[flang][NVPTX] Add initial support to the NVPTX target #71992
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This patch adds initial support to the NVPTX target, enabling Flang to produce OpenMP offload code for NVPTX targets.
llvmbot
added
flang:driver
flang
|
@llvm/pr-subscribers-flang-driver @llvm/pr-subscribers-flang-codegen Author: Fabian Mora (fabianmcg) ChangesThis patch adds initial support to the NVPTX target, enabling Patch is 20.38 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/71992.diff 15 Files Affected:
diff --git a/flang/lib/Frontend/FrontendActions.cpp b/flang/lib/Frontend/FrontendActions.cpp
index 73c00c8679c7ec6..18e312469f3ee63 100644
--- a/flang/lib/Frontend/FrontendActions.cpp
+++ b/flang/lib/Frontend/FrontendActions.cpp
@@ -175,6 +175,45 @@ getExplicitAndImplicitAMDGPUTargetFeatures(CompilerInstance &ci,
return llvm::join(featuresVec, ",");
}
+// Get feature string which represents combined explicit target features
+// for NVPTX and the target features specified by the user/
+// TODO: Have a more robust target conf like `clang/lib/Basic/Targets/NVPTX.cpp`
+static std::string
+getExplicitAndImplicitNVPTXTargetFeatures(CompilerInstance &ci,
+ const TargetOptions &targetOpts,
+ const llvm::Triple triple) {
+ llvm::StringRef cpu = targetOpts.cpu;
+ llvm::StringMap<bool> implicitFeaturesMap;
+ std::string errorMsg;
+ bool ptxVer = false;
+
+ // Add target features specified by the user
+ for (auto &userFeature : targetOpts.featuresAsWritten) {
+ llvm::StringRef userKeyString(llvm::StringRef(userFeature).drop_front(1));
+ implicitFeaturesMap[userKeyString.str()] = (userFeature[0] == '+');
+ // Check if the user provided a PTX version
+ if (userKeyString.startswith("ptx"))
+ ptxVer = true;
+ }
+
+ // Set the default PTX version to `ptx61` if none was provided.
+ // TODO: set the default PTX version based on the chip.
+ if (!ptxVer)
+ implicitFeaturesMap["ptx61"] = true;
+
+ // Set the compute capability.
+ implicitFeaturesMap[cpu.str()] = true;
+
+ llvm::SmallVector<std::string> featuresVec;
+ for (auto &implicitFeatureItem : implicitFeaturesMap) {
+ featuresVec.push_back((llvm::Twine(implicitFeatureItem.second ? "+" : "-") +
+ implicitFeatureItem.first().str())
+ .str());
+ }
+ llvm::sort(featuresVec);
+ return llvm::join(featuresVec, ",");
+}
+
// Produces the string which represents target feature
static std::string getTargetFeatures(CompilerInstance &ci) {
const TargetOptions &targetOpts = ci.getInvocation().getTargetOpts();
@@ -188,6 +227,8 @@ static std::string getTargetFeatures(CompilerInstance &ci) {
// them to the target features specified by the user
if (triple.isAMDGPU()) {
return getExplicitAndImplicitAMDGPUTargetFeatures(ci, targetOpts, triple);
+ } else if (triple.isNVPTX()) {
+ return getExplicitAndImplicitNVPTXTargetFeatures(ci, targetOpts, triple);
}
return llvm::join(targetOpts.featuresAsWritten.begin(),
targetOpts.featuresAsWritten.end(), ",");
diff --git a/flang/lib/Optimizer/CodeGen/Target.cpp b/flang/lib/Optimizer/CodeGen/Target.cpp
index 83e7fa9b440bed2..bb893277cb4d21d 100644
--- a/flang/lib/Optimizer/CodeGen/Target.cpp
+++ b/flang/lib/Optimizer/CodeGen/Target.cpp
@@ -621,6 +621,33 @@ struct TargetAMDGPU : public GenericTarget<TargetAMDGPU> {
};
} // namespace
+//===----------------------------------------------------------------------===//
+// NVPTX linux target specifics.
+//===----------------------------------------------------------------------===//
+
+namespace {
+struct TargetNVPTX : public GenericTarget<TargetNVPTX> {
+ using GenericTarget::GenericTarget;
+
+ // Default size (in bits) of the index type for strings.
