不支持外部呼叫 - CUDA [英] External calls are not supported - CUDA

问题描述

目的是调用另一个文件中可用的设备函数，当我编译全局内核时，它显示以下错误 *不支持外部调用（发现非内联调用_Z6GoldenSectionCUDA ）* 。

Objective is to call a device function available in another file, when i compile the global kernel it shows the following error *External calls are not supported (found non-inlined call to _Z6GoldenSectionCUDA)*.

问题代码（不是完整代码，但出现问题的地方），
cat norm.h

Problematic Code (not the full code but where the problem arises), cat norm.h

# ifndef NORM_H_
# define NORM_H_
# include<stdio.h>

__device__ double invcdf(double prob, double mean, double stddev);

#endif

cat norm.cu

cat norm.cu

# include <norm.h>

__device__ double invcdf(double prob, double mean, double stddev) {
    return (mean + stddev*normcdfinv(prob));
       }

cat test.cu

cat test.cu

# include <norm.h>
# include <curand.h>
# include <curand_kernel.h>

__global__ void phase2Kernel(double* out_profit, struct strategyHolder* strategy) {
       curandState seedValue;
       curand_init(threadIdx.x, 0, 0, &seedValue);
       double randomD = invcdf(curand_uniform_double( &seedValue ), 300, 80);
    }

nvcc -c norm.cu -o norm.o -I

nvcc -c test.cu -o test.o -I。

nvcc -c norm.cu -o norm.o -I"."
nvcc -c test.cu -o test.o -I"."

推荐答案

重新尝试单独编译，这需要一些特殊的命令行选项。请参见 NVCC手册，但这里是如何让你的示例编译。我已经针对sm_20，但你可以根据你有什么GPU目标sm_20或更高版本。

You're trying to do separate compilation, which needs some special command line options. See the NVCC manual for details, but here's how to get your example to compile. I've targeted sm_20, but you can target sm_20 or later depending on what GPU you have. Separate compilation is not possible on older devices (sm_1x).

• 您不需要声明 __ device __ 作为 extern 在你的头文件，但如果你有任何静态设备变量，他们需要被声明为 extern



• 通过编译如下所示生成设备的可重定位代码（ -dc 等效于 -c ，请参阅手册了解详情）

• You don't need to declare the __device__ function as extern in your header file, but if you have any static device variables they will need to be declared as extern

• Generate relocatable code for the device by compiling as shown below (-dc is the device equivalent of -c, see the manual for more information)

nvcc -arch=sm_20 -dc norm.cu -o norm.o -I.
nvcc -arch=sm_20 -dc test.cu -o test.o -I.

在最终主机链接之前通过调用nvlink链接代码的设备部分

Link the device parts of the code by calling nvlink before the final host link

nvlink -arch=sm_20 norm.o test.o -o final.o

这篇关于不支持外部呼叫 - CUDA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！