[llvm-branch-commits] [clang] [llvm] [HLSL] Define RasterizerOrderedBuffer resource (PR #74897)
https://github.com/bogner updated https://github.com/llvm/llvm-project/pull/74897 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [llvm] [HLSL] Define RasterizerOrderedBuffer resource (PR #74897)
https://github.com/bogner updated https://github.com/llvm/llvm-project/pull/74897 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang] [llvm] [HLSL] Define RasterizerOrderedBuffer resource (PR #74897)
@@ -0,0 +1,20 @@ +// RUN: %clang_cc1 -triple dxil-pc-shadermodel6.0-compute -x hlsl -emit-llvm -disable-llvm-passes -o - %s | FileCheck %s bogner wrote: That's a fair point. I went ahead and changed the triple to specify a pixel shader to avoid any confusion / future validation issues https://github.com/llvm/llvm-project/pull/74897 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [mlir] [compiler-rt] [llvm] [clang-tools-extra] [libcxx] [clang] [HLSL] Define RasterizerOrderedBuffer resource (PR #74897)
https://github.com/bogner updated https://github.com/llvm/llvm-project/pull/74897 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [clang-tools-extra] [mlir] [libcxx] [compiler-rt] [clang] [llvm] [HLSL] Define RasterizerOrderedBuffer resource (PR #74897)
https://github.com/bogner updated https://github.com/llvm/llvm-project/pull/74897 ___ llvm-branch-commits mailing list llvm-branch-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
[llvm-branch-commits] [flang] [flang] Move OpenMP-related code from `FirConverter` to `OpenMPMixin` (PR #74866)
https://github.com/kparzysz updated
https://github.com/llvm/llvm-project/pull/74866
>From 27fab0c65445893fb27baead5573bad2dd690dfc Mon Sep 17 00:00:00 2001
From: Krzysztof Parzyszek
Date: Fri, 8 Dec 2023 09:13:11 -0600
Subject: [PATCH] [flang] Move OpenMP-related code from `FirConverter` to
`OpenMPMixin`
This improves the separation of the generic Fortran lowering and the
lowering of OpenMP constructs.
The mixin is intended to be derived from via CRTP:
```
class FirConverter : public OpenMPMixin ...
```
The primary goal of the mixin is to implement `genFIR` functions
that the derived converter can then call via
```
std::visit([this](auto &&s) { genFIR(s); });
```
The mixin is also expecting a handful of functions to be present
in the derived class, most importantly `genFIR(Evaluation&)`, plus
getter classes for the op builder, symbol table, etc.
The pre-existing PFT-lowering functionality is preserved.
---
flang/lib/Lower/Bridge.cpp | 84 +-
flang/lib/Lower/ConverterMixin.h | 28
flang/lib/Lower/FirConverter.h | 38 +-
flang/lib/Lower/OpenMP.cpp | 118 ++-
flang/lib/Lower/OpenMPMixin.h| 66 +
5 files changed, 235 insertions(+), 99 deletions(-)
create mode 100644 flang/lib/Lower/ConverterMixin.h
create mode 100644 flang/lib/Lower/OpenMPMixin.h
diff --git a/flang/lib/Lower/Bridge.cpp b/flang/lib/Lower/Bridge.cpp
index 5aaba233744b2d..0a476a38d8d2de 100644
--- a/flang/lib/Lower/Bridge.cpp
+++ b/flang/lib/Lower/Bridge.cpp
@@ -170,7 +170,7 @@ void FirConverter::run(Fortran::lower::pft::Program &pft) {
});
finalizeOpenACCLowering();
- finalizeOpenMPLowering(globalOmpRequiresSymbol);
+ OpenMPBase::finalize(globalOmpRequiresSymbol);
}
/// Generate FIR for Evaluation \p eval.