+ static constexpr int defaultWidth = 64;
+
+ CodeGenSpecifics::Marshalling
+ complexArgumentType(mlir::Location loc, mlir::Type eleTy) const override {
+ CodeGenSpecifics::Marshalling marshal;
+ TODO(loc, "handle complex argument types");
+ return marshal;
+ }
+
+ CodeGenSpecifics::Marshalling
+ complexReturnType(mlir::Location loc, mlir::Type eleTy) const override {
+ CodeGenSpecifics::Marshalling marshal;
+ TODO(loc, "handle complex return types");
+ return marshal;
+ }
+};
+} // namespace
+
//===----------------------------------------------------------------------===//
// LoongArch64 linux target specifics.
//===----------------------------------------------------------------------===//
@@ -708,6 +735,9 @@ fir::CodeGenSpecifics::get(mlir::MLIRContext *ctx, llvm::Triple &&trp,
case llvm::Triple::ArchType::amdgcn:
return std::make_unique<TargetAMDGPU>(ctx, std::move(trp),
std::move(kindMap));
+ case llvm::Triple::ArchType::nvptx64:
+ return std::make_unique<TargetNVPTX>(ctx, std::move(trp),
+ std::move(kindMap));
case llvm::Triple::ArchType::loongarch64:
return std::make_unique<TargetLoongArch64>(ctx, std::move(trp),
std::move(kindMap));
diff --git a/flang/test/Driver/omp-driver-offload.f90 b/flang/test/Driver/omp-driver-offload.f90
index bfdc3f6f4d4726b..ad50723b0e3a795 100644
--- a/flang/test/Driver/omp-driver-offload.f90
+++ b/flang/test/Driver/omp-driver-offload.f90
@@ -67,6 +67,11 @@
! RUN: -fopenmp-targets=amdgcn-amd-amdhsa \
! RUN: -fopenmp-assume-threads-oversubscription \
! RUN: | FileCheck %s --check-prefixes=CHECK-THREADS-OVS
+! RUN: %flang -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-assume-threads-oversubscription \
+! RUN: | FileCheck %s --check-prefixes=CHECK-THREADS-OVS
! CHECK-THREADS-OVS: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" {{.*}} "-fopenmp-is-target-device" "-fopenmp-assume-threads-oversubscription" {{.*}}.f90"
! RUN: %flang -### %s -o %t 2>&1 \
@@ -74,6 +79,11 @@
! RUN: -fopenmp-targets=amdgcn-amd-amdhsa \
! RUN: -fopenmp-assume-teams-oversubscription \
! RUN: | FileCheck %s --check-prefixes=CHECK-TEAMS-OVS
+! RUN: %flang -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-assume-teams-oversubscription \
+! RUN: | FileCheck %s --check-prefixes=CHECK-TEAMS-OVS
! CHECK-TEAMS-OVS: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" {{.*}} "-fopenmp-is-target-device" "-fopenmp-assume-teams-oversubscription" {{.*}}.f90"
! RUN: %flang -### %s -o %t 2>&1 \
@@ -81,6 +91,11 @@
! RUN: -fopenmp-targets=amdgcn-amd-amdhsa \
! RUN: -fopenmp-assume-no-nested-parallelism \
! RUN: | FileCheck %s --check-prefixes=CHECK-NEST-PAR
+! RUN: %flang -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-assume-no-nested-parallelism \
+! RUN: | FileCheck %s --check-prefixes=CHECK-NEST-PAR
! CHECK-NEST-PAR: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" {{.*}} "-fopenmp-is-target-device" "-fopenmp-assume-no-nested-parallelism" {{.*}}.f90"
! RUN: %flang -### %s -o %t 2>&1 \
@@ -88,6 +103,11 @@
! RUN: -fopenmp-targets=amdgcn-amd-amdhsa \
! RUN: -fopenmp-assume-no-thread-state \
! RUN: | FileCheck %s --check-prefixes=CHECK-THREAD-STATE
+! RUN: %flang -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-assume-no-thread-state \
+! RUN: | FileCheck %s --check-prefixes=CHECK-THREAD-STATE
! CHECK-THREAD-STATE: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" {{.*}} "-fopenmp-is-target-device" "-fopenmp-assume-no-thread-state" {{.*}}.f90"
! RUN: %flang -### %s -o %t 2>&1 \
