[AMDGPU] Add s_wait_xcnt gfx1250 instruction

[rampitec]

No description provided.

[rampitec]

• [AMDGPU] Add s_wait_xcnt gfx1250 instruction #145086 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[llvmbot]

@llvm/pr-subscribers-mc

@llvm/pr-subscribers-backend-amdgpu

Author: Stanislav Mekhanoshin (rampitec)

Changes

---

Full diff: https://github.com/llvm/llvm-project/pull/145086.diff

5 Files Affected:

• (modified) llvm/lib/Target/AMDGPU/AMDGPU.td (+10)
• (modified) llvm/lib/Target/AMDGPU/GCNSubtarget.h (+5)
• (modified) llvm/lib/Target/AMDGPU/SOPInstructions.td (+10)
• (added) llvm/test/MC/AMDGPU/gfx1250_asm_sopp.s (+14)
• (added) llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_sopp.txt (+10)

diff --git a/llvm/lib/Target/AMDGPU/AMDGPU.td b/llvm/lib/Target/AMDGPU/AMDGPU.td
index 5593284591415..4b17e1c808b50 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPU.td
+++ b/llvm/lib/Target/AMDGPU/AMDGPU.td
@@ -1112,6 +1112,12 @@ def FeaturePointSampleAccel : SubtargetFeature<"point-sample-accel",
   "Has point sample acceleration feature"
 >;
 
+def FeatureWaitXcnt : SubtargetFeature<"wait-xcnt",
+  "HasWaitXcnt",
+  "true",
+  "Has s_wait_xcnt instruction"
+>;
+
 //===------------------------------------------------------------===//
 // Subtarget Features (options and debugging)
 //===------------------------------------------------------------===//
@@ -1927,6 +1933,7 @@ def FeatureISAVersion12_50 : FeatureSet<
    FeatureAshrPkInsts,
    FeatureSupportsSRAMECC,
    FeatureMaxHardClauseLength63,
+   FeatureWaitXcnt,
    FeatureAtomicFMinFMaxF64GlobalInsts,
    FeatureAtomicFMinFMaxF64FlatInsts,
    FeatureFlatBufferGlobalAtomicFaddF64Inst,
@@ -2591,6 +2598,9 @@ def HasPrngInst : Predicate<"Subtarget->hasPrngInst()">,
 def HasBVHDualAndBVH8Insts : Predicate<"Subtarget->hasBVHDualAndBVH8Insts()">,
   AssemblerPredicate<(all_of FeatureBVHDualAndBVH8Insts)>;
 
+def HasWaitXcnt : Predicate<"Subtarget->hasWaitXcnt()">,
+  AssemblerPredicate<(all_of FeatureWaitXcnt)>;
+
 def HasFP8ConversionScaleInsts : Predicate<"Subtarget->hasFP8ConversionScaleInsts()">,
   AssemblerPredicate<(all_of FeatureFP8ConversionScaleInsts)>;
 
diff --git a/llvm/lib/Target/AMDGPU/GCNSubtarget.h b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
index dd57cc96e41c5..4ec60dc2752e4 100644
--- a/llvm/lib/Target/AMDGPU/GCNSubtarget.h
+++ b/llvm/lib/Target/AMDGPU/GCNSubtarget.h
@@ -202,6 +202,7 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
 
   bool HasNoSdstCMPX = false;
   bool HasVscnt = false;
+  bool HasWaitXcnt = false;
   bool HasGetWaveIdInst = false;
   bool HasSMemTimeInst = false;
   bool HasShaderCyclesRegister = false;
@@ -1368,6 +1369,10 @@ class GCNSubtarget final : public AMDGPUGenSubtargetInfo,
     return HasMinimum3Maximum3PKF16;
   }
 
+  /// \returns true if the target has s_wait_xcnt insertion. Supported for
+  /// GFX1250.
+  bool hasWaitXCnt() const { return HasWaitXcnt; }
+
   bool hasPointSampleAccel() const { return HasPointSampleAccel; }
 
