Matthew Poremba
Matthew Poremba
Release of MI300X simulation capability: - Implements the required MI300X features over MI200 (currently only architecture flat scratch). - Make the gpu-compute model use MI200 features when MI300X / gfx942...
This adds two failsafes which may cause a panic on some machines. First, check the host machine has the KVM XCR capability before calling getXCRs or setXCRs. Second, ensure the...
**Describe the bug** The AVX / YMM register state is not saved or restored in gem5 with the X86KvmCPU leading to crashes on checkpoint restoration when AVX is enabled in...
There are no GPUFS tests despite being usable for several years. All of the old resources are available so we could do something like the code below. The last release...
Still working on this but stuck on this error: `Error in unproxying param 'memories' of board` Hoping y'all stlib experts can help me out here because I don't see anything...
Move the packages from http/user-data to the install script similar to (https://github.com/gem5/gem5-resources/pull/31). Bump PyTorch to version 2.3.0. This is the first version to support MI300. An MI300 config should be...
Scratch memory requests that are larger than one dword are using a different memory layout than global instructions. Rather than being placed contiguously, each dword is interleaved 64 lanes *...
This PR has 3 commits: - Update scaling methods to scale by multiplication or division when upcasting or downcasting respectively. - Preserve the sign when a microscaling conversion results in...
Allow for multiple RubySystem objects. Not to be confused with multiple Ruby **protocols**. This has been a topic in the past for having one System object with two nodes communicating,...
There are several parts to this PR to work towards #1349 . (1) Make RubySystem::getBlockSizeBytes non-static by providing ways to access the block size or passing the block size explicitly...