diff options
Diffstat (limited to 'libcuda/cuda_runtime_api.cc')
| -rw-r--r-- | libcuda/cuda_runtime_api.cc | 64 |
1 files changed, 50 insertions, 14 deletions
diff --git a/libcuda/cuda_runtime_api.cc b/libcuda/cuda_runtime_api.cc index 14e3329..f3b1113 100644 --- a/libcuda/cuda_runtime_api.cc +++ b/libcuda/cuda_runtime_api.cc @@ -479,6 +479,8 @@ void cuda_runtime_api::cuobjdumpRegisterFatBinary(unsigned int handle, const cha fatbinmap[handle] = filename; } + + /******************************************************************************* * Add internal cuda runtime API call to accept gpgpu_context * *******************************************************************************/ @@ -1977,20 +1979,21 @@ __host__ cudaError_t CUDARTAPI cudaLaunch( const char *hostFun ) __host__ cudaError_t CUDARTAPI cudaLaunchKernel ( const char* hostFun, dim3 gridDim, dim3 blockDim, const void** args, size_t sharedMem, cudaStream_t stream ) { - if(g_debug_execution >= 3){ - announce_call(__my_func__); - } - CUctx_st *context = GPGPUSim_Context(); - function_info *entry = context->get_kernel(hostFun); - - cudaConfigureCallInternal(gridDim, blockDim, sharedMem, stream); - for(unsigned i = 0; i < entry->num_args(); i++){ - std::pair<size_t, unsigned> p = entry->get_param_config(i); - cudaSetupArgumentInternal(args[i], p.first, p.second); - } + if(g_debug_execution >= 3){ + announce_call(__my_func__); + } + CUctx_st *context = GPGPUSim_Context(); + function_info *entry = context->get_kernel(hostFun); +#if CUDART_VERSION < 10000 + cudaConfigureCallInternal(gridDim, blockDim, sharedMem, stream); +#endif + for(unsigned i = 0; i < entry->num_args(); i++){ + std::pair<size_t, unsigned> p = entry->get_param_config(i); + cudaSetupArgumentInternal(args[i], p.first, p.second); + } - cudaLaunchInternal(hostFun); - return g_last_cudaError = cudaSuccess; + cudaLaunchInternal(hostFun); + return g_last_cudaError = cudaSuccess; } @@ -2843,9 +2846,43 @@ extern "C" { void** CUDARTAPI __cudaRegisterFatBinary( void *fatCubin ) { + if(g_debug_execution >= 3){ + announce_call(__my_func__); + } return cudaRegisterFatBinaryInternal(fatCubin); } +void CUDARTAPI __cudaRegisterFatBinaryEnd( void **fatCubinHandle ) +{ + if(g_debug_execution >= 3){ + announce_call(__my_func__); + } +} + +unsigned CUDARTAPI __cudaPushCallConfiguration(dim3 gridDim, + dim3 blockDim, + size_t sharedMem = 0, + struct CUstream_st *stream = 0) +{ + if(g_debug_execution >= 3){ + announce_call(__my_func__); + } + cudaConfigureCallInternal(gridDim, blockDim, sharedMem, stream); +} + +cudaError_t CUDARTAPI __cudaPopCallConfiguration( + dim3 *gridDim, + dim3 *blockDim, + size_t *sharedMem, + void *stream +) +{ + if(g_debug_execution >= 3){ + announce_call(__my_func__); + } + return g_last_cudaError = cudaSuccess; +} + void CUDARTAPI __cudaRegisterFunction( void **fatCubinHandle, const char *hostFun, @@ -2902,7 +2939,6 @@ void __cudaUnregisterFatBinary(void **fatCubinHandle) if(g_debug_execution >= 3){ announce_call(__my_func__); } - ; } cudaError_t cudaDeviceReset ( void ) { |