GitHub-Proxy/eigen
1
0
Fork 0
mirror of https://gitlab.com/libeigen/eigen.git synced 2025-04-22 01:29:35 +08:00
Code Issues Packages Projects Releases Wiki Activity

Get rid of nested template specialization in TensorReductionGpu.h, which was broken by c6953f799b01d36f4236b64f351cc1446e0abe17.

Browse Source
This commit is contained in:
Rasmus Munk Larsen 2020-10-13 23:53:11 +00:00
parent c6953f799b
commit 61fc78bbda
1 changed files with 4 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
unsupported/Eigen/CXX11/src/Tensor/TensorReductionGpu.h
View File

@ -80,8 +80,8 @@ __device__ inline double atomicExchCustom(double* address, double val) {
} }
#ifdef EIGEN_HAS_GPU_FP16 #ifdef EIGEN_HAS_GPU_FP16
template <template <typename T> class R> template <typename R>
__device__ inline void atomicReduce(half2* output, half2 accum, R<half>& reducer) { __device__ inline void atomicReduce(half2* output, half2 accum, R& reducer) {
unsigned int oldval = *reinterpret_cast<unsigned int*>(output); unsigned int oldval = *reinterpret_cast<unsigned int*>(output);
unsigned int newval = oldval; unsigned int newval = oldval;
reducer.reducePacket(accum, reinterpret_cast<half2*>(&newval)); reducer.reducePacket(accum, reinterpret_cast<half2*>(&newval));
@ -99,9 +99,8 @@ __device__ inline void atomicReduce(half2* output, half2 accum, R<half>& reducer
} }
} }
// reduction should be associative since reduction is not atomic in wide vector but atomic in half2 operations // reduction should be associative since reduction is not atomic in wide vector but atomic in half2 operations
template <template <typename T> class R> template <typename R>
__device__ inline void atomicReduce(Packet4h2* output, Packet4h2 accum, __device__ inline void atomicReduce(Packet4h2* output, Packet4h2 accum, R& reducer) {
R<half>& reducer) {
half2* houtput=reinterpret_cast<half2*>(output); half2* houtput=reinterpret_cast<half2*>(output);
half2* haccum=reinterpret_cast<half2*>(&accum); half2* haccum=reinterpret_cast<half2*>(&accum);
for(int i=0;i<4;++i){ for(int i=0;i<4;++i){