Skip to content
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

avoid embedded macro directive in macro function argument because it'… #959

Merged
merged 1 commit into from
Jun 1, 2020
+52 −57
Merged

avoid embedded macro directive in macro function argument because it'… #959

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
6 changes: 3 additions & 3 deletions Src/Base/AMReX_Arena.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,7 +78,7 @@ Arena::allocate_system (std::size_t nbytes)
AMREX_HIP_OR_CUDA_OR_DPCPP
(AMREX_HIP_SAFE_CALL(hipMalloc(&p, nbytes));,
AMREX_CUDA_SAFE_CALL(cudaMallocManaged(&p, nbytes));,
p = sycl::malloc_shared(nbytes, Gpu::Device::syclDevice(), Gpu::Device::syclContext()););
p = sycl::malloc_shared(nbytes, Gpu::Device::syclDevice(), Gpu::Device::syclContext()));
if (arena_info.device_set_readonly)
{
Gpu::Device::mem_advise_set_readonly(p, nbytes);
Expand All @@ -94,7 +94,7 @@ Arena::allocate_system (std::size_t nbytes)
AMREX_HIP_OR_CUDA_OR_DPCPP
(AMREX_HIP_SAFE_CALL ( hipMalloc(&p, nbytes));,
AMREX_CUDA_SAFE_CALL(cudaMalloc(&p, nbytes));,
p = sycl::malloc_device(nbytes, Gpu::Device::syclDevice(), Gpu::Device::syclContext()););
p = sycl::malloc_device(nbytes, Gpu::Device::syclDevice(), Gpu::Device::syclContext()));
}
}
#else
Expand Down Expand Up @@ -126,7 +126,7 @@ Arena::deallocate_system (void* p, std::size_t nbytes)
AMREX_HIP_OR_CUDA_OR_DPCPP
(AMREX_HIP_SAFE_CALL ( hipFree(p));,
AMREX_CUDA_SAFE_CALL(cudaFree(p));,
sycl::free(p,Gpu::Device::syclContext()););
sycl::free(p,Gpu::Device::syclContext()));
}
#else
if (p && arena_info.device_use_hostalloc) munlock(p, nbytes);
Expand Down
32 changes: 16 additions & 16 deletions Src/Base/AMReX_GpuAsyncArray.H
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,14 +14,14 @@
#define HIPRT_CB
#endif

#if defined(AMREX_USE_GPU) && !defined(AMREX_USE_DPCPP)
#if defined(AMREX_USE_CUDA) || defined(AMREX_USE_HIP)
extern "C" {
AMREX_HIP_OR_CUDA(
void HIPRT_CB amrex_asyncarray_delete ( hipStream_t stream, hipError_t error, void* p);,
#if ( defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10) )
void CUDART_CB amrex_asyncarray_delete (void* p);)
#else
void CUDART_CB amrex_asyncarray_delete (cudaStream_t stream, cudaError_t error, void* p);)
#if defined(AMREX_USE_HIP)
void HIPRT_CB amrex_asyncarray_delete ( hipStream_t stream, hipError_t error, void* p);
#elif defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10)
void CUDART_CB amrex_asyncarray_delete (void* p);
#elif defined(AMREX_USE_CUDA)
void CUDART_CB amrex_asyncarray_delete (cudaStream_t stream, cudaError_t error, void* p);
#endif
}
#endif
Expand Down Expand Up @@ -78,15 +78,15 @@ public:
T** p = static_cast<T**>(std::malloc(2*sizeof(T*)));
p[0] = d_data;
p[1] = h_data;
AMREX_HIP_OR_CUDA(
AMREX_HIP_SAFE_CALL ( hipStreamAddCallback(Gpu::gpuStream(),
amrex_asyncarray_delete, p, 0));,
#if ( defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10) )
AMREX_CUDA_SAFE_CALL(cudaLaunchHostFunc(Gpu::gpuStream(),
amrex_asyncarray_delete, p)););
#else
AMREX_CUDA_SAFE_CALL(cudaStreamAddCallback(Gpu::gpuStream(),
amrex_asyncarray_delete, p, 0)););
#if defined(AMREX_USE_HIP)
AMREX_HIP_SAFE_CALL ( hipStreamAddCallback(Gpu::gpuStream(),
amrex_asyncarray_delete, p, 0));
#elif defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10)
AMREX_CUDA_SAFE_CALL(cudaLaunchHostFunc(Gpu::gpuStream(),
amrex_asyncarray_delete, p));
#elif defined(AMREX_USE_CUDA)
AMREX_CUDA_SAFE_CALL(cudaStreamAddCallback(Gpu::gpuStream(),
amrex_asyncarray_delete, p, 0));
#endif
Gpu::callbackAdded();
#else
Expand Down
17 changes: 6 additions & 11 deletions Src/Base/AMReX_GpuAsyncArray.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,20 +1,15 @@
#include <AMReX_GpuAsyncArray.H>
#include <mutex>

