diff --git a/examples_cuda/cuda_ispc.h b/examples_cuda/cuda_ispc.h index 7b034411..e143d90d 100644 --- a/examples_cuda/cuda_ispc.h +++ b/examples_cuda/cuda_ispc.h @@ -151,7 +151,8 @@ static CUmodule loadModule( #endif { // Load the PTX from the string myPtx (64-bit) - fprintf(stderr, "Loading ptx..\n"); + if (print_log) + fprintf(stderr, "Loading ptx..\n"); myErr = cuLinkAddData(*lState, CU_JIT_INPUT_PTX, (void*)module, strlen(module)+1, 0, 0, 0, 0); myErr = cuLinkAddFile(*lState, CU_JIT_INPUT_LIBRARY, cudadevrt_lib, 0,0,0); // PTX May also be loaded from file, as per below. @@ -177,7 +178,8 @@ static CUmodule loadModule( // Destroy the linker invocation checkCudaErrors(cuLinkDestroy(*lState)); - fprintf(stderr, " loadModule took %g ms \n", 1e3*(rtc() - t0)); + if (print_log) + fprintf(stderr, " loadModule took %g ms \n", 1e3*(rtc() - t0)); return cudaModule; } static void unloadModule(CUmodule &cudaModule) diff --git a/examples_cuda/options/options.cpp b/examples_cuda/options/options.cpp index 597218e3..6aad3dc6 100644 --- a/examples_cuda/options/options.cpp +++ b/examples_cuda/options/options.cpp @@ -143,6 +143,7 @@ int main(int argc, char *argv[]) { // Binomial options, serial implementation // double binomial_serial = 1e30; +#if 0 for (int i = 0; i < 3; ++i) { reset_and_start_timer(); const double t0 = rtc(); @@ -159,6 +160,7 @@ int main(int argc, char *argv[]) { printf("\t\t\t\t(%.2fx speedup from ISPC, %.2fx speedup from ISPC + tasks)\n", binomial_serial / binomial_ispc, binomial_serial / binomial_tasks); +#endif // // Black-Scholes options pricing model, ispc implementation, 1 thread @@ -200,6 +202,7 @@ int main(int argc, char *argv[]) { // Black-Scholes options pricing model, serial implementation // double bs_serial = 1e30; +#if 0 for (int i = 0; i < 3; ++i) { reset_and_start_timer(); const double t0 = rtc(); @@ -216,6 +219,7 @@ int main(int argc, char *argv[]) { printf("\t\t\t\t(%.2fx speedup from ISPC, %.2fx speedup from ISPC + tasks)\n", bs_serial / bs_ispc, bs_serial / bs_ispc_tasks); +#endif return 0; } diff --git a/examples_cuda/options/options.ispc b/examples_cuda/options/options.ispc index 8cf69cf5..57af076b 100644 --- a/examples_cuda/options/options.ispc +++ b/examples_cuda/options/options.ispc @@ -76,10 +76,12 @@ export void black_scholes_ispc_tasks(uniform float Sa[], uniform float Xa[], uniform float Ta[], uniform float ra[], uniform float va[], uniform float result[], uniform int count) { - uniform int nTasks = max((int)64, (int)count/16384); + uniform int nTasks = 2048; //count/16384; //max((int)64, (int)count/16384); launch[nTasks] bs_task(Sa, Xa, Ta, ra, va, result, count); } +/********/ + export void black_scholes_ispc(uniform float Sa[], uniform float Xa[], uniform float Ta[], @@ -149,6 +151,6 @@ binomial_put_ispc_tasks(uniform float Sa[], uniform float Xa[], uniform float Ta[], uniform float ra[], uniform float va[], uniform float result[], uniform int count) { - uniform int nTasks = max((int)64, (int)count/16384); + uniform int nTasks = 2048; //count/16384; //max((int)64, (int)count/16384); launch[nTasks] binomial_task(Sa, Xa, Ta, ra, va, result, count); }