[Fix XPU CI] [Inductor UT] Fix test cases broken by community. (#165714)

Fixes #165719, Fixes #165771

Pull Request resolved: https://github.com/pytorch/pytorch/pull/165714
Approved by: https://github.com/jansel

test/inductor/test_mem_estimation.py

@@ -13,6 +13,7 @@ from torch._inductor.fx_passes.memory_estimator import (
 from torch._inductor.test_case import run_tests, TestCase as InductorTestCase
 from torch._subclasses.fake_tensor import FakeTensorMode
 from torch.fx.experimental.proxy_tensor import make_fx
+from torch.testing._internal.inductor_utils import GPU_TYPE, HAS_GPU
 from torch.utils._python_dispatch import TorchDispatchMode
 from torch.utils._pytree import tree_map_only
 from torch.utils.weak import WeakIdKeyDictionary
@@ -23,7 +24,7 @@ def tensor_storage_id(tensor):
 
 
 def device_filter(device):
-    return device.type == "cuda"
+    return device.type == GPU_TYPE
 
 
 class FakeTensorMemoryProfilerMode(TorchDispatchMode):
@@ -83,10 +84,10 @@ class TestMemoryProfilingResNet(InductorTestCase):
 
         def create_inputs_and_weights():
             """Create inputs and weights on CUDA."""
-            x = torch.randn(32, 1000, device="cuda")
-            w1 = torch.randn(500, 1000, device="cuda")
-            w2 = torch.randn(100, 500, device="cuda")
-            w3 = torch.randn(10, 100, device="cuda")
+            x = torch.randn(32, 1000, device=GPU_TYPE)
+            w1 = torch.randn(500, 1000, device=GPU_TYPE)
+            w2 = torch.randn(100, 500, device=GPU_TYPE)
+            w3 = torch.randn(10, 100, device=GPU_TYPE)
             return x, w1, w2, w3
 
         def fn(x, w1, w2, w3):
@@ -128,10 +129,10 @@ class TestMemoryProfilingResNet(InductorTestCase):
 
         def create_inputs_and_weights():
             """Create inputs and weights on CUDA."""
-            x = torch.randn(8, 3, 224, 224, device="cuda")
-            conv1_weight = torch.randn(64, 3, 3, 3, device="cuda")
-            conv2_weight = torch.randn(128, 64, 3, 3, device="cuda")
-            linear_weight = torch.randn(10, 128 * 56 * 56, device="cuda")
+            x = torch.randn(8, 3, 224, 224, device=GPU_TYPE)
+            conv1_weight = torch.randn(64, 3, 3, 3, device=GPU_TYPE)
+            conv2_weight = torch.randn(128, 64, 3, 3, device=GPU_TYPE)
+            linear_weight = torch.randn(10, 128 * 56 * 56, device=GPU_TYPE)
             return x, conv1_weight, conv2_weight, linear_weight
 
         def fn(x, conv1_weight, conv2_weight, linear_weight):
@@ -175,9 +176,9 @@ class TestMemoryTracker(InductorTestCase):
 
         def create_inputs_and_weights():
             """Create inputs and weights on CUDA."""
-            x = torch.randn(32, 100, device="cuda")
-            w1 = torch.randn(100, 50, device="cuda")
-            w2 = torch.randn(50, 10, device="cuda")
+            x = torch.randn(32, 100, device=GPU_TYPE)
+            w1 = torch.randn(100, 50, device=GPU_TYPE)
+            w2 = torch.randn(50, 10, device=GPU_TYPE)
             return x, w1, w2
 
         def fn(x, w1, w2):
@@ -240,7 +241,7 @@ class TestMemoryTracker(InductorTestCase):
 
         with FakeTensorMode():
             # Create input
-            primals_1 = torch.randn(1000, 1000, device="cuda")
+            primals_1 = torch.randn(1000, 1000, device=GPU_TYPE)
 
             # Trace the function
             fx_graph = make_fx(foo)(primals_1)
@@ -340,4 +341,5 @@ class TestMemoryTracker(InductorTestCase):
 
 
 if __name__ == "__main__":
-    run_tests(needs="filelock")
+    if HAS_GPU:
+        run_tests(needs="filelock")

test/inductor/test_native_matmul.py

@@ -152,5 +152,6 @@ if HAS_GPU:
     torch.set_default_device(GPU_TYPE)
 
 if __name__ == "__main__":
-    if HAS_GPU:
+    # TODO: support native matmul on xpu
+    if HAS_GPU and GPU_TYPE != "xpu":
         run_tests()

test/inductor/test_torchinductor_opinfo.py

@@ -286,7 +286,14 @@ inductor_expected_failures_single_sample["xpu"] = {
     "tan": {f16},
     "torch.ops.aten._flash_attention_forward": {f16},
     "torch.ops.aten._efficient_attention_forward": {f16, f32},
-    "to_sparse": {f32, f64},
+    "to_sparse": {
+        b8,
+        f16,
+        f32,
+        f64,
+        i32,
+        i64,
+    },  # align with cuda.
     "linalg.eig": {f32, f64},
     ("linalg.pinv", "singular"): {f64},
     # could not create a primitive