#ifdef AMREX_USE_GPU

#ifdef __HIP_PLATFORM_HCC__
#define HIPRT_CB
#endif

#if !defined(AMREX_USE_DPCPP)
extern "C" {
AMREX_HIP_OR_CUDA(
void HIPRT_CB amrex_asyncarray_delete ( hipStream_t stream, hipError_t error, void* p),
#if ( defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10) )
void CUDART_CB amrex_asyncarray_delete (void* p))
#else
void CUDART_CB amrex_asyncarray_delete (cudaStream_t stream, cudaError_t error, void* p))
#if defined(AMREX_USE_HIP)
void HIPRT_CB amrex_asyncarray_delete ( hipStream_t stream, hipError_t error, void* p)
#elif defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10)
void CUDART_CB amrex_asyncarray_delete (void* p)
#elif defined(AMREX_USE_CUDA)
void CUDART_CB amrex_asyncarray_delete (cudaStream_t stream, cudaError_t error, void* p)
#endif
{
void** pp = (void**)p;
Expand Down
2 changes: 1 addition & 1 deletion Src/Base/AMReX_GpuDevice.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,7 +124,7 @@ Device::Initialize ()
AMREX_HIP_OR_CUDA_OR_DPCPP
( amrex::Print() << "Initializing HIP...\n";,
amrex::Print() << "Initializing CUDA...\n";,
amrex::Print() << "Initializing oneAPI...\n"; );
amrex::Print() << "Initializing oneAPI...\n"; )
}

// XL CUDA Fortran support needs to be initialized
Expand Down
30 changes: 15 additions & 15 deletions Src/Base/AMReX_GpuElixir.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,12 +18,12 @@ namespace {
#if defined(AMREX_USE_GPU) && !defined(AMREX_USE_DPCPP)

extern "C" {
AMREX_HIP_OR_CUDA(
void HIPRT_CB amrex_elixir_delete ( hipStream_t stream, hipError_t error, void* p),
#if ( defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10) )
void CUDART_CB amrex_elixir_delete (void* p))
#else
void CUDART_CB amrex_elixir_delete (cudaStream_t stream, cudaError_t error, void* p))
#if defined(AMREX_USE_HIP)
void HIPRT_CB amrex_elixir_delete ( hipStream_t stream, hipError_t error, void* p)
#elif defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10)
void CUDART_CB amrex_elixir_delete (void* p)
#elif defined(AMREX_USE_CUDA)
void CUDART_CB amrex_elixir_delete (cudaStream_t stream, cudaError_t error, void* p)
#endif
{
void** pp = (void**)p;
Expand All @@ -49,15 +49,15 @@ Elixir::clear () noexcept
void** p = static_cast<void**>(std::malloc(2*sizeof(void*)));
p[0] = m_p;
p[1] = (void*)m_arena;
AMREX_HIP_OR_CUDA(
AMREX_HIP_SAFE_CALL ( hipStreamAddCallback(Gpu::gpuStream(),
amrex_elixir_delete, p, 0));,
#if ( defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10) )
AMREX_CUDA_SAFE_CALL(cudaLaunchHostFunc(Gpu::gpuStream(),
amrex_elixir_delete, p)););
#else
AMREX_CUDA_SAFE_CALL(cudaStreamAddCallback(Gpu::gpuStream(),
amrex_elixir_delete, p, 0)););
#if defined(AMREX_USE_HIP)
AMREX_HIP_SAFE_CALL ( hipStreamAddCallback(Gpu::gpuStream(),
amrex_elixir_delete, p, 0));
#elif defined(__CUDACC__) && (__CUDACC_VER_MAJOR__ >= 10)
AMREX_CUDA_SAFE_CALL(cudaLaunchHostFunc(Gpu::gpuStream(),
amrex_elixir_delete, p));
#elif defined(AMREX_USE_CUDA)
AMREX_CUDA_SAFE_CALL(cudaStreamAddCallback(Gpu::gpuStream(),
amrex_elixir_delete, p, 0));
#endif
Gpu::callbackAdded();
#elif defined(AMREX_USE_DPCPP)
Expand Down
2 changes: 1 addition & 1 deletion Src/Base/AMReX_GpuReduce.H
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -110,7 +110,7 @@ struct warpReduce
//#ifdef AMREX_USE_CUDA
for (int offset = warpSize/2; offset > 0; offset /= 2) {
AMREX_HIP_OR_CUDA(T y = __shfl_down(x, offset);,
T y = __shfl_down_sync(0xffffffff, x, offset); );
T y = __shfl_down_sync(0xffffffff, x, offset); )
x = F()(x,y);
}
//#endif
Expand Down
14 changes: 7 additions & 7 deletions Src/Base/AMReX_Random.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -127,10 +127,10 @@ amrex::RandomNormal (amrex::Real mean, amrex::Real stddev)
int i = get_state(tid);
#ifdef BL_USE_FLOAT
AMREX_HIP_OR_CUDA( rand = stddev * hiprand_normal(&d_states_d_ptr[i]) + mean;,
rand = stddev * curand_normal(&d_states_d_ptr[i]) + mean; );
rand = stddev * curand_normal(&d_states_d_ptr[i]) + mean; )
#else
AMREX_HIP_OR_CUDA( rand = stddev * hiprand_normal_double(&d_states_d_ptr[i]) + mean;,
rand = stddev * curand_normal_double(&d_states_d_ptr[i]) + mean; );
rand = stddev * curand_normal_double(&d_states_d_ptr[i]) + mean; )
#endif
__threadfence();
free_state(tid);
Expand Down Expand Up @@ -172,10 +172,10 @@ amrex::Random ()
// std::uniform_real_distribution in [0.0, 1.0)
#ifdef BL_USE_FLOAT
AMREX_HIP_OR_CUDA( rand = 1.0f - hiprand_uniform(&d_states_d_ptr[i]);,
rand = 1.0f - curand_uniform(&d_states_d_ptr[i]); );
rand = 1.0f - curand_uniform(&d_states_d_ptr[i]); )
#else
AMREX_HIP_OR_CUDA( rand = 1.0 - hiprand_uniform_double(&d_states_d_ptr[i]);,
rand = 1.0 - curand_uniform_double(&d_states_d_ptr[i]); );
rand = 1.0 - curand_uniform_double(&d_states_d_ptr[i]); )
#endif

