CudaErrorUnknown code=30 on any cuda call - c++

Question

I've installed cuda toolkit and I can run the samples without a problem. Now, I want to use cuda on my project and in my project I use cmake. So, in order to demonstrate my problem I created a simple example. I have 3 files, my main, which is "teste.cpp", a cuda file "hello_world.cu" and it's header. The only thing my main has is the call for a function at hello_world.cu, like this:
#include <iostream>
#include "hello_world.h"
using namespace std;
int main(int argc, char** argv)
{
teste(argc, argv);
return 0;
}
My hello_world.cu is a exact copy of the "clock" sample. So, looks like this:
// CUDA runtime
#include </usr/local/cuda-9.0/include/cuda_runtime.h>
// helper functions and utilities to work with CUDA
#include </usr/local/cuda-9.0/samples/common/inc/helper_functions.h>
#include </usr/local/cuda-9.0/samples/common/inc/helper_cuda.h>
#define NUM_BLOCKS 64
#define NUM_THREADS 256
__global__ static void timedReduction(const float *input, float *output, clock_t *timer)
{
// __shared__ float shared[2 * blockDim.x];
extern __shared__ float shared[];
const int tid = threadIdx.x;
const int bid = blockIdx.x;
if (tid == 0) timer[bid] = clock();
// Copy input.
shared[tid] = input[tid];
shared[tid + blockDim.x] = input[tid + blockDim.x];
// Perform reduction to find minimum.
for (int d = blockDim.x; d > 0; d /= 2)
{
__syncthreads();
if (tid < d)
{
float f0 = shared[tid];
float f1 = shared[tid + d];
if (f1 < f0)
{
shared[tid] = f1;
}
}
}
// Write result.
if (tid == 0) output[bid] = shared[0];
__syncthreads();
if (tid == 0) timer[bid+gridDim.x] = clock();
}
int teste(int argc, char** argv) {
printf("CUDA Clock sample\n");
// This will pick the best possible CUDA capable device
int dev = findCudaDevice(argc, (const char **)argv);
float *dinput = NULL;
float *doutput = NULL;
clock_t *dtimer = NULL;
clock_t timer[NUM_BLOCKS * 2];
float input[NUM_THREADS * 2];
for (int i = 0; i < NUM_THREADS * 2; i++)
{
input[i] = (float)i;
}
checkCudaErrors(cudaMalloc((void **)&dinput, sizeof(float) * NUM_THREADS * 2));
checkCudaErrors(cudaMalloc((void **)&dinput, sizeof(float) * NUM_THREADS * 2));
checkCudaErrors(cudaMalloc((void **)&doutput, sizeof(float) * NUM_BLOCKS));
checkCudaErrors(cudaMalloc((void **)&dtimer, sizeof(clock_t) * NUM_BLOCKS * 2));
checkCudaErrors(cudaMemcpy(dinput, input, sizeof(float) * NUM_THREADS * 2, cudaMemcpyHostToDevice));
timedReduction<<<NUM_BLOCKS, NUM_THREADS, sizeof(float) * 2 *NUM_THREADS>>>(dinput, doutput, dtimer);
checkCudaErrors(cudaMemcpy(timer, dtimer, sizeof(clock_t) * NUM_BLOCKS * 2, cudaMemcpyDeviceToHost));
checkCudaErrors(cudaFree(dinput));
checkCudaErrors(cudaFree(doutput));
checkCudaErrors(cudaFree(dtimer));
long double avgElapsedClocks = 0;
for (int i = 0; i < NUM_BLOCKS; i++)
{
avgElapsedClocks += (long double) (timer[i + NUM_BLOCKS] - timer[i]);
}
avgElapsedClocks = avgElapsedClocks/NUM_BLOCKS;
printf("Average clocks/block = %Lf\n", avgElapsedClocks);
return EXIT_SUCCESS;
}
My CMakeLists.txt looks like this:
cmake_minimum_required(VERSION 2.8)
set(CUDA_HOST_COMPILER /usr/bin/g++-4.9)
find_package(CUDA QUIET REQUIRED)
# Pass options to NVCC
set(
CUDA_NVCC_FLAGS
${CUDA_NVCC_FLAGS};
-O3 -std=c++11 -g
)
# For compilation ...
# Specify target & source files to compile it from
cuda_add_executable(
helloworld
teste.cpp
hello_world.cu
)
The code compiles, and when i run it i get this output:
CUDA Clock sample
GPU Device 0: "GeForce GTX 950M" with compute capability 5.0
CUDA error at /home/cesar/Documents/cuda_teste/hello_world.cu:69 code=30(cudaErrorUnknown) "cudaMalloc((void **)&dinput, sizeof(float) * NUM_THREADS * 2)"
Why do I get this error here, using cmake? If I go to the samples directory and try the 'clock' example directly everything works fine.. So is it a problem on my CMakeLists.txt?

Answer 1

I managed to find the solution.
On my CMakeLists.txt i needed to add a flag to my NVCC with "-arch=sm_50", in my case it is sm_50 due to my graphic card having a compute capability 5.0, if any one have the same error and want to try this, you have to check your compute capability