The processing case when the number of elements exceeds the number of threads in Cuda

Project Package Download

When a vector element exceeds the number of threads

Number of vector elements (=2.x)/(x *)

1 /*2 * Copyright 1993-2010 NVIDIA Corporation. All rights reserved.3 *4 * NVIDIA Corporation and its licensors retain all intellectual property and5 * Proprietary rights in and to this software and related documentation.6 * Any use, reproduction, disclosure, or distribution of this software7 * and related documentation without an express license agreement from8 * NVIDIA Corporation is strictly prohibited.9 *Ten * Refer to the applicable NVIDIA End User License Agreement (EULA) One * Associated with this source code for terms and conditions that govern A * Your use of the This NVIDIA software. - * - */ the  -  -#include".. /common/book.h" -#include"cuda.h" +#include"cuda_runtime.h" -#include"Device_launch_parameters.h" +  A #defineN (33 * 1024) at  -__global__voidAddint*a,int*b,int*c) { -     intTid = threadidx.x + blockidx.x *blockdim.x; -      while(Tid <N) { -C[tid] = A[tid] +B[tid]; -Tid + = blockdim.x *griddim.x; in     } - } to  + intMainvoid) { -     int*a, *b, *C; the     int*dev_a, *dev_b, *Dev_c; *  $     //allocate the memory on the CPUPanax NotoginsengA = (int*) malloc (N *sizeof(int)); -B = (int*) malloc (N *sizeof(int)); thec = (int*) malloc (N *sizeof(int)); +  A     //allocate the memory on the GPU theHandle_error (Cudamalloc (void* *) &dev_a, N *sizeof(int))); +Handle_error (Cudamalloc (void* *) &dev_b, N *sizeof(int))); -Handle_error (Cudamalloc (void* *) &dev_c, N *sizeof(int))); $  $     //fill the arrays ' a ' and ' B ' on the CPU -      for(inti =0; i<n; i++) { -A[i] =i; theB[i] =2*i; -     }Wuyi  the     //Copy the Arrays ' a ' and ' B ' to the GPU -Handle_error (cudamemcpy (Dev_a, A, N *sizeof(int), Wu cudamemcpyhosttodevice)); -Handle_error (cudamemcpy (Dev_b, B, N *sizeof(int), About cudamemcpyhosttodevice)); $  -     /* - when a vector element exceeds the number of threads - number of vector elements (=2.x)/(x *) A     */ +Add << < -, ->> >(Dev_a, Dev_b, Dev_c); the  -     //Copy the array ' C ' back from the GPU to the CPU $Handle_error (cudamemcpy (c, Dev_c, N *sizeof(int), the cudamemcpydevicetohost)); the  the     //Verify that the GPU does the work we requested the     BOOLSuccess =true; -      for(inti =0; i<n; i++) { in         if((A[i] + b[i])! =C[i]) { theprintf"Error:%d +%d! =%d\n", A[i], b[i], c[i]); theSuccess =false; About         } the     } the     if(success) printf ("We did it!\n"); the  +     //The memory we allocated on the GPU - Handle_error (Cudafree (dev_a)); the Handle_error (Cudafree (Dev_b));Bayi Handle_error (Cudafree (Dev_c)); the  the     //Free the memory we allocated on the CPU - Free (a); - Free (b); the Free (c); the  the     return 0; the}

The processing case when the number of elements exceeds the number of threads in Cuda