__threadfence();
Expand Down Expand Up @@ -217,7 +217,7 @@ amrex::RandomPoisson (amrex::Real lambda)
const auto i = get_state(tid);

AMREX_HIP_OR_CUDA( rand = hiprand_poisson(&d_states_d_ptr[i], lambda);,
rand = curand_poisson(&d_states_d_ptr[i], lambda););
rand = curand_poisson(&d_states_d_ptr[i], lambda);)

__threadfence();
free_state(tid);
Expand Down Expand Up @@ -259,7 +259,7 @@ amrex::Random_int (unsigned int n)
int i = get_state(tid);
do {
AMREX_HIP_OR_CUDA( rand = hiprand(&d_states_d_ptr[i]);,
rand = curand(&d_states_d_ptr[i]); );
rand = curand(&d_states_d_ptr[i]); )
} while (rand > (RAND_M - RAND_M % n));
__threadfence();
free_state(tid);
Expand Down Expand Up @@ -415,7 +415,7 @@ amrex::ResizeRandomSeed (int N)
int loc = idx + PrevSize;

AMREX_HIP_OR_CUDA( hiprand_init(seed, seqstart, 0, &d_states_d_ptr[loc]);,
curand_init(seed, seqstart, 0, &d_states_d_ptr[loc]); );
curand_init(seed, seqstart, 0, &d_states_d_ptr[loc]); )
});

#endif
Expand Down
4 changes: 2 additions & 2 deletions Src/Base/AMReX_Scan.H
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -479,7 +479,7 @@ T PrefixSum (int n, FIN && fin, FOUT && fout, Type type)
// Scan within a warp
for (int i = 1; i <= Gpu::Device::warp_size; i *= 2) {
AMREX_HIP_OR_CUDA( T s = __shfl_up(x,i);,
T s = __shfl_up_sync(0xffffffff, x, i); );
T s = __shfl_up_sync(0xffffffff, x, i); )
if (lane >= i) x += s;
}

Expand Down Expand Up @@ -585,7 +585,7 @@ T PrefixSum (int n, FIN && fin, FOUT && fout, Type type)

for (int i = Gpu::Device::warp_size/2; i > 0; i /= 2) {
AMREX_HIP_OR_CUDA( x += __shfl_down(x,i);,
x += __shfl_down_sync(0xffffffff, x, i); );
x += __shfl_down_sync(0xffffffff, x, i); )
}
}

Expand Down
2 changes: 1 addition & 1 deletion Tests/GPU/Locking/main.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,7 +153,7 @@ void blockCountingTest ()
#endif

AMREX_HIP_OR_CUDA( hipPeekAtLastError();,
cudaPeekAtLastError(); );
cudaPeekAtLastError(); )

Gpu::Device::synchronize();

Expand Down