[intel] update skiplist and use xfail mark after '65ad3b7'

Signed-off-by: Anatoly Myachev <[email protected]>

Author: anmyachev

python/test/unit/language/test_core.py

@@ -6016,7 +6016,7 @@ def compute_scratch_buffer_shape(src_layout, dst_layout, shape):
 @pytest.mark.parametrize("dst_layout", filter_layouts(layouts))
 def test_convert2d(M, N, src_layout, interm_layout, dst_layout, dtype, device, tmp_path: pathlib.Path):
     if str(src_layout) == str(dst_layout):
-        pytest.skip()
+        pytest.xfail("Do not convert same layout")
     if is_hip() or is_xpu():
         try:
             scratch_shape = compute_scratch_buffer_shape(src_layout, dst_layout, (M, N))

python/test/unit/language/test_matmul.py

@@ -120,7 +120,7 @@ def test_simple_matmul(dtype_src_str, dtype_dst_str, BLOCK_M, BLOCK_N, BLOCK_K,
     if EPILOGUE_SUBTILE and not is_xpu() and (is_hip() or NUM_CTAS > 1 or BLOCK_N >= 512):
         pytest.skip("creates convert layout too big to fit in smem")
     if LAYOUT_16x256 and (not is_cuda() or torch.cuda.get_device_capability()[0] < 10):
-        pytest.skip("skip forcing tmem layout on non blackwell targets.")
+        pytest.xfail("skip forcing tmem layout on non blackwell targets.")
     M, N, K = 1024, 512, 256
     torch.manual_seed(42)
     precision = "tf32" if dtype_src_str == "tensorfloat32" else "ieee"