@@ -977,70 +977,6 @@ void FirConverter::genFIR(const
Fortran::parser::OpenACCRoutineConstruct &acc) {
// Handled by genFIR(const Fortran::parser::OpenACCDeclarativeConstruct &)
}
-void FirConverter::genFIR(const Fortran::parser::OpenMPConstruct &omp) {
- mlir::OpBuilder::InsertPoint insertPt = builder->saveInsertionPoint();
- localSymbols.pushScope();
- genOpenMPConstruct(*this, bridge.getSemanticsContext(), getEval(), omp);
-
- const Fortran::parser::OpenMPLoopConstruct *ompLoop =
- std::get_if(&omp.u);
- const Fortran::parser::OpenMPBlockConstruct *ompBlock =
- std::get_if(&omp.u);
-
- // If loop is part of an OpenMP Construct then the OpenMP dialect
- // workshare loop operation has already been created. Only the
- // body needs to be created here and the do_loop can be skipped.
- // Skip the number of collapsed loops, which is 1 when there is a
- // no collapse requested.
-
- Fortran::lower::pft::Evaluation *curEval = &getEval();
- const Fortran::parser::OmpClauseList *loopOpClauseList = nullptr;
- if (ompLoop) {
-loopOpClauseList = &std::get(
-std::get(ompLoop->t).t);
-int64_t collapseValue =
Fortran::lower::getCollapseValue(*loopOpClauseList);
-
-curEval = &curEval->getFirstNestedEvaluation();
-for (int64_t i = 1; i < collapseValue; i++) {
- curEval = &*std::next(curEval->getNestedEvaluations().begin());
-}
- }
-
- for (Fortran::lower::pft::Evaluation &e : curEval->getNestedEvaluations())
-genFIR(e);
-
- if (ompLoop) {
-genOpenMPReduction(*this, *loopOpClauseList);
- } else if (ompBlock) {
-const auto &blockStart =
-std::get(ompBlock->t);
-const auto &blockClauses =
-std::get(blockStart.t);
-genOpenMPReduction(*this, blockClauses);
- }
-
- localSymbols.popScope();
- builder->restoreInsertionPoint(insertPt);
-
- // Register if a target region was found
- ompDeviceCodeFound =
- ompDeviceCodeFound || Fortran::lower::isOpenMPTargetConstruct(omp);
-}
-
-void FirConverter::genFIR(
-const Fortran::parser::OpenMPDeclarativeConstruct &ompDecl) {
- mlir::OpBuilder::InsertPoint insertPt = builder->saveInsertionPoint();
- // Register if a declare target construct intended for a target device was
- // found
- ompDeviceCodeFound =
- ompDeviceCodeFound ||
- Fortran::lower::isOpenMPDeviceDeclareTarget(*this, getEval(), ompDecl);
- genOpenMPDeclarativeConstruct(*this, getEval(), ompDecl);
- for (Fortran::lower::pft::Evaluation &e : getEval().getNestedEvaluations())
-genFIR(e);
- builder->restoreInsertionPoint(insertPt);
-}
-
void FirConverter::genFIR(const Fortran::parser::OpenStmt &stmt) {