@@ -95,6 +115,11 @@
! RUN: -fopenmp-targets=amdgcn-amd-amdhsa \
! RUN: -fopenmp-target-debug \
! RUN: | FileCheck %s --check-prefixes=CHECK-TARGET-DEBUG
+! RUN: %flang -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-target-debug \
+! RUN: | FileCheck %s --check-prefixes=CHECK-TARGET-DEBUG
! CHECK-TARGET-DEBUG: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" {{.*}} "-fopenmp-is-target-device" "-fopenmp-target-debug" {{.*}}.f90"
! RUN: %flang -### %s -o %t 2>&1 \
@@ -102,6 +127,11 @@
! RUN: -fopenmp-targets=amdgcn-amd-amdhsa \
! RUN: -fopenmp-target-debug \
! RUN: | FileCheck %s --check-prefixes=CHECK-TARGET-DEBUG
+! RUN: %flang -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-target-debug \
+! RUN: | FileCheck %s --check-prefixes=CHECK-TARGET-DEBUG
! CHECK-TARGET-DEBUG-EQ: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" {{.*}} "-fopenmp-is-target-device" "-fopenmp-target-debug=111" {{.*}}.f90"
! RUN: %flang -S -### %s -o %t 2>&1 \
@@ -111,6 +141,13 @@
! RUN: -fopenmp-assume-teams-oversubscription -fopenmp-assume-no-nested-parallelism \
! RUN: -fopenmp-assume-no-thread-state \
! RUN: | FileCheck %s --check-prefixes=CHECK-RTL-ALL
+! RUN: %flang -S -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-target-debug -fopenmp-assume-threads-oversubscription \
+! RUN: -fopenmp-assume-teams-oversubscription -fopenmp-assume-no-nested-parallelism \
+! RUN: -fopenmp-assume-no-thread-state \
+! RUN: | FileCheck %s --check-prefixes=CHECK-RTL-ALL
! CHECK-RTL-ALL: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" {{.*}} "-fopenmp-is-target-device" "-fopenmp-target-debug" "-fopenmp-assume-teams-oversubscription"
! CHECK-RTL-ALL: "-fopenmp-assume-threads-oversubscription" "-fopenmp-assume-no-thread-state" "-fopenmp-assume-no-nested-parallelism"
! CHECK-RTL-ALL: {{.*}}.f90"
@@ -120,6 +157,11 @@
! RUN: -fopenmp-targets=amdgcn-amd-amdhsa \
! RUN: -fopenmp-version=45 \
! RUN: | FileCheck %s --check-prefixes=CHECK-OPENMP-VERSION
+! RUN: %flang -### %s -o %t 2>&1 \
+! RUN: -fopenmp --offload-arch=sm_70 \
+! RUN: -fopenmp-targets=nvptx64-nvidia-cuda \
+! RUN: -fopenmp-version=45 \
+! RUN: | FileCheck %s --check-prefixes=CHECK-OPENMP-VERSION
! CHECK-OPENMP-VERSION: "{{[^"]*}}flang-new" "-fc1" {{.*}} "-fopenmp" "-fopenmp-version=45" {{.*}}.f90"
! Test diagnostic error when host IR file is non-existent
diff --git a/flang/test/Fir/target-rewrite-boxchar.fir b/flang/test/Fir/target-rewrite-boxchar.fir
index e66fa6041630380..b87cb35b46eb6c4 100644
--- a/flang/test/Fir/target-rewrite-boxchar.fir
+++ b/flang/test/Fir/target-rewrite-boxchar.fir
@@ -3,6 +3,7 @@
// RUN: fir-opt --target-rewrite="target=aarch64-unknown-linux-gnu" %s | FileCheck %s --check-prefix=INT64
// RUN: fir-opt --target-rewrite="target=powerpc64le-unknown-linux-gnu" %s | FileCheck %s --check-prefix=INT64
// RUN: fir-opt --target-rewrite="target=amdgcn-amd-amdhsa" %s | FileCheck %s --check-prefix=INT64
+// RUN: fir-opt --target-rewrite="target=nvptx64-nvidia-cuda" %s | FileCheck %s --check-prefix=INT64
// RUN: fir-opt --target-rewrite="target=loongarch64-unknown-linux-gnu" %s | FileCheck %s --check-prefix=INT64
// Test that we rewrite the signatures and bodies of functions that take boxchar
diff --git a/flang/test/Lower/OpenMP/FIR/omp-is-gpu.f90 b/flang/test/Lower/OpenMP/FIR/omp-is-gpu.f90
index b702fc2c5a7e253..ac8d24974801570 100644
--- a/flang/test/Lower/OpenMP/FIR/omp-is-gpu.f90
+++ b/flang/test/Lower/OpenMP/FIR/omp-is-gpu.f90
@@ -1,9 +1,11 @@
