update siplists and code after rebase

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

Author: anmyachev

python/triton/language/target_info.py

@@ -22,12 +22,6 @@ def is_cuda():
     return target is not None and target.backend == "cuda"
 
 
-@constexpr_function
-def is_xpu():
-    target = current_target()
-    return target is not None and target.backend == "xpu"
-
-
 @constexpr_function
 def cuda_capability_geq(major, minor=0):
     """

python/triton_kernels/tests/test_matmul.py

@@ -297,8 +297,6 @@ def test_op(m, n, k, split_k, do_gather, do_scatter, fused_scatter, has_y_gammas
     elif is_xpu():
         if split_k > 1:
             pytest.skip("splitK hasn't been fully tested on INTEL GPU.")
-        if "float8_e4m3fn" in act_dtype_str and "float8_e4m3fn" in weight_dtype_str:
-            pytest.skip("FIXME")
 
     if "float8_e4m3fnuz" in (weight_dtype_str, act_dtype_str) and not is_hip_cdna3():
         pytest.skip("float8_e4m3fnuz only tested on AMD CDNA3 Platform")

python/triton_kernels/triton_kernels/matmul_ogs.py

@@ -588,7 +588,7 @@ def matmul_ogs_torch(x, w, bias,
         else:
             idx = gather_indx.src_indx[lo:hi] // n_expts_act
         batch = i if is_input_batched else 0
-        out = torch.matmul(round_x(x[batch, idx, :], torch.arange(lo, hi, device="xpu")).float(),
+        out = torch.matmul(round_x(x[batch, idx, :], torch.arange(lo, hi, device=device)).float(),
                            w[i].float())
         if bias is not None:
             out += bias[i, :] if betas is None else bias[i, :] * betas[lo:hi, None]

python/triton_kernels/triton_kernels/target_info.py

@@ -6,7 +6,6 @@
     cuda_capability_geq,
     is_cuda,
     is_hip,
-    is_xpu,
     is_hip_cdna3,
     is_hip_cdna4,
     is_xpu,

scripts/skiplist/default/triton_kernels.txt

@@ -1,3 +1,11 @@
-tests/test_matmul.py::test_op
+# https://github.com/intel/intel-xpu-backend-for-triton/issues/5074
+tests/test_matmul.py::test_op[False-False-False-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-False-False-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-False-True-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-False-True-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-False-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-False-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-True-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-True-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
 tests/test_matmul.py::test_fused_act
 tests/test_matmul.py::test_zero_reduction_dim

scripts/skiplist/xe2/triton_kernels.txt

@@ -1,3 +1,11 @@
-tests/test_matmul.py::test_op
+# https://github.com/intel/intel-xpu-backend-for-triton/issues/5074
+tests/test_matmul.py::test_op[False-False-False-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-False-False-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-False-True-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-False-True-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-False-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-False-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-True-True-False-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
+tests/test_matmul.py::test_op[False-True-True-True-False-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-1-False-None]
 tests/test_matmul.py::test_fused_act
 tests/test_matmul.py::test_zero_reduction_dim