[PATCH] D147159: [AMDGPU][GFX940] Simplify definitions of GLOBAL_ATOMIC

[Mariusz Sikora via Phabricator via llvm-commits]

mariusz-sikora-at-amd created this revision.
Herald added subscribers: kosarev, foad, kerbowa, hiraditya, tpr, dstuttard, yaxunl, jvesely, kzhuravl, arsenm.
Herald added a project: All.
mariusz-sikora-at-amd requested review of this revision.
Herald added subscribers: llvm-commits, wdng.
Herald added a project: LLVM.

This change aims to remove unnecessary SubtargetPredicate override
for Flat_Global_Atomics for gfx940.

In current implementation we override SubtargetPredicate twice.
First time when creating definition for Pseudo Instruction in which
SubtargetPredicate is changed to 'isGFX90APlus' and later when
creating Real instruction SubtargetPredicate is changed to 'isGFX940Plus'.
This should not be needed because multiclass 'FLAT_Global_Atomic_Pseudo'
sets up SubtargetPredicate to 'HasFlatGlobalInsts' and all ASICs Generations
should add this feature if they support it.

Additionaly FLAT_Global_Real_Atomics_gfx940 sets up AssemblerPredicate to
'isGFX940Plus' which later is added to final list of all Predicates
for Real instructions in 'list<Predicate> Predicates'


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D147159

Files:
  llvm/lib/Target/AMDGPU/FLATInstructions.td


Index: llvm/lib/Target/AMDGPU/FLATInstructions.td
===================================================================
--- llvm/lib/Target/AMDGPU/FLATInstructions.td
+++ llvm/lib/Target/AMDGPU/FLATInstructions.td
@@ -725,11 +725,12 @@
   defm FLAT_ATOMIC_ADD_F64   : FLAT_Atomic_Pseudo<"flat_atomic_add_f64", VReg_64, f64>;
   defm FLAT_ATOMIC_MIN_F64   : FLAT_Atomic_Pseudo<"flat_atomic_min_f64", VReg_64, f64>;
   defm FLAT_ATOMIC_MAX_F64   : FLAT_Atomic_Pseudo<"flat_atomic_max_f64", VReg_64, f64>;
-  defm GLOBAL_ATOMIC_ADD_F64 : FLAT_Global_Atomic_Pseudo<"global_atomic_add_f64", VReg_64, f64>;
-  defm GLOBAL_ATOMIC_MIN_F64 : FLAT_Global_Atomic_Pseudo<"global_atomic_min_f64", VReg_64, f64>;
-  defm GLOBAL_ATOMIC_MAX_F64 : FLAT_Global_Atomic_Pseudo<"global_atomic_max_f64", VReg_64, f64>;
 } // End SubtargetPredicate = isGFX90APlus
 
+defm GLOBAL_ATOMIC_ADD_F64 : FLAT_Global_Atomic_Pseudo<"global_atomic_add_f64", VReg_64, f64>;
+defm GLOBAL_ATOMIC_MIN_F64 : FLAT_Global_Atomic_Pseudo<"global_atomic_min_f64", VReg_64, f64>;
+defm GLOBAL_ATOMIC_MAX_F64 : FLAT_Global_Atomic_Pseudo<"global_atomic_max_f64", VReg_64, f64>;
+
 let SubtargetPredicate = HasAtomicFlatPkAdd16Insts in {
   defm FLAT_ATOMIC_PK_ADD_F16    : FLAT_Atomic_Pseudo<"flat_atomic_pk_add_f16",  VGPR_32, v2f16>;
   defm FLAT_ATOMIC_PK_ADD_BF16   : FLAT_Atomic_Pseudo<"flat_atomic_pk_add_bf16", VGPR_32, v2f16>;
@@ -1942,22 +1943,24 @@
 }
 
 let SubtargetPredicate = isGFX940Plus in {
-  // These instructions are encoded differently on gfx90* and gfx940.
-  defm GLOBAL_ATOMIC_ADD_F32     : FLAT_Global_Real_Atomics_gfx940 <0x04d>;
-  defm GLOBAL_ATOMIC_PK_ADD_F16  : FLAT_Global_Real_Atomics_gfx940 <0x04e>;
 
   defm FLAT_ATOMIC_ADD_F64       : FLAT_Real_Atomics_gfx940<0x4f, FLAT_ATOMIC_ADD_F64>;
   defm FLAT_ATOMIC_MIN_F64       : FLAT_Real_Atomics_gfx940<0x50, FLAT_ATOMIC_MIN_F64>;
   defm FLAT_ATOMIC_MAX_F64       : FLAT_Real_Atomics_gfx940<0x51, FLAT_ATOMIC_MAX_F64>;
-  defm GLOBAL_ATOMIC_ADD_F64     : FLAT_Global_Real_Atomics_gfx940<0x4f>;
-  defm GLOBAL_ATOMIC_MIN_F64     : FLAT_Global_Real_Atomics_gfx940<0x50>;
-  defm GLOBAL_ATOMIC_MAX_F64     : FLAT_Global_Real_Atomics_gfx940<0x51>;
   defm FLAT_ATOMIC_ADD_F32       : FLAT_Real_Atomics_vi<0x4d, FLAT_ATOMIC_ADD_F32>;
   defm FLAT_ATOMIC_PK_ADD_F16    : FLAT_Real_Atomics_vi<0x4e, FLAT_ATOMIC_PK_ADD_F16>;
   defm FLAT_ATOMIC_PK_ADD_BF16   : FLAT_Real_Atomics_vi<0x52, FLAT_ATOMIC_PK_ADD_BF16>;
   defm GLOBAL_ATOMIC_PK_ADD_BF16 : FLAT_Global_Real_Atomics_vi<0x52>;
 } // End SubtargetPredicate = isGFX940Plus
 
+// These instructions are encoded differently on gfx90* and gfx940.
+defm GLOBAL_ATOMIC_ADD_F32     : FLAT_Global_Real_Atomics_gfx940<0x04d>;
+defm GLOBAL_ATOMIC_PK_ADD_F16  : FLAT_Global_Real_Atomics_gfx940<0x04e>;
+
+defm GLOBAL_ATOMIC_ADD_F64     : FLAT_Global_Real_Atomics_gfx940<0x4f>;
+defm GLOBAL_ATOMIC_MIN_F64     : FLAT_Global_Real_Atomics_gfx940<0x50>;
+defm GLOBAL_ATOMIC_MAX_F64     : FLAT_Global_Real_Atomics_gfx940<0x51>;
+
 //===----------------------------------------------------------------------===//
 // GFX10.
 //===----------------------------------------------------------------------===//


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D147159.509361.patch
Type: text/x-patch
Size: 3172 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230329/f977563a/attachment.bin>