ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-11 17:00:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

Misc fixes (#994)

Browse Source 
* reinterpret_cast to const char* in dumpBufferToFile to be compatible with both const and non-const input pointers

* Add seed input to GeneratorTensor_4 for normal_distribution generator

* Add GetTypeString() for DeviceElementwiseImpl

* Add HIP_CHECK_ERROR macro
This commit is contained in:
Qianfeng
2023-10-20 00:26:04 +08:00
committed by GitHub
parent 82f3a835d5
commit b4fc4d0b8d
4 changed files with 41 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
library/include/ck/library/utility/host_common_util.hpp
View File

@@ -22,7 +22,7 @@ static inline void dumpBufferToFile(const char* fileName, T* data, size_t dataNu
std::ofstream outFile(fileName, std::ios::binary);
if(outFile)
{
outFile.write(reinterpret_cast<char*>(data), dataNumItems * sizeof(T));
outFile.write(reinterpret_cast<const char*>(data), dataNumItems * sizeof(T));
outFile.close();
std::cout << "Write output to file " << fileName << std::endl;
}

5
library/include/ck/library/utility/host_tensor_generator.hpp
View File

@@ -200,10 +200,11 @@ struct GeneratorTensor_3<ck::bf8_t>
template <typename T>
struct GeneratorTensor_4
{
std::default_random_engine generator;
std::mt19937 generator;
std::normal_distribution<float> distribution;
GeneratorTensor_4(float mean, float stddev) : generator(1), distribution(mean, stddev){};
GeneratorTensor_4(float mean, float stddev, unsigned int seed = 1)
: generator(seed), distribution(mean, stddev){};
template <typename... Is>
T operator()(Is...)