mlir::Value iostat = genOpenStatement(*this, stmt);
genIoConditionBranches(getEval(), stmt.v, iostat);
@@ -3752,13 +3688,7 @@ void FirConverter::instantiateVar(const
Fortran::lower::pft::Variable &var,
Fortran::lower::AggregateStoreMap &storeMap)
{
Fortran::lower::instantiateVariable(*this, var, localSymbols, storeMap);
if (var.hasSymbol()) {
-if (var.getSymbol().test(
-Fortran::semantics::Symbol::Flag::OmpThreadprivate))
- For
[llvm-branch-commits] [llvm] [BPF] expand cttz, ctlz for i32, i64 (PR #73668)
https://github.com/inclyc updated
https://github.com/llvm/llvm-project/pull/73668
>From c9afc897d51cf7ec93d7d5f7a43ed53346bb282d Mon Sep 17 00:00:00 2001
From: Yingchi Long
Date: Wed, 29 Nov 2023 00:23:49 +0800
Subject: [PATCH] [BPF] expand cttz, ctlz for i32, i64
Fixes: https://github.com/llvm/llvm-project/issues/62252
---
llvm/lib/Target/BPF/BPFISelLowering.cpp | 9 +-
llvm/test/CodeGen/BPF/cttz-ctlz.ll | 304
2 files changed, 308 insertions(+), 5 deletions(-)
create mode 100644 llvm/test/CodeGen/BPF/cttz-ctlz.ll
diff --git a/llvm/lib/Target/BPF/BPFISelLowering.cpp
b/llvm/lib/Target/BPF/BPFISelLowering.cpp
index 45645df52f26a4..996fb590443b5f 100644
--- a/llvm/lib/Target/BPF/BPFISelLowering.cpp
+++ b/llvm/lib/Target/BPF/BPFISelLowering.cpp
@@ -111,6 +111,10 @@ BPFTargetLowering::BPFTargetLowering(const TargetMachine
&TM,
setOperationAction(ISD::SRL_PARTS, VT, Expand);
setOperationAction(ISD::SRA_PARTS, VT, Expand);
setOperationAction(ISD::CTPOP, VT, Expand);
+setOperationAction(ISD::CTTZ, VT, Expand);
+setOperationAction(ISD::CTLZ, VT, Expand);
+setOperationAction(ISD::CTTZ_ZERO_UNDEF, VT, Expand);
+setOperationAction(ISD::CTLZ_ZERO_UNDEF, VT, Expand);
setOperationAction(ISD::SETCC, VT, Expand);
setOperationAction(ISD::SELECT, VT, Expand);
@@ -123,11 +127,6 @@ BPFTargetLowering::BPFTargetLowering(const TargetMachine
&TM,
STI.getHasJmp32() ? Custom : Promote);
}
- setOperationAction(ISD::CTTZ, MVT::i64, Custom);
- setOperationAction(ISD::CTLZ, MVT::i64, Custom);
- setOperationAction(ISD::CTTZ_ZERO_UNDEF, MVT::i64, Custom);
- setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i64, Custom);
-
setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::i1, Expand);
if (!STI.hasMovsx()) {
setOperationAction(ISD::SIGN_EXTEND_INREG, MVT::i8, Expand);
diff --git a/llvm/test/CodeGen/BPF/cttz-ctlz.ll
b/llvm/test/CodeGen/BPF/cttz-ctlz.ll
new file mode 100644
index 00..f42b2e2d10871b
--- /dev/null
+++ b/llvm/test/CodeGen/BPF/cttz-ctlz.ll
@@ -0,0 +1,304 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
