Home
last modified time | relevance | path

Searched refs:NumThreads (Results 1 – 25 of 65) sorted by relevance

123

/llvm-project/offload/DeviceRTL/src/
H A DParallelism.cpp53 uint32_t NumThreads = mapping::getMaxTeamThreads(); in determineNumberOfThreads()
55 if (NThreadsICV != 0 && NThreadsICV < NumThreads) in determineNumberOfThreads()
56 NumThreads = NThreadsICV; in determineNumberOfThreads()
61 return NumThreads; in determineNumberOfThreads()
63 if (NumThreads < mapping::getWarpSize()) in determineNumberOfThreads()
64 NumThreads = 1; in determineNumberOfThreads()
66 NumThreads = (NumThreads & ~((uint32_t)mapping::getWarpSize() - 1)); in determineNumberOfThreads()
68 return NumThreads; in determineNumberOfThreads()
92 uint32_t NumThreads in __kmpc_parallel_spmd()
52 uint32_t NumThreads = mapping::getMaxTeamThreads(); determineNumberOfThreads() local
91 uint32_t NumThreads = determineNumberOfThreads(num_threads); __kmpc_parallel_spmd() local
169 uint32_t NumThreads = determineNumberOfThreads(num_threads); __kmpc_parallel_51() local
[all...]
H A DReduction.cpp72 uint32_t NumThreads = omp_get_num_threads(); in nvptx_parallel_reduce_nowait()
73 if (NumThreads == 1) in nvptx_parallel_reduce_nowait()
90 (NumThreads + mapping::getWarpSize() - 1) / mapping::getWarpSize(); in nvptx_parallel_reduce_nowait()
97 if ((NumThreads % mapping::getWarpSize() == 0) || in nvptx_parallel_reduce_nowait()
100 else if (NumThreads > 1) // Only SPMD execution mode comes thru this case. in nvptx_parallel_reduce_nowait()
103 /*LaneCount=*/NumThreads % mapping::getWarpSize(), in nvptx_parallel_reduce_nowait()
110 if (NumThreads > mapping::getWarpSize()) { in nvptx_parallel_reduce_nowait()
140 if (NumThreads > mapping::getWarpSize()) { in nvptx_parallel_reduce_nowait()
142 (NumThreads + mapping::getWarpSize() - 1) / mapping::getWarpSize(); in nvptx_parallel_reduce_nowait()
198 uint32_t NumThreads in __kmpc_nvptx_teams_reduce_nowait_v2()
74 uint32_t NumThreads = omp_get_num_threads(); nvptx_parallel_reduce_nowait() local
197 uint32_t NumThreads = omp_get_num_threads(); __kmpc_nvptx_teams_reduce_nowait_v2() local
[all...]
H A DWorkshare.cpp700 Ty NumBlocks, Ty BId, Ty NumThreads, in NormalizedLoopNestNoChunk() argument
703 Ty KernelIteration = NumBlocks * NumThreads; in NormalizedLoopNestNoChunk()
706 Ty IV = BId * NumThreads + TId; in NormalizedLoopNestNoChunk()
731 Ty ThreadChunk, Ty NumThreads, Ty TId, in NormalizedLoopNestChunked() argument
772 Ty NumIters, Ty NumThreads, Ty ThreadChunk) { in For() argument
796 ASSERT(NumThreads >= NumIters, "Broken assumption"); in For()
802 ThreadChunk, NumThreads, TId, NumIters, in For()
805 NormalizedLoopNestNoChunk(LoopBody, Arg, NumBlocks, BId, NumThreads, TId, in For()
822 Ty NumThreads = 1; in Distribute() local
832 BlockChunk = NumThreads; in Distribute()
858 DistributeFor(IdentTy * Loc,void (* LoopBody)(Ty,void *),void * Arg,Ty NumIters,Ty NumThreads,Ty BlockChunk,Ty ThreadChunk) DistributeFor() argument
[all...]
