sriharikarnam

[Clarification] Provision for AMD HIP backend for MXNet

1

In the process of merging HIP Port of MXNet, we are creating a provision for CUDA and HIP backend for mshadow as well. mshadow being sub-module for MXNet. Please refer...

Runtime issue for thrust::copy API

5

**Background:** Porting Thrust library to ROCm Platform **Error Description**: The unit test case fails when copying stl vector to device vector for mixed data types (int to float) and giving...

**Issue** Accuracy issue related to nan loss is observed on executing the mxnet training application on gpu where as on cpu we get the approriate output. For example: **Training on...

Memory access fault error on ROCm 1.8.151 for MXNet HIP Port

8

**Background:** Porting Mxnet Deep Learning framework to ROCm Platform **Issue :** While executing some applications on HIP/ROCm platform we are facing below mentioned issue.(ROCm 1.8.151) This issue is reproducible in...

HSADevice::CreateKernel(): Unable to create kernel

4

[Sample_Code&logs.zip](https://github.com/RadeonOpenCompute/hcc/files/1651604/Sample_Code.logs.zip) Below error is seen while executing thrust testing testcases on HIP/ROCm path. **HSADevice::CreateKernel(): Unable to create kernel** Issue Description: Unable to create kernel is seen when device kernel is...

Error: 'amdgpu_flat_work_group_size' attribute requires an integer constant

1

**Background:** Porting Thrust to ROCm Platform **Issue:** Compilaton error 'amdgpu_flat_work_group_size' attribute requires an integer constant ../Thrust/thrust/system/cuda/detail/bulk/detail/cuda_launcher/triple_chevron_launcher.hpp:63:1: **error:** 'amdgpu_flat_work_group_size' attribute requires an integer constant bulk_launch_bounds(block_size, 0) [launch_bounds_error_Thrust.txt] (https://github.com/RadeonOpenCompute/hcc/files/1344054/launch_bounds_error_Thrust.txt) ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Here block_size...

[Clarification] Provision for AMD HIP backend for MXNet

1

In the process of merging HIP Port of MXNet, we are creating a provision for CUDA and HIP backend for CUB as well. CUB being sub-module for MXNet. Please refer...

Error : no matching function for call to 'atomicAdd'

9

Background: Porting Mxnet Deep Learning framework to ROCm Platform Issue : While generating 'libmxnet.so' file for HIP/ROCm platform we are facing below mentioned compilation issue. Error: src/operator/./nn/./pool.cuh:339:7: error: no matching...