p21095348
/
InfiniTensor
forked from jiuyuan/InfiniTensor
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix: fix cuda conv_fp16 run fail (#105)

This commit is contained in:
zhangyunze 2023-08-10 15:22:18 +08:00 committed by GitHub
parent 57ac94d893
commit bd9e1aeb3f
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/bang/bang_runtime.cc
View File

@ -13,8 +13,8 @@ void BangRuntimeObj::runWithoutSync(const Graph &graph, bool tune = false,
std::map<OpType, int> opCnt;
for (auto &op : graph->getOperators()) {
// HACK: set correct data type
auto kernelAttrs = KernelAttrs{device, op->getOpType().underlying(),
DataType::Float32};
auto kernelAttrs =
KernelAttrs{device, op->getOpType().underlying(), op->getDType()};
Kernel *kernel = kernelRegistry.getKernel(kernelAttrs);
auto perfKey = PerfEngine::Key{kernelAttrs, op->getOpPerfKey()};
auto perfData = perfEngine.getPerfData(perfKey);

4
src/cuda/cuda_runtime.cc
View File

@ -11,8 +11,8 @@ void CudaRuntimeObj::runWithoutSync(const Graph &graph) const {
auto &perfEngine = PerfEngine::getInstance();
for (auto &op : graph->getOperators()) {
// HACK: set correct data type
auto kernelAttrs = KernelAttrs{device, op->getOpType().underlying(),
DataType::Float32};
auto kernelAttrs =
KernelAttrs{device, op->getOpType().underlying(), op->getDType()};
Kernel *kernel = kernelRegistry.getKernel(kernelAttrs);
auto perfKey = PerfEngine::Key{kernelAttrs, op->getOpPerfKey()};
auto perfData = perfEngine.getPerfData(perfKey);