Skip to content

[AMDGPU] Support SIProgramInfo MCExpr for comments and remarks #94350

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Jun 10, 2024
Merged

[AMDGPU] Support SIProgramInfo MCExpr for comments and remarks #94350

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
629302a
[AMDGPU] Support SIProgramInfo MCExpr for comments and remarks
JanekvO Jun 4, 2024
7c6ab8f
SmallString + StringRef, check for nullptr instead of optional
JanekvO Jun 7, 2024
b804bdf
Feedback
JanekvO Jun 7, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
119 changes: 72 additions & 47 deletions llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -400,6 +400,36 @@ void AMDGPUAsmPrinter::emitCommonFunctionComments(
false);
}

SmallString<128> AMDGPUAsmPrinter::getMCExprStr(const MCExpr *Value) {
SmallString<128> Str;
raw_svector_ostream OSS(Str);
int64_t IVal;
if (Value->evaluateAsAbsolute(IVal)) {
OSS << static_cast<uint64_t>(IVal);
} else {
Value->print(OSS, MAI);
}
return Str;
}

void AMDGPUAsmPrinter::emitCommonFunctionComments(
const MCExpr *NumVGPR, const MCExpr *NumAGPR, const MCExpr *TotalNumVGPR,
const MCExpr *NumSGPR, const MCExpr *ScratchSize, uint64_t CodeSize,
const AMDGPUMachineFunction *MFI) {
OutStreamer->emitRawComment(" codeLenInByte = " + Twine(CodeSize), false);
OutStreamer->emitRawComment(" NumSgprs: " + getMCExprStr(NumSGPR), false);
OutStreamer->emitRawComment(" NumVgprs: " + getMCExprStr(NumVGPR), false);
if (NumAGPR && TotalNumVGPR) {
OutStreamer->emitRawComment(" NumAgprs: " + getMCExprStr(NumAGPR), false);
OutStreamer->emitRawComment(" TotalNumVgprs: " + getMCExprStr(TotalNumVGPR),
false);
}
OutStreamer->emitRawComment(" ScratchSize: " + getMCExprStr(ScratchSize),
false);
OutStreamer->emitRawComment(" MemoryBound: " + Twine(MFI->isMemoryBound()),
false);
}

uint16_t AMDGPUAsmPrinter::getAmdhsaKernelCodeProperties(
const MachineFunction &MF) const {
const SIMachineFunctionInfo &MFI = *MF.getInfo<SIMachineFunctionInfo>();
Expand Down Expand Up @@ -554,13 +584,10 @@ bool AMDGPUAsmPrinter::runOnMachineFunction(MachineFunction &MF) {

OutStreamer->emitRawComment(" Kernel info:", false);
emitCommonFunctionComments(
getMCExprValue(CurrentProgramInfo.NumArchVGPR, Ctx),
STM.hasMAIInsts() ? getMCExprValue(CurrentProgramInfo.NumAccVGPR, Ctx)
: std::optional<uint32_t>(),
getMCExprValue(CurrentProgramInfo.NumVGPR, Ctx),
getMCExprValue(CurrentProgramInfo.NumSGPR, Ctx),
getMCExprValue(CurrentProgramInfo.ScratchSize, Ctx),
getFunctionCodeSize(MF), MFI);
CurrentProgramInfo.NumArchVGPR,
STM.hasMAIInsts() ? CurrentProgramInfo.NumAccVGPR : nullptr,
CurrentProgramInfo.NumVGPR, CurrentProgramInfo.NumSGPR,
CurrentProgramInfo.ScratchSize, getFunctionCodeSize(MF), MFI);

OutStreamer->emitRawComment(
" FloatMode: " + Twine(CurrentProgramInfo.FloatMode), false);
Expand All @@ -571,43 +598,38 @@ bool AMDGPUAsmPrinter::runOnMachineFunction(MachineFunction &MF) {
" bytes/workgroup (compile time only)", false);

OutStreamer->emitRawComment(
" SGPRBlocks: " +
Twine(getMCExprValue(CurrentProgramInfo.SGPRBlocks, Ctx)),
false);
" SGPRBlocks: " + getMCExprStr(CurrentProgramInfo.SGPRBlocks), false);

