如何访问我的内核中的常量内存？ [英] How can I access my constant memory in my kernel?

问题描述

我无法管理访问我的常量内存中的数据，我不知道为什么。这是我的代码片段：

I can't manage to access the data in my constant memory and I don't know why. Here is a snippet of my code:

#define N 10
__constant__ int constBuf_d[N];

__global__ void foo( int *results, int *constBuf )
{
    int tdx = threadIdx.x;
    int idx = blockIdx.x * blockDim.x + tdx;

    if( idx < N )
    {
         results[idx] = constBuf[idx];
    }
}

// main routine that executes on the host
int main(int argc, char* argv[])
{
    int *results_h = new int[N];
    int *results_d = NULL;

    cudaMalloc((void **)&results_d, N*sizeof(int));

    int arr[10] = { 16, 2, 77, 40, 12, 3, 5, 3, 6, 6 };

    int *cpnt;
    cudaError_t err = cudaGetSymbolAddress((void **)&cpnt, "constBuf_d");

    if( err )
        cout << "error!";

    cudaMemcpyToSymbol((void**)&cpnt, arr, N*sizeof(int), 0, cudaMemcpyHostToDevice);

    foo <<< 1, 256 >>> ( results_d, cpnt );

    cudaMemcpy(results_h, results_d, N*sizeof(int), cudaMemcpyDeviceToHost);

    for( int i=0; i < N; ++i )
        printf("%i ", results_h[i] );
}

由于某种原因，我在results_h中只得到0我使用能力1.1的卡运行CUDA 4.0。

For some reason, I only get "0" in results_h. I'm running CUDA 4.0 with a card with capability 1.1.

任何想法？感谢！

推荐答案

如果您对代码添加适当的错误检查，您会发现 cudaMemcpyToSymbol 出现无效的设备符号错误。您需要通过名称传递符号，或使用 cudaMemcpy 。因此：

If you add proper error checking to your code, you will find that the cudaMemcpyToSymbol is failing with a invalid device symbol error. You either need to pass the symbol by name, or use cudaMemcpy instead. So this:

cudaGetSymbolAddress((void **)&cpnt, "constBuf_d");
cudaMemcpy(cpnt, arr, N*sizeof(int), cudaMemcpyHostToDevice); 

或

cudaMemcpyToSymbol("constBuf_d", arr, N*sizeof(int), 0, cudaMemcpyHostToDevice);

或

cudaMemcpyToSymbol(constBuf_d, arr, N*sizeof(int), 0, cudaMemcpyHostToDevice);

可以工作。话虽如此，将常量内存地址作为参数传递给内核是使用常量内存的错误方式 - 它使编译器无法生成指令以通过常量内存缓存访问内存。比较为您的内核生成的计算能力1.2 PTX：

will work. Having said that, passing a constant memory address as an argument to a kernel is the wrong way to use constant memory - it defeats the compiler from generating instructions to access memory via the constant memory cache. Compare the compute capability 1.2 PTX generated for your kernel:

    .entry _Z3fooPiS_ (
        .param .u32 __cudaparm__Z3fooPiS__results,
        .param .u32 __cudaparm__Z3fooPiS__constBuf)
    {
    .reg .u16 %rh<4>;
    .reg .u32 %r<12>;
    .reg .pred %p<3>;
    .loc    16  7   0
$LDWbegin__Z3fooPiS_:
    mov.u16     %rh1, %ctaid.x;
    mov.u16     %rh2, %ntid.x;
    mul.wide.u16    %r1, %rh1, %rh2;
    cvt.s32.u16     %r2, %tid.x;
    add.u32     %r3, %r2, %r1;
    mov.u32     %r4, 9;
    setp.gt.s32     %p1, %r3, %r4;
    @%p1 bra    $Lt_0_1026;
    .loc    16  14  0
    mul.lo.u32  %r5, %r3, 4;
    ld.param.u32    %r6, [__cudaparm__Z3fooPiS__constBuf];
    add.u32     %r7, %r6, %r5;
    ld.global.s32   %r8, [%r7+0];
    ld.param.u32    %r9, [__cudaparm__Z3fooPiS__results];
    add.u32     %r10, %r9, %r5;
    st.global.s32   [%r10+0], %r8;
$Lt_0_1026:
    .loc    16  16  0
    exit;
$LDWend__Z3fooPiS_:
    } // _Z3fooPiS_

p>

with this kernel:

__global__ void foo2( int *results )
{
    int tdx = threadIdx.x;
    int idx = blockIdx.x * blockDim.x + tdx;

    if( idx < N )
    {
         results[idx] = constBuf_d[idx];
    }
}

产生

    .entry _Z4foo2Pi (
        .param .u32 __cudaparm__Z4foo2Pi_results)
    {
    .reg .u16 %rh<4>;
    .reg .u32 %r<12>;
    .reg .pred %p<3>;
    .loc    16  18  0
$LDWbegin__Z4foo2Pi:
    mov.u16     %rh1, %ctaid.x;
    mov.u16     %rh2, %ntid.x;
    mul.wide.u16    %r1, %rh1, %rh2;
    cvt.s32.u16     %r2, %tid.x;
    add.u32     %r3, %r2, %r1;
    mov.u32     %r4, 9;
    setp.gt.s32     %p1, %r3, %r4;
    @%p1 bra    $Lt_1_1026;
    .loc    16  25  0
    mul.lo.u32  %r5, %r3, 4;
    mov.u32     %r6, constBuf_d;
    add.u32     %r7, %r5, %r6;
    ld.const.s32    %r8, [%r7+0];
    ld.param.u32    %r9, [__cudaparm__Z4foo2Pi_results];
    add.u32     %r10, %r9, %r5;
    st.global.s32   [%r10+0], %r8;
$Lt_1_1026:
    .loc    16  27  0
    exit;
$LDWend__Z4foo2Pi:
    } // _Z4foo2Pi

请注意，在第二种情况， constBuf_d 通过 ld.const.s32 访问，而不是 ld.global.s32 ，因此使用常量内存缓存。

Note that in the second case, constBuf_d is accessed via ld.const.s32, rather than ld.global.s32, so that constant memory cache is used.

这篇关于如何访问我的内核中的常量内存？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！