   /// \returns The maximum number of instructions that can be enclosed in an
diff --git a/llvm/lib/Target/AMDGPU/SOPInstructions.td b/llvm/lib/Target/AMDGPU/SOPInstructions.td
index 90e65a6950c0a..3f2e764f29268 100644
--- a/llvm/lib/Target/AMDGPU/SOPInstructions.td
+++ b/llvm/lib/Target/AMDGPU/SOPInstructions.td
@@ -1751,6 +1751,11 @@ let OtherPredicates = [HasExportInsts] in
                 [(int_amdgcn_s_wait_kmcnt timm:$simm16)]>;
 } // End SubtargetPredicate = isGFX12Plus, hasSideEffects = 1
 
+let SubtargetPredicate = HasWaitXcnt, hasSideEffects = 1 in {
+  def S_WAIT_XCNT :
+    SOPP_Pseudo<"s_wait_xcnt", (ins s16imm:$simm16), "$simm16">;
+} // End SubtargetPredicate = hasWaitXcnt, hasSideEffects = 1
+
 //===----------------------------------------------------------------------===//
 // SOP1 Patterns
 //===----------------------------------------------------------------------===//
@@ -2560,6 +2565,11 @@ defm S_WAIT_KMCNT           : SOPP_Real_32_gfx12<0x047>;
 defm S_WAIT_LOADCNT_DSCNT   : SOPP_Real_32_gfx12<0x048>;
 defm S_WAIT_STORECNT_DSCNT  : SOPP_Real_32_gfx12<0x049>;
 
+//===----------------------------------------------------------------------===//
+// SOPP - GFX1250 only.
+//===----------------------------------------------------------------------===//
+defm S_WAIT_XCNT      : SOPP_Real_32_gfx12<0x045>;
+
 //===----------------------------------------------------------------------===//
 // SOPP - GFX11, GFX12.
 //===----------------------------------------------------------------------===//
diff --git a/llvm/test/MC/AMDGPU/gfx1250_asm_sopp.s b/llvm/test/MC/AMDGPU/gfx1250_asm_sopp.s
new file mode 100644
index 0000000000000..1aca88771c1f9
--- /dev/null
+++ b/llvm/test/MC/AMDGPU/gfx1250_asm_sopp.s
@@ -0,0 +1,14 @@
+// RUN: llvm-mc -triple=amdgcn -show-encoding -mcpu=gfx1250 %s | FileCheck --check-prefix=GFX1250 %s
+// RUN: not llvm-mc -triple=amdgcn -mcpu=gfx1200 -show-encoding %s 2>&1 | FileCheck --check-prefixes=GFX12-ERR --implicit-check-not=error: -strict-whitespace %s
+
+s_wait_xcnt 0x0
+// GFX1250: [0x00,0x00,0xc5,0xbf]
+// GFX12-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+s_wait_xcnt 0x7
+// GFX1250: [0x07,0x00,0xc5,0xbf]
+// GFX12-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
+
+s_wait_xcnt 0xf
+// GFX1250: [0x0f,0x00,0xc5,0xbf]
+// GFX12-ERR: :[[@LINE-2]]:{{[0-9]+}}: error: instruction not supported on this GPU
diff --git a/llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_sopp.txt b/llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_sopp.txt
new file mode 100644
index 0000000000000..e785fe9cc6d58
--- /dev/null
+++ b/llvm/test/MC/Disassembler/AMDGPU/gfx1250_dasm_sopp.txt
@@ -0,0 +1,10 @@
+# RUN: llvm-mc -triple=amdgcn -mcpu=gfx1250 -disassemble -show-encoding < %s | FileCheck -check-prefixes=GFX1250 %s
+
+# GFX1250: s_wait_xcnt 0x0 ; encoding: [0x00,0x00,0xc5,0xbf]
+0x00,0x00,0xc5,0xbf
+
+# GFX1250: s_wait_xcnt 0x7 ; encoding: [0x07,0x00,0xc5,0xbf]
+0x07,0x00,0xc5,0xbf
+
+# GFX1250: s_wait_xcnt 0xf ; encoding: [0x0f,0x00,0xc5,0xbf]
+0x0f,0x00,0xc5,0xbf