diff options
Diffstat (limited to 'include/llvm/IR/IntrinsicsNVVM.td')
| -rw-r--r-- | include/llvm/IR/IntrinsicsNVVM.td | 84 |
1 files changed, 28 insertions, 56 deletions
diff --git a/include/llvm/IR/IntrinsicsNVVM.td b/include/llvm/IR/IntrinsicsNVVM.td index f035ac3c90ee..68f123df0430 100644 --- a/include/llvm/IR/IntrinsicsNVVM.td +++ b/include/llvm/IR/IntrinsicsNVVM.td @@ -11,6 +11,27 @@ // //===----------------------------------------------------------------------===// +// The following intrinsics were once defined here, but are now auto-upgraded +// to target-generic LLVM intrinsics. +// +// * llvm.nvvm.brev32 --> llvm.bitreverse.i32 +// * llvm.nvvm.brev64 --> llvm.bitreverse.i64 +// * llvm.nvvm.clz.i --> llvm.ctlz.i32 +// * llvm.nvvm.clz.ll --> trunc i64 llvm.ctlz.i64(x) to i32 +// * llvm.nvvm.popc.i --> llvm.ctpop.i32 +// * llvm.nvvm.popc.ll --> trunc i64 llvm.ctpop.i64 to i32 +// * llvm.nvvm.abs.i --> select(x >= -x, x, -x) +// * llvm.nvvm.abs.ll --> ibid. +// * llvm.nvvm.max.i --> select(x sge y, x, y) +// * llvm.nvvm.max.ll --> ibid. +// * llvm.nvvm.max.ui --> select(x uge y, x, y) +// * llvm.nvvm.max.ull --> ibid. +// * llvm.nvvm.max.i --> select(x sle y, x, y) +// * llvm.nvvm.max.ll --> ibid. +// * llvm.nvvm.max.ui --> select(x ule y, x, y) +// * llvm.nvvm.max.ull --> ibid. +// * llvm.nvvm.h2f --> llvm.convert.to.fp16.f32 + def llvm_anyi64ptr_ty : LLVMAnyPointerType<llvm_i64_ty>; // (space)i64* // @@ -18,16 +39,6 @@ def llvm_anyi64ptr_ty : LLVMAnyPointerType<llvm_i64_ty>; // (space)i64* // let TargetPrefix = "nvvm" in { - def int_nvvm_clz_i : GCCBuiltin<"__nvvm_clz_i">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem]>; - def int_nvvm_clz_ll : GCCBuiltin<"__nvvm_clz_ll">, - Intrinsic<[llvm_i32_ty], [llvm_i64_ty], [IntrNoMem]>; - - def int_nvvm_popc_i : GCCBuiltin<"__nvvm_popc_i">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem]>; - def int_nvvm_popc_ll : GCCBuiltin<"__nvvm_popc_ll">, - Intrinsic<[llvm_i32_ty], [llvm_i64_ty], [IntrNoMem]>; - def int_nvvm_prmt : GCCBuiltin<"__nvvm_prmt">, Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty], [IntrNoMem, Commutative]>; @@ -36,34 +47,6 @@ let TargetPrefix = "nvvm" in { // Min Max // - def int_nvvm_min_i : GCCBuiltin<"__nvvm_min_i">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], - [IntrNoMem, Commutative]>; - def int_nvvm_min_ui : GCCBuiltin<"__nvvm_min_ui">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], - [IntrNoMem, Commutative]>; - - def int_nvvm_min_ll : GCCBuiltin<"__nvvm_min_ll">, - Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty], - [IntrNoMem, Commutative]>; - def int_nvvm_min_ull : GCCBuiltin<"__nvvm_min_ull">, - Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty], - [IntrNoMem, Commutative]>; - - def int_nvvm_max_i : GCCBuiltin<"__nvvm_max_i">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], - [IntrNoMem, Commutative]>; - def int_nvvm_max_ui : GCCBuiltin<"__nvvm_max_ui">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty], - [IntrNoMem, Commutative]>; - - def int_nvvm_max_ll : GCCBuiltin<"__nvvm_max_ll">, - Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty], - [IntrNoMem, Commutative]>; - def int_nvvm_max_ull : GCCBuiltin<"__nvvm_max_ull">, - Intrinsic<[llvm_i64_ty], [llvm_i64_ty, llvm_i64_ty], - [IntrNoMem, Commutative]>; - def int_nvvm_fmin_f : GCCBuiltin<"__nvvm_fmin_f">, Intrinsic<[llvm_float_ty], [llvm_float_ty, llvm_float_ty], [IntrNoMem, Commutative]>; @@ -201,15 +184,6 @@ let TargetPrefix = "nvvm" in { [IntrNoMem, Commutative]>; // -// Brev -// - - def int_nvvm_brev32 : GCCBuiltin<"__nvvm_brev32">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem]>; - def int_nvvm_brev64 : GCCBuiltin<"__nvvm_brev64">, - Intrinsic<[llvm_i64_ty], [llvm_i64_ty], [IntrNoMem]>; - -// // Sad // @@ -242,16 +216,10 @@ let TargetPrefix = "nvvm" in { // Abs // - def int_nvvm_abs_i : GCCBuiltin<"__nvvm_abs_i">, - Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrNoMem]>; - def int_nvvm_abs_ll : GCCBuiltin<"__nvvm_abs_ll">, - Intrinsic<[llvm_i64_ty], [llvm_i64_ty], [IntrNoMem]>; - def int_nvvm_fabs_ftz_f : GCCBuiltin<"__nvvm_fabs_ftz_f">, Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>; def int_nvvm_fabs_f : GCCBuiltin<"__nvvm_fabs_f">, Intrinsic<[llvm_float_ty], [llvm_float_ty], [IntrNoMem]>; - def int_nvvm_fabs_d : GCCBuiltin<"__nvvm_fabs_d">, Intrinsic<[llvm_double_ty], [llvm_double_ty], [IntrNoMem]>; @@ -700,9 +668,6 @@ let TargetPrefix = "nvvm" in { def int_nvvm_f2h_rn : GCCBuiltin<"__nvvm_f2h_rn">, Intrinsic<[llvm_i16_ty], [llvm_float_ty], [IntrNoMem]>; - def int_nvvm_h2f : GCCBuiltin<"__nvvm_h2f">, - Intrinsic<[llvm_float_ty], [llvm_i16_ty], [IntrNoMem]>; - // // Bitcast // @@ -768,6 +733,13 @@ let TargetPrefix = "nvvm" in { // intrinsics in this file, this one is a user-facing API. def int_nvvm_barrier0 : GCCBuiltin<"__syncthreads">, Intrinsic<[], [], [IntrConvergent]>; + // Synchronize all threads in the CTA at barrier 'n'. + def int_nvvm_barrier_n : GCCBuiltin<"__nvvm_bar_n">, + Intrinsic<[], [llvm_i32_ty], [IntrConvergent]>; + // Synchronize 'm', a multiple of warp size, (arg 2) threads in + // the CTA at barrier 'n' (arg 1). + def int_nvvm_barrier : GCCBuiltin<"__nvvm_bar">, + Intrinsic<[], [llvm_i32_ty, llvm_i32_ty], [IntrConvergent]>; def int_nvvm_barrier0_popc : GCCBuiltin<"__nvvm_bar0_popc">, Intrinsic<[llvm_i32_ty], [llvm_i32_ty], [IntrConvergent]>; def int_nvvm_barrier0_and : GCCBuiltin<"__nvvm_bar0_and">, |