Device Agnostic Pipeline #140
Conversation
There was a problem hiding this comment.
What's the purpose of this PR?
Have an example a user can try out on any HW and get decent performance?
We don't expect anyone to really use this for anything right?
The GEMM isn't really device agonistic, is it? It's more that likely any GPU a user would run on has features assumed by the SM_70 pipeline.
If so would it be sufficient to only add the example, but not the new builder and instead directly use the SM_70 builder as a baseline for most current GPUs?
examples/sycl/device_agnostic/device_agnostic_collective_builder.cpp
Outdated
Show resolved
Hide resolved
include/cutlass/gemm/collective/builders/device_agnostic_mma_builder.inl
Outdated
Show resolved
Hide resolved
The SM70 mainloop is device agnostic, it implements a tiled GEMM algorithm, with data being blocked in shared memory and registers. With us passing the UniversalCopy and UniversalMMA, this would become a truly device agnostic gemm.
SM70 does not have a collective builder. Also, I believe the idea is that the API accepts something like a |
…e_agnostic_pipeline
…pipeline # Conflicts: # examples/CMakeLists.txt
aacostadiaz
force-pushed
the
atharva/device_agnostic_pipeline
branch
from
46b102a to
044bcf7
Compare
include/cutlass/epilogue/collective/builders/device_agnostic_builder.inl
Outdated
Show resolved
Hide resolved
examples/sycl/device_agnostic/device_agnostic_collective_builder.cpp
Outdated
Show resolved
Hide resolved
This pipeline serves as an entry point for others in the community to build their own backends and setup their infrastructure. We don’t expect to get performance, just functionality and we are not planning to support more functionalities than what is in this PR. Just basic GEMM with the default epilogue. |
Adds
CollectiveMMAand aCollectiveBuilderAPI for device agnostic pipeline.This piggybacks off of the SM_70 2 stage gemm pipeline, with blocking in SMem and RMem, to get somewhat performant
gemmon any device.