HSAFoundation / HSA-Drivers-Linux-AMD

These drivers have been superseded by ROCm Platform now hosted at Radeon Open Compute GitHub Repo
https://github.com/RadeonOpenCompute
Other
61 stars 15 forks source link

Querying the agent name failed. #20

Open Mindwalk3r opened 8 years ago

Mindwalk3r commented 8 years ago

Hi, I have a problem when running vector_copy. When I run the program, I get the following output:

Initializing the hsa runtime succeeded. Checking finalizer 1.0 extension support succeeded. Generating function table for finalizer succeeded. Getting a gpu agent succeeded. Querying the agent name failed.

Running kdf_check_installation yields:

Carrizo detected:.....................................Yes Carrizo type supported:.............................Yes amdgpu module is loaded:........................Yes amdkfd module is loaded:..........................Yes AMD IOMMU V2 module is loaded:............Yes KDF device exists:.....................................Yes KFD device has correct permissions:........Yes Valid GPU ID is detected:..........................Yes

Can run HSA:..........................................YES

I have followed the installation process from http://gpuopen.com/getting-started-with-boltzmann-components-platforms-installation/ Any help or ideas to fix this issue would be greatly appreciated.

Thanks!

matcheydj commented 8 years ago

kfd_check_installation skips that part. Valid GPU ID is detected, then it goes right to Can run HSA instead of checking around for what would be the target of hsa_agent_get_info(agent, HSA_AGENT_INFO_NAME, name)

People seem to be hardcoding it? Pretty sure you can just comment out of vector_copy.c:

char name[64] = { 0 }; err = hsa_agent_get_info(agent, HSA_AGENT_INFO_NAME, name); check(Querying the agent name, err); printf("The agent name is %s.\n", name);

Mindwalk3r commented 8 years ago

Thanks for the reply. Unfortunately, the change (commented the lines) caused a segmentation fault instead.

I might also mention that using HCC and testing a sample program (can be found here: http://gpuopen.com/a-brief-intro-to-boltzmann-hcc/) the following error occurred:

There is no device can be used to do the computation

Not sure if this is related, but might be worth mentioning.