UTC_ARGS: --version 4
+; RUN: llc < %s -march=bpf | FileCheck %s
+
+; test that we can expand CTTZ & CTLZ
+
+declare i32 @llvm.cttz.i32(i32, i1)
+
+define i32 @cttz_i32_zdef(i32 %a) {
+; CHECK-LABEL: cttz_i32_zdef:
+; CHECK: # %bb.0:
+; CHECK-NEXT:r2 = r1
+; CHECK-NEXT:r2 = -r2
+; CHECK-NEXT:r1 &= r2
+; CHECK-NEXT:r1 *= 125613361
+; CHECK-NEXT:r2 = 4160749568 ll
+; CHECK-NEXT:r1 &= r2
+; CHECK-NEXT:r1 >>= 27
+; CHECK-NEXT:r2 = {{\.?LCPI[0-9]+_[0-9]+}} ll
+; CHECK-NEXT:r2 += r1
+; CHECK-NEXT:r0 = *(u8 *)(r2 + 0)
+; CHECK-NEXT:exit
+%ret = call i32 @llvm.cttz.i32(i32 %a, i1 1)
+ret i32 %ret
+}
+
+
+define i32 @cttz_i32(i32 %a) {
+; CHECK-LABEL: cttz_i32:
+; CHECK: # %bb.0:
+; CHECK-NEXT:r0 = 32
+; CHECK-NEXT:r2 = r1
+; CHECK-NEXT:r2 <<= 32
+; CHECK-NEXT:r2 >>= 32
+; CHECK-NEXT:if r2 == 0 goto LBB1_2
+; CHECK-NEXT: # %bb.1: # %cond.false
+; CHECK-NEXT:r2 = r1
+; CHECK-NEXT:r2 = -r2
+; CHECK-NEXT:r1 &= r2
+; CHECK-NEXT:r1 *= 125613361
+; CHECK-NEXT:r2 = 4160749568 ll
+; CHECK-NEXT:r1 &= r2
+; CHECK-NEXT:r1 >>= 27
+; CHECK-NEXT:r2 = {{\.?LCPI[0-9]+_[0-9]+}} ll
+; CHECK-NEXT:r2 += r1
+; CHECK-NEXT:r0 = *(u8 *)(r2 + 0)
+; CHECK-NEXT: LBB1_2: # %cond.end
+; CHECK-NEXT:exit
+%ret = call i32 @llvm.cttz.i32(i32 %a, i1 0)
+ret i32 %ret
+}
+
+declare i64 @llvm.cttz.i64(i64, i1)
+
+define i64 @cttz_i64_zdef(i64 %a) {
+; CHECK-LABEL: cttz_i64_zdef:
+; CHECK: # %bb.0:
+; CHECK-NEXT:r2 = r1
+; CHECK-NEXT:r2 = -r2
+; CHECK-NEXT:r1 &= r2
+; CHECK-NEXT:r2 = 151050438420815295 ll
+; CHECK-NEXT:r1 *= r2
+; CHECK-NEXT:r1 >>= 58
+; CHECK-NEXT:r2 = {{\.?LCPI[0-9]+_[0-9]+}} ll
+; CHECK-NEXT:r2 += r1
+; CHECK-NEXT:r0 = *(u8 *)(r2 + 0)
+; CHECK-NEXT:exit
+%ret = call i64 @llvm.cttz.i64(i64 %a, i1 1)
+ret i64 %ret
+}
+
+
+define i64 @cttz_i64(i64 %a) {
+; CHECK-LABEL: cttz_i64:
+; CHECK: # %bb.0:
+; CHECK-NEXT:r0 = 64
+; CHECK-NEXT:if r1 == 0 goto LBB3_2
+; CHECK-NEXT: # %bb.1: # %cond.false
+; CHECK-NEXT:r2 = r1
+; CHECK-NEXT:r2 = -r2
+; CHECK-NEXT:r1 &= r2
+; CHECK-NEXT:r2 = 151050438420815295 ll
+; CHECK-NEXT:r1 *= r2
+; CHECK-NEXT:r1 >>= 58
+; CHECK-NEXT:r2 = {{\.?LCPI[0-9]+_[0-9]+}} ll
+; CHECK-NEXT:r2 += r1
+; CHECK-NEXT:r0 = *(u8 *)(r2 + 0)
+; CHECK-NEXT: LBB3_2: # %cond.end
+; CHECK-NEXT:exit
+%ret = call i64 @llvm.cttz.i64(i64 %a, i1 0)
+ret i64 %ret
+}
+
+
+declare i32 @llvm.ctlz.i32(i32, i1)
+
+define i32 @ctlz_i32_zdef(i32 %a) {
+; CHECK-LABEL: ctlz_i32_zdef:
+; CHECK: # %bb.0:
+; CHECK-NEXT:r2 = 4294967294 ll
+; CHECK-NEXT:r3 = r1
+; CHECK-NEXT:r3 &= r2
+; CHECK-NEXT:r3 >>= 1
+; CHECK-NEXT:r1 |= r3
+;
