Merge pull request #20178 from Flamefire/20240321123703_new_pr_PyTorc…

…h1121

fix PyTorch-1.12.1-foss-2022a-CUDA-11.7.0.eb for Linux 6+

easybuild/easyconfigs/p/PyTorch/PyTorch-1.12.1-foss-2022a-CUDA-11.7.0.eb

@@ -27,6 +27,7 @@ patches = [
     'PyTorch-1.11.0_install-vsx-vec-headers.patch',
     'PyTorch-1.11.1_skip-test_init_from_local_shards.patch',
     'PyTorch-1.12.0_fix-EmbeddingBag-without-fbgemm.patch',
+    'PyTorch-1.12.0_fix-Gloo-compatibility-with-newer-kernel.patch',
     'PyTorch-1.12.1_add-hypothesis-suppression.patch',
     'PyTorch-1.12.1_fix-autograd-thread_shutdown-test.patch',
     'PyTorch-1.12.1_fix-cuda-gcc-version-check.patch',
@@ -47,6 +48,7 @@ patches = [
     'PyTorch-1.12.1_skip-ao-sparsity-test-without-fbgemm.patch',
     'PyTorch-1.12.1_skip-failing-grad-test.patch',
     'PyTorch-1.12.1_skip-test_round_robin.patch',
+    'PyTorch-2.0.1_skip-test_baddbmm_cpu_bfloat16.patch',
 ]
 checksums = [
     '031c71073db73da732b5d01710220564ce6dd88d812ba053f0cc94296401eccb',  # pytorch-v1.12.1.tar.gz
@@ -75,6 +77,8 @@ checksums = [
     '4aeb1b0bc863d4801b0095cbce69f8794066748f0df27c6aaaf729c5ecba04b7',
     {'PyTorch-1.12.0_fix-EmbeddingBag-without-fbgemm.patch':
      '090598592283e3fc46ee08a68b6a6afe07be41b26514afba51834408bf1c98ed'},
+    {'PyTorch-1.12.0_fix-Gloo-compatibility-with-newer-kernel.patch':
+     '0d19c3dd7ea088aae2f18a2552420293a06ed0a0e1e8e2a28d44ff3847de32af'},
     # PyTorch-1.12.1_add-hypothesis-suppression.patch
     'e71ffb94ebe69f580fa70e0de84017058325fdff944866d6bd03463626edc32c',
     # PyTorch-1.12.1_fix-autograd-thread_shutdown-test.patch
@@ -108,6 +112,8 @@ checksums = [
     '1c89e7e67287fe6b9a95480a4178d3653b94d0ab2fe68edf227606c8ae548fdc',  # PyTorch-1.12.1_skip-failing-grad-test.patch
     # PyTorch-1.12.1_skip-test_round_robin.patch
     '63d4849b78605aa088fdff695637d9473ea60dee603a3ff7f788690d70c55349',
+    {'PyTorch-2.0.1_skip-test_baddbmm_cpu_bfloat16.patch':
+     '199005bbbb913837e557358dee31535d8e3f63af9ac7cdcece624ab8e572e28a'},
 ]
 
 osdependencies = [OS_PKG_IBVERBS_DEV]
@@ -164,6 +170,10 @@ excluded_tests = {
     ]
 }
 
+# several tests are known to be flaky, and fail in some contexts (like having multiple GPUs available),
+# so we allow some tests to fail before treating the installation to be faulty
+max_failed_tests = 4
+
 runtest = 'cd test && PYTHONUNBUFFERED=1 %(python)s run_test.py --continue-through-error  --verbose %(excluded_tests)s'
 
 # The readelf sanity check command can be taken out once the TestRPATH test from