Skip to content

Running some CI tests with xlarge resource_class VMs for testing AVX512 support #59166

@imaginary-person

Description

@imaginary-person

🚀 Feature

Currently, the Linux CI checks that are tests are run on large resource_class executors, which are run on machines that don't support AVX512. To test ATen AVX512 kernels, xlarge executors would be required, as these VMs are run on Intel Cascade Lake machines that support AVX512.

Pitch

Some CI checks should be run on xlarge instances, which have AVX512 support. The number of CI tests running on xlarge instances can probably be limited in order to not vastly increase CI costs.

cc @ezyang @seemethere @malfet @walterddr @pytorch/pytorch-dev-infra @ngimel @VitalyFedyunin

Metadata

Metadata

Assignees

No one assigned

    Labels

    module: ciRelated to continuous integrationmodule: vectorizationRelated to SIMD vectorization, e.g., Vec256triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions