RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-19 11:02:59 +02:00
Code Issues Projects Releases Wiki Activity

[Clang][NVPTX] Add NVPTX intrinsics and builtins for CUDA PTX redux.sync instructions

Browse Source 
Adds NVPTX builtins and intrinsics for the CUDA PTX `redux.sync` instructions
for `sm_80` architecture or newer.

PTX ISA description of `redux.sync`:
https://docs.nvidia.com/cuda/parallel-thread-execution/index.html#parallel-synchronization-and-communication-instructions-redux-sync

Authored-by: Steffen Larsen <steffen.larsen@codeplay.com>

Differential Revision: https://reviews.llvm.org/D100124
This commit is contained in:
Steffen Larsen 2021-05-17 09:23:44 -07:00 committed by Artem Belevich
parent 4b94b88699
commit 1e7a7bb573
3 changed files with 124 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

43
include/llvm/IR/IntrinsicsNVVM.td
View File

@ -4204,6 +4204,49 @@ def int_nvvm_match_all_sync_i64p :
Intrinsic<[llvm_i64_ty, llvm_i1_ty], [llvm_i32_ty, llvm_i64_ty], Intrinsic<[llvm_i64_ty, llvm_i1_ty], [llvm_i32_ty, llvm_i64_ty],
[IntrInaccessibleMemOnly, IntrConvergent], "llvm.nvvm.match.all.sync.i64p">; [IntrInaccessibleMemOnly, IntrConvergent], "llvm.nvvm.match.all.sync.i64p">;
//
// REDUX.SYNC
//
// redux.sync.min.u32 dst, src, membermask;
def int_nvvm_redux_sync_umin : GCCBuiltin<"__nvvm_redux_sync_umin">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// redux.sync.max.u32 dst, src, membermask;
def int_nvvm_redux_sync_umax : GCCBuiltin<"__nvvm_redux_sync_umax">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// redux.sync.add.s32 dst, src, membermask;
def int_nvvm_redux_sync_add : GCCBuiltin<"__nvvm_redux_sync_add">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// redux.sync.min.s32 dst, src, membermask;
def int_nvvm_redux_sync_min : GCCBuiltin<"__nvvm_redux_sync_min">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// redux.sync.max.s32 dst, src, membermask;
def int_nvvm_redux_sync_max : GCCBuiltin<"__nvvm_redux_sync_max">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// redux.sync.and.b32 dst, src, membermask;
def int_nvvm_redux_sync_and : GCCBuiltin<"__nvvm_redux_sync_and">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// redux.sync.xor.b32 dst, src, membermask;
def int_nvvm_redux_sync_xor : GCCBuiltin<"__nvvm_redux_sync_xor">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// redux.sync.or.b32 dst, src, membermask;
def int_nvvm_redux_sync_or : GCCBuiltin<"__nvvm_redux_sync_or">,
Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty],
[IntrConvergent, IntrInaccessibleMemOnly]>;
// //
// WMMA instructions // WMMA instructions
// //