-!REQUIRES: amdgpu-registered-target
+!REQUIRES: amdgpu-registered-target, nvptx-registered-target
!RUN: %flang_fc1 -triple amdgcn-amd-amdhsa -emit-fir -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s
+!RUN: %flang_fc1 -triple nvptx64-nvidia-cuda -emit-fir -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s
!RUN: bbc -fopenmp -fopenmp-is-target-device -fopenmp-is-gpu -emit-fir -o - %s | FileCheck %s
!RUN: not %flang_fc1 -triple amdgcn-amd-amdhsa -emit-fir -fopenmp %s -o - 2>&1 | FileCheck %s --check-prefix=FLANG-ERROR
+!RUN: not %flang_fc1 -triple nvptx64-nvidia-cuda -emit-fir -fopenmp %s -o - 2>&1 | FileCheck %s --check-prefix=FLANG-ERROR
!RUN: not bbc -fopenmp -fopenmp-is-gpu -emit-fir %s -o - 2>&1 | FileCheck %s --check-prefix=BBC-ERROR
!CHECK: module attributes {{{.*}}omp.is_gpu = true
diff --git a/flang/test/Lower/OpenMP/FIR/target_cpu_features.f90 b/flang/test/Lower/OpenMP/FIR/target_cpu_features.f90
index c6159342c023aa4..179b71b3f0cfa5c 100644
--- a/flang/test/Lower/OpenMP/FIR/target_cpu_features.f90
+++ b/flang/test/Lower/OpenMP/FIR/target_cpu_features.f90
@@ -1,5 +1,7 @@
-!REQUIRES: amdgpu-registered-target
+!REQUIRES: amdgpu-registered-target, nvptx-registered-target
!RUN: %flang_fc1 -emit-fir -triple amdgcn-amd-amdhsa -target-cpu gfx908 -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s
+!RUN: %flang_fc1 -emit-hlfir -triple nvptx64-nvidia-cuda -target-cpu sm_80 -fopenmp -fopenmp-is-target-device %s -o - | FileCheck --check-prefix=NVPTX %s
+
!===============================================================================
! Target_Enter Simple
@@ -10,6 +12,7 @@
!CHECK-SAME: +dot2-insts,+dot3-insts,+dot4-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,
!CHECK-SAME: +gfx8-insts,+gfx9-insts,+gws,+image-insts,+mai-insts,+s-memrealtime,+s-memtime-inst,
!CHECK-SAME: +wavefrontsize64">
+!NVPTX: omp.target = #omp.target<target_cpu = "sm_80", target_features = "+ptx61,+sm_80">
!CHECK-LABEL: func.func @_QPomp_target_simple()
subroutine omp_target_simple
! Directive needed to prevent subroutine from being filtered out when
diff --git a/flang/test/Lower/OpenMP/omp-is-gpu.f90 b/flang/test/Lower/OpenMP/omp-is-gpu.f90
index 12d0e4e869fba5a..3e6daeb522d7789 100644
--- a/flang/test/Lower/OpenMP/omp-is-gpu.f90
+++ b/flang/test/Lower/OpenMP/omp-is-gpu.f90
@@ -1,9 +1,11 @@
-!REQUIRES: amdgpu-registered-target
+!REQUIRES: amdgpu-registered-target, nvptx-registered-target
!RUN: %flang_fc1 -triple amdgcn-amd-amdhsa -emit-hlfir -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s
+!RUN: %flang_fc1 -triple nvptx64-nvidia-cuda -emit-hlfir -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s
!RUN: bbc -fopenmp -fopenmp-is-target-device -fopenmp-is-gpu -emit-hlfir -o - %s | FileCheck %s
!RUN: not %flang_fc1 -triple amdgcn-amd-amdhsa -emit-hlfir -fopenmp %s -o - 2>&1 | FileCheck %s --check-prefix=FLANG-ERROR
+!RUN: not %flang_fc1 -triple nvptx64-nvidia-cuda -emit-hlfir -fopenmp %s -o - 2>&1 | FileCheck %s --check-prefix=FLANG-ERROR
!RUN: not bbc -fopenmp -fopenmp-is-gpu -emit-hlfir %s -o - 2>&1 | FileCheck %s --check-prefix=BBC-ERROR
!CHECK: module attributes {{{.*}}omp.is_gpu = true
diff --git a/flang/test/Lower/OpenMP/target_cpu_features.f90 b/flang/test/Lower/OpenMP/target_cpu_features.f90
index 46fb14efad5c03c..ea1e5e38fca88ef 100644
--- a/flang/test/Lower/OpenMP/target_cpu_features.f90
+++ b/flang/test/Lower/OpenMP/target_cpu_features.f90
@@ -1,5 +1,6 @@
-!REQUIRES: amdgpu-registered-target
