直接将PTX程序传递给CUDA驱动程序

问题描述

CUDA驱动程序API提供了从文件系统加载包含PTX代码的文件。通常执行以下操作：

The CUDA driver API provides loading the file containing PTX code from the filesystem. One usually does the following:

CUmodule module;
CUfunction function;

const char* module_file = "my_prg.ptx";
const char* kernel_name = "vector_add";

err = cuModuleLoad(&module, module_file);
err = cuModuleGetFunction(&function, module, kernel_name);

如果在运行时（运行中）生成PTX文件并通过文件IO，似乎是

In case one generates the PTX files during runtime (on the fly) going through file IO seems to be a waste (since the driver has to load it back in again).

有没有办法将PTX程序直接传递给CUDA驱动程序（例如，作为C字符串） ）？

Is there a way to pass the PTX program to the CUDA driver directly (e.g. as a C string) ?

推荐答案

取自 ptxjit CUDA示例：

将PTX程序定义为C字符串

Define the PTX program as a C string as

char myPtx32[] = "\n\
    .version 1.4\n\
    .target sm_10, map_f64_to_f32\n\
    .entry _Z8myKernelPi (\n\.param .u32 __cudaparm__Z8myKernelPi_data)\n\
    {\n\
    .reg .u16 %rh<4>;\n\
    .reg .u32 %r<8>;\n\

    // Other stuff

    .loc    28      18      0\n\
    exit;\n\
    }\n\
 ";

然后

 cuModuleLoadDataEx(phModule, myPtx32, 0, 0, 0);

最后

 cuModuleLoadDataEx(phModule, myPtx, 0, 0, 0);

这篇关于直接将PTX程序传递给CUDA驱动程序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！