H A DSynchronization.cpp104 uint32_t NumThreads = omp_get_num_threads();
105 // assert(NumThreads % 32 == 0);
108 uint32_t NumWaves = NumThreads / WarpSize; in atomicAnd()
120 // invariant: NumWaves * WarpSize == NumThreads; in atomicExchange()
223 uint32_t NumThreads = omp_get_num_threads(); in namedBarrier()
224 ASSERT(NumThreads % 32 == 0, nullptr); in namedBarrier()
229 __nvvm_barrier_sync_cnt(BarrierNo, NumThreads); in namedBarrier()
189 uint32_t NumThreads = omp_get_num_threads(); namedBarrier() local
342 uint32_t NumThreads = omp_get_num_threads(); namedBarrier() local
/llvm-project/offload/test/api/
H A Domp_dynamic_shared_memory_mixed.inc9 int Result[N], NumThreads;
13 map(from : Result, NumThreads)
21 NumThreads = omp_get_num_threads();
25 int WrappedTId = (TId + 37) % NumThreads;
33 if (NumThreads < N / 2 || NumThreads > N) {
35 NumThreads);
40 for (int i = 0; i < NumThreads; ++i) {
/llvm-project/compiler-rt/lib/gwp_asan/tests/
H A Dthread_contention.cpp45 void runThreadContentionTest(unsigned NumThreads, unsigned NumIterations, in runThreadContentionTest() argument
50 for (unsigned i = 0; i < NumThreads; ++i) { in runThreadContentionTest()
61 unsigned NumThreads = 4; in TEST_F() local
63 InitNumSlots(NumThreads); in TEST_F()
64 runThreadContentionTest(NumThreads, NumIterations, &GPA); in TEST_F()
H A Dmutex_test.cpp66 static void runSynchronisedTest(unsigned NumThreads, unsigned CounterMax) { in runSynchronisedTest() argument
69 ASSERT_TRUE(CounterMax % NumThreads == 0); in runSynchronisedTest()
75 for (unsigned i = 0; i < NumThreads; ++i) in runSynchronisedTest()
77 CounterMax / NumThreads); in runSynchronisedTest()
/llvm-project/compiler-rt/lib/scudo/standalone/tests/
H A Dcondition_variable_test.cpp18 constexpr scudo::u32 NumThreads = 2; in simpleWaitAndNotifyAll() local
20 std::thread Threads[NumThreads]; in simpleWaitAndNotifyAll()
27 for (scudo::u32 I = 0; I < NumThreads; ++I) { in simpleWaitAndNotifyAll()
32 if (Counter % NumThreads != Id && Counter < CounterMax) in simpleWaitAndNotifyAll()
/llvm-project/llvm/lib/Target/DirectX/
H A DDXILMetadata.cpp
/llvm-project/clang-tools-extra/clangd/unittests/support/
H A DThreadingTests.cpp68 const unsigned NumThreads = 5; in TEST_F() local
78 for (unsigned I = 0; I < NumThreads; ++I) in TEST_F()
90 EXPECT_LE(ComputeCount, NumThreads * NumKeys) in TEST_F()
/llvm-project/offload/plugins-nextgen/common/src/
H A DPluginInterface.cpp529 uint32_t NumThreads[3], in printLaunchInfo()
534 getName(), NumBlocks[0], NumBlocks[1], NumBlocks[2], NumThreads[0], in printLaunchInfoDetails()
535 NumThreads[1], NumThreads[2], getExecutionModeName()); in printLaunchInfoDetails()
536 return printLaunchInfoDetails(GenericDevice, KernelArgs, NumThreads, in printLaunchInfoDetails() argument
542 uint32_t NumThreads[3], in launch()
570 uint32_t NumThreads[3] = {KernelArgs.ThreadLimit[0], in launch()
576 NumThreads[0] = getNumThreads(GenericDevice, NumThreads); in launch()
578 NumThreads[ in launch()
524 printLaunchInfo(GenericDeviceTy & GenericDevice,KernelArgsTy & KernelArgs,uint32_t NumThreads,uint64_t NumBlocks) const printLaunchInfo() argument
556 uint32_t NumThreads = getNumThreads(GenericDevice, KernelArgs.ThreadLimit); launch() local
625 getNumBlocks(GenericDeviceTy & GenericDevice,uint32_t NumTeamsClause[3],uint64_t LoopTripCount,uint32_t & NumThreads,bool IsNumThreadsFromUser) const getNumBlocks() argument
[all...]
/llvm-project/llvm/examples/SpeculativeJIT/
H A DSpeculativeJIT.cpp35 static cl::opt<unsigned> NumThreads("num-threads", cl::Optional,
135 DefaultThreadPool CompileThreads{llvm::hardware_concurrency(NumThreads)};
160 if (NumThreads < 1) { in main()
34 static cl::opt<unsigned> NumThreads("num-threads", cl::Optional, global() variable
/llvm-project/llvm/tools/llvm-dwarfutil/
H A Dllvm-dwarfutil.cpp100 if (opt::Arg *NumThreads = Args.getLastArg(OPT_threads)) in validateAndSetOptions()
101 Options.NumThreads = atoi(NumThreads->getValue()); in validateAndSetOptions()
103 Options.NumThreads = 0; // Use all available hardware threads in validateAndSetOptions()
147 if (Options.NumThreads != 1 && Args.hasArg(OPT_threads)) in validateAndSetOptions()
150 Options.NumThreads = 1; in validateAndSetOptions()
98 if (opt::Arg *NumThreads = Args.getLastArg(OPT_threads)) validateAndSetOptions() local
H A DOptions.h41 int NumThreads = 0; member
/llvm-project/offload/tools/kernelreplay/
H A Dllvm-omp-kernel-replay.cpp74 unsigned NumThreads = in main()
172 TgtArgOffsets.data(), NumArgs.value(), NumTeams, NumThreads, in main()
72 unsigned NumThreads = main() local
/llvm-project/llvm/tools/llvm-gsymutil/
H A Dllvm-gsymutil.cpp96 static unsigned NumThreads;
157 if (!llvm::to_integer(S, NumThreads, 0)) { in parseArgs()
350 NumThreads > 0 ? NumThreads : std::thread::hardware_concurrency(); in handleObjectFile()
93 static unsigned NumThreads; global() variable
/llvm-project/llvm/include/llvm/DebugInfo/GSYM/
H A DDwarfTransformer.h52 /// \param NumThreads The number of threads that the conversion process can
60 llvm::Error convert(uint32_t NumThreads, OutputAggregator &OS);
/llvm-project/llvm/lib/DWARFLinker/Parallel/
H A DDWARFLinkerImpl.h94 void setNumThreads(unsigned NumThreads) override { in setNumThreads() argument
95 GlobalData.Options.Threads = NumThreads; in setNumThreads()
/llvm-project/clang/lib/CodeGen/
H A DCGOpenMPRuntimeGPU.h168 /// \param NumThreads An integer value of threads.
