps.Pfl, bit isDPP16 = 0> :
+class VOP1_DPP<bits<8> op, VOP1_DPP_Pseudo ps, VOPProfile p = ps.Pfl, bit isDPP16 = 0> :
VOP_DPP<ps.OpName, p, isDPP16> {
let hasSideEffects = ps.hasSideEffects;
let Defs = ps.Defs;
@@ -448,8 +460,9 @@ class VOP1_DPP<bits<8> op, VOP1_Pseudo ps, VOPProfile p = ps.Pfl, bit isDPP16 =
let Inst{31-25} = 0x3f;
}
-class VOP1_DPP16<bits<8> op, VOP1_Pseudo ps, VOPProfile p = ps.Pfl> :
- VOP1_DPP<op, ps, p, 1> {
+class VOP1_DPP16<bits<8> op, VOP1_DPP_Pseudo ps, VOPProfile p = ps.Pfl> :
+ VOP1_DPP<op, ps, p, 1>,
+ SIMCInstr <ps.PseudoInstr, SIEncodingFamily.GFX10> {
let AssemblerPredicate = !if(p.HasExt, HasDPP16, DisableInst);
let SubtargetPredicate = HasDPP16;
}
@@ -492,6 +505,7 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
VOP3e_gfx10<{0, 1, 1, op{6-0}}, !cast<VOP3_Pseudo>(NAME#"_e64").Pfl>;
}
multiclass VOP1_Real_sdwa_gfx10<bits<9> op> {
+ foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
def _sdwa_gfx10 :
VOP_SDWA10_Real<!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
VOP1_SDWA9Ae<op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl> {
@@ -499,11 +513,13 @@ let AssemblerPredicate = isGFX10Plus, DecoderNamespace = "GFX10" in {
}
}
multiclass VOP1_Real_dpp_gfx10<bits<9> op> {
- def _dpp_gfx10 : VOP1_DPP16<op{7-0}, !cast<VOP1_Pseudo>(NAME#"_e32")> {
+ foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtDPP>.ret in
+ def _dpp_gfx10 : VOP1_DPP16<op{7-0}, !cast<VOP1_DPP_Pseudo>(NAME#"_dpp")> {
let DecoderNamespace = "SDWA10";
}
}
multiclass VOP1_Real_dpp8_gfx10<bits<9> op> {
+ foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtDPP>.ret in
def _dpp8_gfx10 : VOP1_DPP8<op{7-0}, !cast<VOP1_Pseudo>(NAME#"_e32")> {
let DecoderNamespace = "DPP8";
}
@@ -704,10 +720,12 @@ multiclass VOP1_Real_e32e64_vi <bits<10> op> {
multiclass VOP1_Real_vi <bits<10> op> {
defm NAME : VOP1_Real_e32e64_vi <op>;
+ foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA>.ret in
def _sdwa_vi :
VOP_SDWA_Real <!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
VOP1_SDWAe <op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
+ foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
def _sdwa_gfx9 :
VOP_SDWA9_Real <!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
VOP1_SDWA9Ae <op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
@@ -831,25 +849,25 @@ def V_MOVRELD_B32_V4 : V_MOVRELD_B32_pseudo<VReg_128>;
def V_MOVRELD_B32_V8 : V_MOVRELD_B32_pseudo<VReg_256>;
def V_MOVRELD_B32_V16 : V_MOVRELD_B32_pseudo<VReg_512>;
-let OtherPredicates = [isGFX8GFX9] in {
+let OtherPredicates = [isGFX8Plus] in {
def : GCNPat <
- (i32 (int_amdgcn_mov_dpp i32:$src, imm:$dpp_ctrl, imm:$row_mask, imm:$bank_mask,
- imm:$bound_ctrl)),
+ (i32 (int_amdgcn_mov_dpp i32:$src, timm:$dpp_ctrl, timm:$row_mask, timm:$bank_mask,
+ timm:$bound_ctrl)),
(V_MOV_B32_dpp $src, $src, (as_i32imm $dpp_ctrl),
(as_i32imm $row_mask), (as_i32imm $bank_mask),
(as_i1imm $bound_ctrl))
>;
def : GCNPat <
- (i32 (int_amdgcn_update_dpp i32:$old, i32:$src, imm:$dpp_ctrl, imm:$row_mask,
- imm:$bank_mask, imm:$bound_ctrl)),
+ (i32 (int_amdgcn_update_dpp i32:$old, i32:$src, timm:$dpp_ctrl, timm:$row_mask,
+ timm:$bank_mask, timm:$bound_ctrl)),
(V_MOV_B32_dpp $old, $src, (as_i32imm $dpp_ctrl),
(as_i32imm $row_mask), (as_i32imm $bank_mask),
(as_i1imm $bound_ctrl))
>;
-} // End OtherPredicates = [isGFX8GFX9]
+} // End OtherPredicates = [isGFX8Plus]
let OtherPredicates = [isGFX8Plus] in {
def : GCNPat<
@@ -885,6 +903,7 @@ multiclass VOP1_Real_gfx9 <bits<10> op> {
defm NAME : VOP1_Real_e32e64_vi <op>;
}
+ foreach _ = BoolToList<!cast<VOP1_Pseudo>(NAME#"_e32").Pfl.HasExtSDWA9>.ret in
def _sdwa_gfx9 :
VOP_SDWA9_Real <!cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa")>,
VOP1_SDWA9Ae <op{7-0}, !cast<VOP1_SDWA_Pseudo>(NAME#"_sdwa").Pfl>;
@@ -904,23 +923,7 @@ defm V_SCREEN_PARTITION_4SE_B32 : VOP1_Real_gfx9 <0x37>;
let OtherPredicates = [isGFX10Plus] in {
def : GCNPat <
- (i32 (int_amdgcn_mov_dpp8 i32:$src, imm:$dpp8)),
+ (i32 (int_amdgcn_mov_dpp8 i32:$src, timm:$dpp8)),
(V_MOV_B32_dpp8_gfx10 $src, $src, (as_i32imm $dpp8), (i32 DPP8Mode.FI_0))
>;
-
-def : GCNPat <
- (i32 (int_amdgcn_mov_dpp i32:$src, imm:$dpp_ctrl, imm:$row_mask, imm:$bank_mask,
- imm:$bound_ctrl)),
- (V_MOV_B32_dpp_gfx10 $src, $src, (as_i32imm $dpp_ctrl),
- (as_i32imm $row_mask), (as_i32imm $bank_mask),
- (as_i1imm $bound_ctrl), (i32 0))
->;
-
-def : GCNPat <
- (i32 (int_amdgcn_update_dpp i32:$old, i32:$src, imm:$dpp_ctrl, imm:$row_mask,
- imm:$bank_mask, imm:$bound_ctrl)),
- (V_MOV_B32_dpp_gfx10 $old, $src, (as_i32imm $dpp_ctrl),
- (as_i32imm $row_mask), (as_i32imm $bank_mask),
- (as_i1imm $bound_ctrl), (i32 0))
->;
} // End OtherPredicates = [isGFX10Plus]
generated by cgit v1.2.3 (git 2.25.1) at 2025-07-31 06:31:31 +0000