Hi,
I am playing with Cuda and Emscripten, and I have some trouble with this
syntax : transformKernel<<<dimGrid, dimBlock, 0>>>(dData, width, height,
angle);
It's the part who call the cuda kernel. For solve the problem I was
thinking make an custom emscripten function, unfortunately the function can
take N arguments and any type.
I was trying to solve that using a function template inside an header file,
and write the code in my library_cuda.js
The header file is :
template<typename T0,typename T1,typename T2,typename T3>
extern int cudaRunKernelDim4(const char * kernel_name, const char *
kernel_source, const char* option, dim3 blocksPerGrid, dim3
threadsPerBlock, int num_args, T0 P0, T1 P1, T2 P2, T3 P3);
The c++ code give :
#ifdef __EMSCRIPTEN__
cudaRunKernelDim4<float*,unsigned int,unsigned int,
float>("transformKernel", transformKernel, "", dimGrid, dimBlock , 4,
dData, width, height, angle);
#else
transformKernel<<<dimGrid, dimBlock, 0>>>(dData, width, height, angle);
#endif
The JS code is :
cudaRunKernelDim: function(kernel_name, kernel_source, options,
blocksPerGrid, threadsPerBlock, num_args, P1, P2, P3, P4, P5, P6, P7, P8,
P9, P10) {
....
}
cudaRunKernelDim4: 'cudaRunKernelDim',
But when I build it with emscripten I have an unresolved symbol :
*warning: unresolved symbol:
_Z14cudaRunKernelDim4IPfS0_S0_iEiPKcS2_S2_iiiT_T0_T1_T2_*
*What I am doing wrong ?? *
*Alon do you have any suggestion ??*
*Thanks*
*Tony*
--
You received this message because you are subscribed to the Google Groups
"emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