169 void emitNumThreadsClause(CodeGenFunction &CGF, llvm::Value *NumThreads,
231 /// \param NumThreads The value corresponding to the num_threads clause, if
237 const Expr *IfCond, llvm::Value *NumThreads) override;
/llvm-project/llvm/lib/Target/SPIRV/
H A DSPIRVAsmPrinter.cpp482 SmallVector<StringRef> NumThreads; in outputExecutionMode()
483 Attr.getValueAsString().split(NumThreads, ','); in outputExecutionMode()
484 assert(NumThreads.size() == 3 && "invalid numthreads"); in outputExecutionMode()
487 [[maybe_unused]] bool Result = NumThreads[i].getAsInteger(10, V); in outputExecutionMode()
452 SmallVector<StringRef> NumThreads; outputExecutionModeFromNumthreadsAttribute() local
/llvm-project/llvm/tools/llvm-cov/
H A DCoverageViewOptions.h56 unsigned NumThreads;
55 unsigned NumThreads; global() member
H A DCoverageReport.cpp478 ThreadPoolStrategy S = hardware_concurrency(Options.NumThreads); in prepareFileReports()
479 if (Options.NumThreads == 0) { in prepareFileReports()
595 ThreadPoolStrategy PoolS = hardware_concurrency(Options.NumThreads); in prepareDirectoryReports()
596 if (Options.NumThreads == 0) { in prepareDirectoryReports()
/llvm-project/polly/include/polly/CodeGen/
H A DLoopGeneratorsKMP.h90 /// @param NumThreads The number of threads to use.
91 void createCallPushNumThreads(Value *GlobalThreadID, Value *NumThreads);
/llvm-project/offload/plugins-nextgen/amdgpu/src/
H A Drtl.cpp562 Error launchImpl(GenericDeviceTy &GenericDevice, uint32_t NumThreads[3],
569 KernelArgsTy &KernelArgs, uint32_t NumThreads[3],
719 uint32_t NumThreads[3], uint32_t NumBlocks[3],
746 uint16_t Dims = NumBlocks[2] * NumThreads[2] > 1 in pushKernelLaunch()
748 : 1 + (NumBlocks[1] * NumThreads[1] != 1); in pushKernelLaunch()
751 Packet->workgroup_size_x = NumThreads[0]; in pushKernelLaunch()
752 Packet->workgroup_size_y = NumThreads[1]; in pushKernelLaunch()
753 Packet->workgroup_size_z = NumThreads[2]; in pushKernelLaunch()
755 Packet->grid_size_x = NumBlocks[0] * NumThreads[0]; in pushKernelLaunch()
756 Packet->grid_size_y = NumBlocks[1] * NumThreads[ in pushKernelLaunch()
3270 launchImpl(GenericDeviceTy & GenericDevice,uint32_t NumThreads,uint64_t NumBlocks,KernelArgsTy & KernelArgs,KernelLaunchParamsTy LaunchParams,AsyncInfoWrapperTy & AsyncInfoWrapper) const launchImpl() argument
3344 printLaunchInfoDetails(GenericDeviceTy & GenericDevice,KernelArgsTy & KernelArgs,uint32_t NumThreads,uint64_t NumBlocks) const printLaunchInfoDetails() argument
[all...]
/llvm-project/llvm/include/llvm/Frontend/OpenMP/
H A DOMPIRBuilder.h693 /// \param NumThreads The evaluated 'num_threads' clause expression, if any.
702 Value *NumThreads, omp::ProcBindKind ProcBind,
2041 /// \param NumThreads Number of threads via the 'thread_limit' clause.
2047 Value *NumThreads, Value *HostPtr,
2282 ArrayRef<Value *> NumThreads;
2292 ArrayRef<Value *> NumThreads, Value *DynCGGroupMem, in append()
2296 NumThreads(NumThreads), DynCGGroupMem(DynCGGroupMem), in append()
2191 Value *NumThreads; global() member

123