[ET-VK] Implement SDPA with fused ops

.github/workflows/pull.yml

@@ -971,6 +971,13 @@ jobs:
         ./cmake-out/backends/vulkan/test/custom_ops/q4gsw_linear
         ./cmake-out/backends/vulkan/test/custom_ops/choose_qparams_per_row
 
+        # "Classic" Operator tests
+        PYTHON_EXECUTABLE=python bash backends/vulkan/test/scripts/test_op.sh --build
+        # TODO(ssjia): figure out how to run custom op tests in CI. Currently, they are
+        # failing due to to the libstdc++.so.6 installed with conda not supporting
+        # GLIBCXX_3.4.30. These tests are still run in Meta internal CI.
+        # ./cmake-out/backends/vulkan/test/op_tests/vulkan_sdpa_test
+
         # Run e2e testing for selected operators. More operators will be tested via this
         # route in the future.
         python -m unittest backends/vulkan/test/test_vulkan_delegate.py -k "*pt2e*"

backends/vulkan/op_registry.py

@@ -571,7 +571,7 @@ def register_sdpa_with_kv_cache_op():
 )
 def register_sdpa_ops():
     return OpFeatures(
-        inputs_storage=utils.WIDTH_PACKED_TEXTURE,
+        inputs_storage=utils.CONTIGUOUS_ANY,
         supports_resize=True,
     )