diff --git a/easybuild/easyconfigs/p/PyTorch/PyTorch-1.12.0-foss-2022a.eb b/easybuild/easyconfigs/p/PyTorch/PyTorch-1.12.0-foss-2022a.eb
new file mode 100644
index 000000000000..acaa8d0b12a9
--- /dev/null
+++ b/easybuild/easyconfigs/p/PyTorch/PyTorch-1.12.0-foss-2022a.eb
@@ -0,0 +1,90 @@
+name = 'PyTorch'
+version = '1.12.0'
+
+homepage = 'https://pytorch.org/'
+description = """Tensors and Dynamic neural networks in Python with strong GPU acceleration.
+PyTorch is a deep learning framework that puts Python first."""
+
+toolchain = {'name': 'foss', 'version': '2022a'}
+
+source_urls = [GITHUB_RELEASE]
+sources = ['%(namelower)s-v%(version)s.tar.gz']
+patches = [
+    '%(name)s-1.7.0_avoid-nan-in-test-torch.patch',
+    '%(name)s-1.7.0_disable-dev-shm-test.patch',
+    '%(name)s-1.9.0_limit-world-size-for-zero-redundancy-opt-test.patch',
+    '%(name)s-1.10.0_fix-test-dataloader-fixed-affinity.patch',
+    '%(name)s-1.10.0_skip_cmake_rpath.patch',
+    '%(name)s-1.11.0_increase-distributed-test-timeout.patch',
+    '%(name)s-1.11.0_increase_c10d_gloo_timeout.patch',
+    '%(name)s-1.11.0_disable_failing_jit_cuda_fuser_tests.patch',
+]
+checksums = [
+    {'pytorch-v1.12.0.tar.gz': '46eff236370b759c427b03ff535c3597099043e8e467b8f81f9cd4b258a7a321'},
+    {'PyTorch-1.7.0_avoid-nan-in-test-torch.patch': 'b899aa94d9e60f11ee75a706563312ccefa9cf432756c470caa8e623991c8f18'},
+    {'PyTorch-1.7.0_disable-dev-shm-test.patch': '622cb1eaeadc06e13128a862d9946bcc1f1edd3d02b259c56a9aecc4d5406b8a'},
+    {'PyTorch-1.9.0_limit-world-size-for-zero-redundancy-opt-test.patch':
+     'ff573660913ce055e24cfd194ce747ba5685091c631cfd443eae2a99d56b57ea'},
+    {'PyTorch-1.10.0_fix-test-dataloader-fixed-affinity.patch':
+     '313dca681f45ce3bc7c4557fdcdcbe0b77216d2c708fa30a2ec0e22c44876707'},
+    {'PyTorch-1.10.0_skip_cmake_rpath.patch': 'ac05943bb205623f91ef140aa00869efc5fe844184bd666bebf5405808610448'},
+    {'PyTorch-1.11.0_increase-distributed-test-timeout.patch':
+     '087ad20163a1291773ae3457569b80523080eb3731e210946459b2333a919f3f'},
+    {'PyTorch-1.11.0_increase_c10d_gloo_timeout.patch':
+     '20cd4a8663f74ab326fdb032b926bf5c7e94d9750c515ab9050927ba00cf1953'},
+    {'PyTorch-1.11.0_disable_failing_jit_cuda_fuser_tests.patch':
+     'e7bfe120a8b3fe2b40dac6839852a5fbab3cb3429fbe44a0fc3a1800adaaee51'},
+]
+
+builddependencies = [
+    ('CMake', '3.23.1'),
+    ('hypothesis', '6.46.7'),
+]
+dependencies = [
+    ('Ninja', '1.10.2'),  # Required for JIT compilation of C++ extensions
+    ('Python', '3.10.4'),
+    ('protobuf', '3.19.4'),
+    ('protobuf-python', '3.19.4'),
+    ('pybind11', '2.9.2'),
+    ('SciPy-bundle', '2022.05'),
+    ('PyYAML', '6.0'),
+    ('MPFR', '4.1.0'),
+    ('GMP', '6.2.1'),
+    ('numactl', '2.0.14'),
+    ('FFmpeg', '4.4.2'),
+    ('Pillow', '9.1.1'),
+    ('expecttest', '0.1.3'),
+]
+
+osdependencies = [('libibverbs-dev', 'libibverbs-devel', 'rdma-core-devel')]
+
+excluded_tests = {
+    '': [
+        # Bad tests: https://github.com/pytorch/pytorch/issues/60260
+        'distributed/elastic/utils/distributed_test',
+        'distributed/elastic/multiprocessing/api_test',
+        # These tests fail on A10s at the very least, they time out forever no matter how long the timeout is.
+        # Possibly related to NCCL 2.8.3: https://docs.nvidia.com/deeplearning/nccl/release-notes/rel_2-8-3.html
+        # 'distributed/test_distributed_fork',
+        'distributed/test_distributed_spawn',
+        # Fails on A10s: https://github.com/pytorch/pytorch/issues/63079
+        'test_optim',
+        # Test from this suite timeout often. The process group backend is deprecated anyway
+        # 'distributed/rpc/test_process_group_agent',
+        # This test fails constently when run as part of the test suite, but succeeds when run interactively
+        'test_model_dump',
+        # These tests appear flaky, possibly related to number of GPUs that are used
+        'distributed/fsdp/test_fsdp_memory',
+        'distributed/fsdp/test_fsdp_overlap',
+    ]
+}
+
+runtest = "cd test && PYTHONUNBUFFERED=1 %(python)s run_test.py --continue-through-error  --verbose %(excluded_tests)s"
+
+# several tests are known to be flaky, and fail in some contexts (like having multiple GPUs available),
+# so we allow up to 400 (out of ~90k) tests to fail before treating the installation to be faulty
+max_failed_tests = 400
+
+tests = ['%(name)s-check-cpp-extension.py']
+
+moduleclass = 'ai'