+!REQUIRES: amdgpu-registered-target, nvptx-registered-target
!RUN: %flang_fc1 -emit-hlfir -triple amdgcn-amd-amdhsa -target-cpu gfx908 -fopenmp -fopenmp-is-target-device %s -o - | FileCheck %s
+!RUN: %flang_fc1 -emit-hlfir -triple nvptx64-nvidia-cuda -target-cpu sm_80 -fopenmp -fopenmp-is-target-device %s -o - | FileCheck --check-prefix=NVPTX %s
!===============================================================================
! Target_Enter Simple
@@ -10,6 +11,7 @@
!CHECK-SAME: +dot2-insts,+dot3-insts,+dot4-insts,+dot5-insts,+dot6-insts,+dot7-insts,+dpp,
!CHECK-SAME: +gfx8-insts,+gfx9-insts,+gws,+image-insts,+mai-insts,+s-memrealtime,+s-memtime-inst,
!CHECK-SAME: +wavefrontsize64">
+!NVPTX: omp.target = #omp.target<target_cpu = "sm_80", target_features = "+ptx61,+sm_80">
!CHECK-LABEL: func.func @_QPomp_target_simple()
subroutine omp_target_simple
! Directive needed to prevent subroutine from being filtered out when
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90
index 58f5379e330ec03..993b91d4eb623e9 100644
--- a/openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-1D-array-section.f90
@@ -1,7 +1,6 @@
! Basic offloading test of arrays with provided lower
! and upper bounds as specified by OpenMP's sectioning
-! REQUIRES: flang, amdgcn-amd-amdhsa
-! UNSUPPORTED: nvptx64-nvidia-cuda
+! REQUIRES: flang, amdgcn-amd-amdhsa, nvptx64-nvidia-cuda
! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
! UNSUPPORTED: aarch64-unknown-linux-gnu
! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90
index e3df7983e6b5c18..669d3674926f696 100644
--- a/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array-section.f90
@@ -1,7 +1,6 @@
! Basic offloading test of a regular array explicitly
! passed within a target region
-! REQUIRES: flang, amdgcn-amd-amdhsa
-! UNSUPPORTED: nvptx64-nvidia-cuda
+! REQUIRES: flang, amdgcn-amd-amdhsa, nvptx64-nvidia-cuda
! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
! UNSUPPORTED: aarch64-unknown-linux-gnu
! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90
index abc2763d4a30cca..c87d6ee24aed3ef 100644
--- a/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-3D-array.f90
@@ -1,7 +1,6 @@
! Basic offloading test of a regular array explicitly
! passed within a target region
-! REQUIRES: flang, amdgcn-amd-amdhsa
-! UNSUPPORTED: nvptx64-nvidia-cuda
+! REQUIRES: flang, amdgcn-amd-amdhsa, nvptx64-nvidia-cuda
! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
! UNSUPPORTED: aarch64-unknown-linux-gnu
! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
diff --git a/openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90 b/openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90
index d3c799ff3334f4d..9b10e4c7650d05a 100644
--- a/openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90
+++ b/openmp/libomptarget/test/offloading/fortran/basic-target-region-array.f90
@@ -1,7 +1,6 @@
! Basic offloading test of a regular array explicitly
! passed within a target region
-! REQUIRES: flang, amdgcn-amd-amdhsa
-! UNSUPPORTED: nvptx64-nvidia-cuda
+! REQUIRES: flang, amdgcn-amd-amdhsa, nvptx64-nvidia-cuda
! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
! UNSUPPORTED: aarch64-unknown-linux-gnu
! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
diff --git a/openmp/libomptarget/test/offloading/fortran/basic_target_region.f90 b/openmp/libomptarget/test/offloading/fortran/basic_target_region.f90
index 295452b0698a660..6423ac765670d48 100644
