Closed Anndrey24 closed 4 months ago
@tvm-bot rerun
Failed to re-run CI in https://github.com/apache/tvm/actions/runs/9147913719
Resolved the conflict!
Thanks @Anndrey24 and @lhutton1, this is now merged!
Thanks @Anndrey24 @lhutton1 @ekalda .
Seems we have a breakage/flaky likely related to this pr https://ci.tlcpack.ai/blue/organizations/jenkins/tvm-arm/detail/main/1980/pipeline (in lint,arm, and cpu jobs).
I created a temp revert, https://github.com/apache/tvm/pull/17038 to unblock the ci, if there is an alternative fix that would also be good, eitherway we followup with a redo quickly.
This commit adds a scalable
arm_cpu
conv2d NHWC schedule for fp32 which generates SME instructions by using the tensor intrinsics introduced in #16921.Alongside the SME schedule, the logic of the TE schedule
schedule_conv2d_gemm_native()
for both non-scalable and scalable vector implementations has also been translated into the new TIR schedule. This means that the TE compute definitioncompute_conv2d_NHWC_hybrid()
is now compatible with both the original TE schedules (e.g.schedule_conv2d_NHWC_hybrid()
) and the newly introduced TIR scheduleschedule_conv2d_NHWC_hybrid_TIR()
. The corresponding TOPI test has been extended to reflect that.cc @ekalda @lhutton1