OutStreamer->emitRawComment(
" VGPRBlocks: " +
Twine(getMCExprValue(CurrentProgramInfo.VGPRBlocks, Ctx)),
false);
" VGPRBlocks: " + getMCExprStr(CurrentProgramInfo.VGPRBlocks), false);

OutStreamer->emitRawComment(
" NumSGPRsForWavesPerEU: " +
Twine(
getMCExprValue(CurrentProgramInfo.NumSGPRsForWavesPerEU, Ctx)),
getMCExprStr(CurrentProgramInfo.NumSGPRsForWavesPerEU),
false);
OutStreamer->emitRawComment(
" NumVGPRsForWavesPerEU: " +
Twine(
getMCExprValue(CurrentProgramInfo.NumVGPRsForWavesPerEU, Ctx)),
getMCExprStr(CurrentProgramInfo.NumVGPRsForWavesPerEU),
false);

if (STM.hasGFX90AInsts())
if (STM.hasGFX90AInsts()) {
const MCExpr *AdjustedAccum = MCBinaryExpr::createAdd(
CurrentProgramInfo.AccumOffset, MCConstantExpr::create(1, Ctx), Ctx);
AdjustedAccum = MCBinaryExpr::createMul(
AdjustedAccum, MCConstantExpr::create(4, Ctx), Ctx);
OutStreamer->emitRawComment(
" AccumOffset: " +
Twine((getMCExprValue(CurrentProgramInfo.AccumOffset, Ctx) + 1) *
4),
false);
" AccumOffset: " + getMCExprStr(AdjustedAccum), false);
}

OutStreamer->emitRawComment(
" Occupancy: " +
Twine(getMCExprValue(CurrentProgramInfo.Occupancy, Ctx)),
false);
" Occupancy: " + getMCExprStr(CurrentProgramInfo.Occupancy), false);

OutStreamer->emitRawComment(
" WaveLimiterHint : " + Twine(MFI->needsWaveLimiter()), false);

OutStreamer->emitRawComment(
" COMPUTE_PGM_RSRC2:SCRATCH_EN: " +
Twine(getMCExprValue(CurrentProgramInfo.ScratchEnable, Ctx)),
getMCExprStr(CurrentProgramInfo.ScratchEnable),
false);
OutStreamer->emitRawComment(" COMPUTE_PGM_RSRC2:USER_SGPR: " +
Twine(CurrentProgramInfo.UserSGPR),
Expand All @@ -628,20 +650,25 @@ bool AMDGPUAsmPrinter::runOnMachineFunction(MachineFunction &MF) {
Twine(CurrentProgramInfo.TIdIGCompCount),
false);

