Cherry-picks for release branch 3.6.x

.github/README.md

@@ -20,7 +20,7 @@ Intel® XPU Backend for Triton\* is a out of tree backend module for [Triton](ht
   * Latest [Long Term Support (LTS) Release](https://dgpu-docs.intel.com/driver/installation.html)
   * Latest [The Kobuk team Intel® Graphics PPA](https://launchpad.net/~kobuk-team/+archive/ubuntu/intel-graphics)
 * Toolchain:
-  * [Intel® Deep Learning Essentials 2025.2.1](https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit-download.html?packages=dl-essentials&dl-lin=offline&dl-essentials-os=linux)
+  * [Intel® Deep Learning Essentials 2025.3.1](https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit-download.html?packages=dl-essentials&dl-lin=offline&dl-essentials-os=linux)
 
 Note that Intel® XPU Backend for Triton\* is not compatible with Intel® Extension for PyTorch\* and Intel® oneAPI Base Toolkit\*.
 
@@ -37,7 +37,7 @@ You can check existing installation by running one of the [tutorials](/python/tu
 ## Prerequisites
 
 1. Latest [Rolling Release](https://dgpu-docs.intel.com/driver/installation-rolling.html) or [Long Term Support Release](https://dgpu-docs.intel.com/driver/installation.html) of GPU driver
-2. [Intel® Deep Learning Essentials 2025.2.1](https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit-download.html?packages=dl-essentials&dl-lin=offline&dl-essentials-os=linux)
+2. [Intel® Deep Learning Essentials 2025.3.1](https://www.intel.com/content/www/us/en/developer/tools/oneapi/base-toolkit-download.html?packages=dl-essentials&dl-lin=offline&dl-essentials-os=linux)
 
 ## Install PyTorch and Triton from nightly wheels
 

.github/pins/e2e_reference_torch-xpu-ops.txt

@@ -1 +1 @@
-9aac5a1ddf50d75f929d572df51bb368b32da14e
+1e69f40b3c03492eb3dd7e03462a5566f29674d3

.github/pins/pytorch.txt

@@ -1 +1 @@
-d33d125c9413c5043aa5f74fad909a576288242d
+01f94d4096060597e2815efe385255ac19c9c787

.github/workflows/spirvrunner-test.yml

@@ -20,7 +20,7 @@ jobs:
     name: Tests
     runs-on:
       - rolling
-      - dle-2025.2
+      - runner-0.0.22
     timeout-minutes: 75 # equal to max + 3*std over the last 600 successful runs
     steps:
       - name: Checkout repository

scripts/skiplist/default/triton_kernels.txt

@@ -1,62 +1,17 @@
 # https://github.com/intel/intel-xpu-backend-for-triton/issues/5074
-tests/test_matmul.py::test_op[False-False-False-False-False-None-128-1000-700-700-ragged-float16-float16-8-2-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-False-False-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-False-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-False-False-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-False-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-True-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-False-True-True-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-False-True-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-128-1000-700-700-ragged-float16-float16-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-True-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-False-True-True-True-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-False-False-None-128-1000-700-700-ragged-float16-float16-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-True-False-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-False-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-True-False-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-False-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-True-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-False-True-True-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-False-False-None-128-1000-700-700-ragged-float16-float16-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-True-True-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-False-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-False-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-128-1000-700-700-ragged-float16-float16-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-False]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-128-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-False-None-16-1000-704-800-ragged-mxfloat8_e4m3fn-mxfloat4_e2m1-8-2-9-True-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-True-None-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
-tests/test_matmul.py::test_op[False-True-True-True-True-None-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-False-True-None0-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-False-True-None0-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-False-True-None1-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-False-True-None1-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-True-True-None0-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-True-True-None0-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-True-True-None1-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-False-True-True-None1-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-False-True-None0-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-False-True-None0-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-False-True-None1-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-False-True-None1-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-True-True-None0-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-True-True-None0-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-True-True-None1-128-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]
+tests/test_matmul.py::test_op[False-True-True-True-None1-16-1000-400-400-ragged-float8_e4m3fn-float8_e4m3fn-3-1-1-False-None-False-False-False-True]

scripts/test-triton.sh

@@ -45,8 +45,6 @@ OPTION:
     --skip-list SKIPLIST
     --extra-skip-list-suffixes SEMICOLON-SEPARATED LIST OF SUFFIXES
     --select-from-file SELECTFILE
-    --test-expr EXPRESSION
-    --debug-fail
 "
 
 err() {
@@ -88,8 +86,6 @@ TRITON_TEST_IGNORE_ERRORS=false
 SKIP_PIP=false
 SKIP_PYTORCH=false
 TEST_UNSKIP=false
-TEST_FILTER_EXPRESSION=""
-TEST_DEBUG_FAIL=false
 
 while (( $# != 0 )); do
   case "$1" in
@@ -277,14 +273,6 @@ while (( $# != 0 )); do
       TRITON_TEST_SELECTFILE="$(realpath "$2")"
       shift 2
       ;;
-    --test-expr)
-      TEST_FILTER_EXPRESSION="$2"
-      shift 2
-      ;;
-    --debug-fail)
-      TEST_DEBUG_FAIL=true
-      shift
-      ;;
     --help)
       echo "$HELP"
       exit 0
@@ -354,44 +342,12 @@ run_unit_tests() {
 }
 
 run_pytest_command() {
-  local pytest_args=()
-  local pytest_expr=""
-
-  # Parse args to separate -k expression
-  local args=("$@")
-  for ((i=0; i<${#args[@]}; i++)); do
-    if [[ "${args[i]}" == "-k" ]]; then
-      pytest_expr="${args[i+1]}"
-      i=$((i + 1))
-      continue
-    fi
-    pytest_args+=("${args[i]}")
-  done
-
-  # Combine with TEST_FILTER_EXPRESSION
-  if [[ -n "$TEST_FILTER_EXPRESSION" ]]; then
-    if [[ -n "$pytest_expr" ]]; then
-      pytest_expr="($pytest_expr) and ($TEST_FILTER_EXPRESSION)"
-    else
-      pytest_expr="$TEST_FILTER_EXPRESSION"
-    fi
-  fi
-
-  # Apply -k expression if any
-  if [[ -n "$pytest_expr" ]]; then
-    pytest_args+=("-k" "$pytest_expr")
-  fi
-
-  if [[ "$TEST_DEBUG_FAIL" == true ]]; then
-    pytest_args+=("--pdb")
-  fi
-
-  if [[ -n "$TRITON_TEST_SELECTFILE" ]] || [[ -n "$pytest_expr" ]]; then
-    if pytest "${pytest_args[@]}" --collect-only > /dev/null 2>&1; then
-      pytest "${pytest_args[@]}"
+  if [[ -n "$TRITON_TEST_SELECTFILE" ]]; then
+    if pytest "$@" --collect-only > /dev/null 2>&1; then
+      pytest "$@"
     fi
   else
-    pytest "${pytest_args[@]}"
+    pytest "$@"
   fi
 }