FIX SVE CI latency regression (facebookresearch#3918)

Summary:
Sometimes between Sept 25 to Oct 2, downloading and linking against `openblas=*=*openmp*` package to run tests have caused a 4-7x slow down. Link it with the regular openblas package which is not compiled with `USE_OPENMP=1`. We will set the openblas omp threads via the environment variable `OPENBLAS_NUM_THREADS` according to https://github.com/OpenMathLib/OpenBLAS/wiki/Faq#multi-threaded

Pull Request resolved: facebookresearch#3918

Test Plan: SVE CI should finish within 40 minutes

Reviewed By: ramilbakhshyiev

Differential Revision: D64059860

Pulled By: mengdilin

fbshipit-source-id: 3ba2bda5fce5122f051421f459692f15ad5360a4

Author: mengdilin

.github/actions/build_cmake/action.yml

@@ -36,12 +36,7 @@ runs:
 
         # install base packages for ARM64
         if [ "${{ runner.arch }}" = "ARM64" ]; then
-          # installing libmamba-solver is necessary for openblas=*=*openmp*
-          conda install --solver=classic conda-forge::conda-libmamba-solver conda-forge::libmamba conda-forge::libmambapy conda-forge::libarchive
-
-          # installing openblas that was compiled with USE_OPENMP=1.
-          # Context: https://github.com/facebookresearch/faiss/wiki/Troubleshooting#surprising-faiss-openmp-and-openblas-interaction
-          conda install -y -q -c conda-forge openblas=*=*openmp* gxx_linux-aarch64 sysroot_linux-aarch64
+          conda install -y -q -c conda-forge openblas gxx_linux-aarch64 sysroot_linux-aarch64
         fi
 
         # install base packages for X86_64

.github/workflows/build.yml

@@ -124,6 +124,9 @@ jobs:
         uses: ./.github/actions/build_cmake
         with:
           opt_level: sve
+        env:
+          # Context: https://github.com/facebookresearch/faiss/wiki/Troubleshooting#surprising-faiss-openmp-and-openblas-interaction
+          OPENBLAS_NUM_THREADS: '1'
   linux-x86_64-conda:
     name: Linux x86_64 (conda)
     needs: linux-x86_64-cmake