--- a/openmp/libomptarget/test/offloading/fortran/basic_target_region.f90
+++ b/openmp/libomptarget/test/offloading/fortran/basic_target_region.f90
@@ -1,6 +1,5 @@
! Basic offloading test with a target region
-! REQUIRES: flang, amdgcn-amd-amdhsa
-! UNSUPPORTED: nvptx64-nvidia-cuda
+! REQUIRES: flang, amdgcn-amd-amdhsa, nvptx64-nvidia-cuda
! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
! UNSUPPORTED: aarch64-unknown-linux-gnu
! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
diff --git a/openmp/libomptarget/test/offloading/fortran/declare-target-array-in-target-region.f90 b/openmp/libomptarget/test/offloading/fortran/declare-target-array-in-target-region.f90
index f5e3ae00653a9ab..d2e59d93a0209ec 100644
--- a/openmp/libomptarget/test/offloading/fortran/declare-target-array-in-target-region.f90
+++ b/openmp/libomptarget/test/offloading/fortran/declare-target-array-in-target-region.f90
@@ -1,8 +1,7 @@
! Offloading test with a target region mapping a declare target
! Fortran array writing some values to it and checking the host
! correctly receives the updates made on the device.
-! REQUIRES: flang, amdgcn-amd-amdhsa
-! UNSUPPORTED: nvptx64-nvidia-cuda
+! REQUIRES: flang, amdgcn-amd-amdhsa, nvptx64-nvidia-cuda
! UNSUPPORTED: nvptx64-nvidia-cuda-LTO
! UNSUPPORTED: aarch64-unknown-linux-gnu
! UNSUPPORTED: aarch64-unknown-linux-gnu-LTO
diff --git a/openmp/libomptarget/test/offloading/fortran/double-target-call-with-declare-target.f90 b/openmp/libomptarget/test/offloading/fortran/double-target-call-with-declare-target.f90
index b4c793ca06cf798..884acb275a0eb47 100644
--- a/openmp/libomptarget/test/offloading/fortran/double-target-call-with-declare-target.f90
+++ b/openmp/libomptarget/test/offloading/fortran/double-target-call-with-declare-target.f90
@@ -2,8 +2,7 @@
! declare target For...
[truncated]
|
fabianmcg
requested review from
clementval,
kiranchandramohan,
jdoerfert and
jeanPerier
|
Ping for review. |
kiranchandramohan
requested review from
skatrak,
jsjodin,
agozillon,
TIFitis and
banach-space
kiranchandramohan
approved these changes
Nov 16, 2023
| @@ -175,6 +175,45 @@ getExplicitAndImplicitAMDGPUTargetFeatures(CompilerInstance &ci, | |||
| return llvm::join(featuresVec, ","); | |||
| } | |||
|
|
|||
| // Get feature string which represents combined explicit target features | |||
| // for NVPTX and the target features specified by the user/ | |||
| // TODO: Have a more robust target conf like `clang/lib/Basic/Targets/NVPTX.cpp` | |||
There was a problem hiding this comment.
If this code can be refactored and placed in llvm/lib/Frontend/Driver then both clang and flang can share it.
There was a problem hiding this comment.
Sure, it should be doable, AMDGPU also needs a similar refactor, however, I'm thinking (and waiting) in adding them to llvm/offload as MLIR would also benefit from these refactors.
Yes, on an NVIDIA V100. |
sr-tream
pushed a commit
to sr-tream/llvm-project
that referenced
this pull request
Nov 20, 2023
This patch adds initial support to the NVPTX target, enabling `flang` to produce OpenMP offload code for NVPTX targets.
zahiraam
pushed a commit
to zahiraam/llvm-project
that referenced
this pull request
Nov 20, 2023
This patch adds initial support to the NVPTX target, enabling `flang` to produce OpenMP offload code for NVPTX targets.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This patch adds initial support to the NVPTX target, enabling
flangto produce OpenMP offload code for NVPTX targets.