portelli/Grid
1
0
Fork 0
mirror of https://github.com/paboyle/Grid.git synced 2025-06-17 15:27:06 +01:00
Code Releases Activity

Sycl happier

Browse Source
This commit is contained in:
u61464
2021-03-03 11:21:43 -08:00
parent 442336bd96
commit 679d1d22f7
13 changed files with 64 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
Grid/tensors/Tensor_SIMT.h
View File

@ -66,7 +66,7 @@ void coalescedWriteNonTemporal(vobj & __restrict__ vec,const vobj & __restrict__
#ifndef GRID_SYCL
// Use the scalar as our own complex on GPU
// Use the scalar as our own complex on GPU ... thrust::complex or std::complex
template<class vsimd,IfSimd<vsimd> = 0> accelerator_inline
typename vsimd::scalar_type
coalescedRead(const vsimd & __restrict__ vec,int lane=acceleratorSIMTlane(vsimd::Nsimd()))
@ -96,6 +96,8 @@ void coalescedWrite(vsimd & __restrict__ vec,
p[lane]=extracted;
}
#else
// For SyCL have option to use GpuComplex from inside the vector type in SIMT loops
// Faster for some reason
template<class vsimd,IfSimd<vsimd> = 0> accelerator_inline
typename vsimd::vector_type::datum
coalescedRead(const vsimd & __restrict__ vec,int lane=acceleratorSIMTlane(vsimd::Nsimd()))