16
lib/Target/NVPTX/NVPTXIntrinsics.td
View File

@ -274,6 +274,22 @@ defm MATCH_ALLP_SYNC_32 : MATCH_ALLP_SYNC<Int32Regs, "b32", int_nvvm_match_all_s
defm MATCH_ALLP_SYNC_64 : MATCH_ALLP_SYNC<Int64Regs, "b64", int_nvvm_match_all_sync_i64p, defm MATCH_ALLP_SYNC_64 : MATCH_ALLP_SYNC<Int64Regs, "b64", int_nvvm_match_all_sync_i64p,
i64imm>; i64imm>;
multiclass REDUX_SYNC<string BinOp, string PTXType, Intrinsic Intrin> {
def : NVPTXInst<(outs Int32Regs:$dst), (ins Int32Regs:$src, Int32Regs:$mask),
"redux.sync." # BinOp # "." # PTXType # " $dst, $src, $mask;",
[(set Int32Regs:$dst, (Intrin Int32Regs:$src, Int32Regs:$mask))]>,
Requires<[hasPTX70, hasSM80]>;
}
defm REDUX_SYNC_UMIN : REDUX_SYNC<"min", "u32", int_nvvm_redux_sync_umin>;
defm REDUX_SYNC_UMAX : REDUX_SYNC<"max", "u32", int_nvvm_redux_sync_umax>;
defm REDUX_SYNC_ADD : REDUX_SYNC<"add", "s32", int_nvvm_redux_sync_add>;
defm REDUX_SYNC_MIN : REDUX_SYNC<"min", "s32", int_nvvm_redux_sync_min>;
defm REDUX_SYNC_MAX : REDUX_SYNC<"max", "s32", int_nvvm_redux_sync_max>;
defm REDUX_SYNC_AND : REDUX_SYNC<"and", "b32", int_nvvm_redux_sync_and>;
defm REDUX_SYNC_XOR : REDUX_SYNC<"xor", "b32", int_nvvm_redux_sync_xor>;
defm REDUX_SYNC_OR : REDUX_SYNC<"or", "b32", int_nvvm_redux_sync_or>;
} // isConvergent = true } // isConvergent = true
//----------------------------------- //-----------------------------------

65
test/CodeGen/NVPTX/redux-sync.ll Normal file
View File

@ -0,0 +1,65 @@
; RUN: llc < %s -march=nvptx64 -mcpu=sm_80 -mattr=+ptx70 | FileCheck %s
declare i32 @llvm.nvvm.redux.sync.umin(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_min_u32
define i32 @redux_sync_min_u32(i32 %src, i32 %mask) {
; CHECK: redux.sync.min.u32
%val = call i32 @llvm.nvvm.redux.sync.umin(i32 %src, i32 %mask)
ret i32 %val
}
declare i32 @llvm.nvvm.redux.sync.umax(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_max_u32
define i32 @redux_sync_max_u32(i32 %src, i32 %mask) {
; CHECK: redux.sync.max.u32
%val = call i32 @llvm.nvvm.redux.sync.umax(i32 %src, i32 %mask)
ret i32 %val
}
declare i32 @llvm.nvvm.redux.sync.add(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_add_s32
define i32 @redux_sync_add_s32(i32 %src, i32 %mask) {
; CHECK: redux.sync.add.s32
%val = call i32 @llvm.nvvm.redux.sync.add(i32 %src, i32 %mask)
ret i32 %val
}
declare i32 @llvm.nvvm.redux.sync.min(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_min_s32
define i32 @redux_sync_min_s32(i32 %src, i32 %mask) {
; CHECK: redux.sync.min.s32
%val = call i32 @llvm.nvvm.redux.sync.min(i32 %src, i32 %mask)
ret i32 %val
}
declare i32 @llvm.nvvm.redux.sync.max(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_max_s32
define i32 @redux_sync_max_s32(i32 %src, i32 %mask) {
; CHECK: redux.sync.max.s32
%val = call i32 @llvm.nvvm.redux.sync.max(i32 %src, i32 %mask)
ret i32 %val
}
declare i32 @llvm.nvvm.redux.sync.and(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_and_b32
define i32 @redux_sync_and_b32(i32 %src, i32 %mask) {
; CHECK: redux.sync.and.b32
%val = call i32 @llvm.nvvm.redux.sync.and(i32 %src, i32 %mask)
ret i32 %val
}
declare i32 @llvm.nvvm.redux.sync.xor(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_xor_b32
define i32 @redux_sync_xor_b32(i32 %src, i32 %mask) {
; CHECK: redux.sync.xor.b32
%val = call i32 @llvm.nvvm.redux.sync.xor(i32 %src, i32 %mask)
ret i32 %val
}
declare i32 @llvm.nvvm.redux.sync.or(i32, i32)
; CHECK-LABEL: .func{{.*}}redux_sync_or_b32
define i32 @redux_sync_or_b32(i32 %src, i32 %mask) {
; CHECK: redux.sync.or.b32
%val = call i32 @llvm.nvvm.redux.sync.or(i32 %src, i32 %mask)
ret i32 %val
}