int64_t PGMRSrc3;
assert(STM.hasGFX90AInsts() ||
getMCExprValue(CurrentProgramInfo.ComputePGMRSrc3GFX90A, Ctx) == 0);
(CurrentProgramInfo.ComputePGMRSrc3GFX90A->evaluateAsAbsolute(
PGMRSrc3) &&
static_cast<uint64_t>(PGMRSrc3) == 0));
if (STM.hasGFX90AInsts()) {
OutStreamer->emitRawComment(
" COMPUTE_PGM_RSRC3_GFX90A:ACCUM_OFFSET: " +
Twine((AMDHSA_BITS_GET(
getMCExprValue(CurrentProgramInfo.ComputePGMRSrc3GFX90A, Ctx),
amdhsa::COMPUTE_PGM_RSRC3_GFX90A_ACCUM_OFFSET))),
getMCExprStr(MCKernelDescriptor::bits_get(
CurrentProgramInfo.ComputePGMRSrc3GFX90A,
amdhsa::COMPUTE_PGM_RSRC3_GFX90A_ACCUM_OFFSET_SHIFT,
amdhsa::COMPUTE_PGM_RSRC3_GFX90A_ACCUM_OFFSET, Ctx)),
false);
OutStreamer->emitRawComment(
" COMPUTE_PGM_RSRC3_GFX90A:TG_SPLIT: " +
Twine((AMDHSA_BITS_GET(
getMCExprValue(CurrentProgramInfo.ComputePGMRSrc3GFX90A, Ctx),
amdhsa::COMPUTE_PGM_RSRC3_GFX90A_TG_SPLIT))),
getMCExprStr(MCKernelDescriptor::bits_get(
CurrentProgramInfo.ComputePGMRSrc3GFX90A,
amdhsa::COMPUTE_PGM_RSRC3_GFX90A_TG_SPLIT_SHIFT,
amdhsa::COMPUTE_PGM_RSRC3_GFX90A_TG_SPLIT, Ctx)),
false);
}
}
Expand Down Expand Up @@ -1463,28 +1490,26 @@ void AMDGPUAsmPrinter::emitResourceUsageRemarks(
// remarks to simulate newlines. If and when clang does accept newlines, this
// formatting should be aggregated into one remark with newlines to avoid
// printing multiple diagnostic location and diag opts.
MCContext &MCCtx = MF.getContext();
EmitResourceUsageRemark("FunctionName", "Function Name",
MF.getFunction().getName());
EmitResourceUsageRemark("NumSGPR", "SGPRs",
getMCExprValue(CurrentProgramInfo.NumSGPR, MCCtx));
EmitResourceUsageRemark(
"NumVGPR", "VGPRs",
getMCExprValue(CurrentProgramInfo.NumArchVGPR, MCCtx));
getMCExprStr(CurrentProgramInfo.NumSGPR));
EmitResourceUsageRemark("NumVGPR", "VGPRs",
getMCExprStr(CurrentProgramInfo.NumArchVGPR));
if (hasMAIInsts) {
EmitResourceUsageRemark(
"NumAGPR", "AGPRs",
getMCExprValue(CurrentProgramInfo.NumAccVGPR, MCCtx));
EmitResourceUsageRemark("NumAGPR", "AGPRs",
getMCExprStr(CurrentProgramInfo.NumAccVGPR));
}
EmitResourceUsageRemark(
"ScratchSize", "ScratchSize [bytes/lane]",
getMCExprValue(CurrentProgramInfo.ScratchSize, MCCtx));
EmitResourceUsageRemark("ScratchSize", "ScratchSize [bytes/lane]",
getMCExprStr(CurrentProgramInfo.ScratchSize));
int64_t DynStack;
bool DynStackEvaluatable =
CurrentProgramInfo.DynamicCallStack->evaluateAsAbsolute(DynStack);
StringRef DynamicStackStr =
getMCExprValue(CurrentProgramInfo.DynamicCallStack, MCCtx) ? "True"
: "False";
DynStackEvaluatable && DynStack ? "True" : "False";
EmitResourceUsageRemark("DynamicStack", "Dynamic Stack", DynamicStackStr);
EmitResourceUsageRemark("Occupancy", "Occupancy [waves/SIMD]",
getMCExprValue(CurrentProgramInfo.Occupancy, MCCtx));
getMCExprStr(CurrentProgramInfo.Occupancy));
EmitResourceUsageRemark("SGPRSpill", "SGPRs Spill",
CurrentProgramInfo.SGPRSpill);
EmitResourceUsageRemark("VGPRSpill", "VGPRs Spill",
Expand Down
6 changes: 6 additions & 0 deletions llvm/lib/Target/AMDGPU/AMDGPUAsmPrinter.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,6 +65,11 @@ class AMDGPUAsmPrinter final : public AsmPrinter {
uint32_t TotalNumVGPR, uint32_t NumSGPR,
uint64_t ScratchSize, uint64_t CodeSize,
const AMDGPUMachineFunction *MFI);
void emitCommonFunctionComments(const MCExpr *NumVGPR, const MCExpr *NumAGPR,
const MCExpr *TotalNumVGPR,
const MCExpr *NumSGPR,
const MCExpr *ScratchSize, uint64_t CodeSize,
const AMDGPUMachineFunction *MFI);
void emitResourceUsageRemarks(const MachineFunction &MF,
const SIProgramInfo &CurrentProgramInfo,
bool isModuleEntryFunction, bool hasMAIInsts);
Expand All @@ -79,6 +84,7 @@ class AMDGPUAsmPrinter final : public AsmPrinter {
void initTargetStreamer(Module &M);

static uint64_t getMCExprValue(const MCExpr *Value, MCContext &Ctx);
SmallString<128> getMCExprStr(const MCExpr *Value);

public:
explicit AMDGPUAsmPrinter(TargetMachine &TM,
Expand Down
Loading