Merge from 'sycl' to 'sycl-web' (9 commits)

Author: iclsrc

.github/workflows/sycl-linux-precommit.yml

@@ -44,6 +44,7 @@ jobs:
     uses: ./.github/workflows/sycl-detect-changes.yml
 
   build:
+    name: Self build
     needs: [detect_changes]
     if: always() && success()
     uses: ./.github/workflows/sycl-linux-build.yml

.github/workflows/sycl-post-commit.yml

@@ -34,7 +34,7 @@ jobs:
     uses: ./.github/workflows/sycl-detect-changes.yml
 
   build-lin:
-    name: Linux (Self build + no-assertions)
+    name: Linux (GCC + no-assertions)
     if: github.repository == 'intel/llvm'
     uses: ./.github/workflows/sycl-linux-build.yml
     with:

devops/dependencies-igc-dev.json

@@ -1,10 +1,10 @@
 {
   "linux": {
     "igc_dev": {
-      "github_tag": "igc-dev-e22e2a8",
-      "version": "e22e2a8",
-      "updated_at": "2025-07-31T01:11:27Z",
-      "url": "https://api.github.com/repos/intel/intel-graphics-compiler/actions/artifacts/3654137925/zip",
+      "github_tag": "igc-dev-04d2e53",
+      "version": "04d2e53",
+      "updated_at": "2025-08-04T02:10:56Z",
+      "url": "https://api.github.com/repos/intel/intel-graphics-compiler/actions/artifacts/3677914099/zip",
       "root": "{DEPS_ROOT}/opencl/runtime/linux/oclgpu"
     }
   }

devops/scripts/benchmarks/benches/compute.py

@@ -51,7 +51,7 @@ def git_url(self) -> str:
         return "https://github.com/intel/compute-benchmarks.git"
 
     def git_hash(self) -> str:
-        return "83b9ae3ebb3563552409f3a317cdc1cf3d3ca6bd"
+        return "c9e135d4f26dd6badd83009f92f25d6285fc1e21"
 
     def setup(self) -> None:
         if options.sycl is None:
@@ -73,6 +73,8 @@ def setup(self) -> None:
             f"-DBUILD_SYCL=ON",
             f"-DSYCL_COMPILER_ROOT={options.sycl}",
             f"-DALLOW_WARNINGS=ON",
+            f"-DCMAKE_CXX_COMPILER=clang++",
+            f"-DCMAKE_C_COMPILER=clang",
         ]
 
         if options.ur_adapter == "cuda":
@@ -204,6 +206,17 @@ def benchmarks(self) -> list[Benchmark]:
 
             # Add GraphApiSubmitGraph benchmarks
             for in_order_queue in [0, 1]:
+                benches.append(
+                    GraphApiSubmitGraph(
+                        self,
+                        runtime,
+                        in_order_queue,
+                        self.submit_graph_num_kernels[-1],
+                        0,
+                        useEvents=0,
+                        useHostTasks=1,
+                    )
+                )
                 for num_kernels in self.submit_graph_num_kernels:
                     for measure_completion_time in [0, 1]:
                         for use_events in [0, 1]:
@@ -215,6 +228,7 @@ def benchmarks(self) -> list[Benchmark]:
                                     num_kernels,
                                     measure_completion_time,
                                     use_events,
+                                    useHostTasks=0,
                                 )
                             )
 
@@ -840,22 +854,25 @@ def __init__(
         numKernels,
         measureCompletionTime,
         useEvents,
+        useHostTasks,
     ):
         self.inOrderQueue = inOrderQueue
         self.numKernels = numKernels
         self.measureCompletionTime = measureCompletionTime
         self.useEvents = useEvents
+        self.useHostTasks = useHostTasks
         self.ioq_str = "in order" if self.inOrderQueue else "out of order"
         self.measure_str = (
             " with measure completion" if self.measureCompletionTime else ""
         )
         self.use_events_str = f" with events" if self.useEvents else ""
+        self.host_tasks_str = f" use host tasks" if self.useHostTasks else ""
         super().__init__(
             bench, f"graph_api_benchmark_{runtime.value}", "SubmitGraph", runtime
         )
 
     def explicit_group(self):
-        return f"SubmitGraph {self.ioq_str}{self.measure_str}{self.use_events_str}, {self.numKernels} kernels"
+        return f"SubmitGraph {self.ioq_str}{self.measure_str}{self.use_events_str}{self.host_tasks_str}, {self.numKernels} kernels"
 
     def description(self) -> str:
         return (
@@ -864,10 +881,10 @@ def description(self) -> str:
         )
 
     def name(self):
-        return f"graph_api_benchmark_{self.runtime.value} SubmitGraph{self.use_events_str} numKernels:{self.numKernels} ioq {self.inOrderQueue} measureCompletion {self.measureCompletionTime}"
+        return f"graph_api_benchmark_{self.runtime.value} SubmitGraph{self.use_events_str}{self.host_tasks_str} numKernels:{self.numKernels} ioq {self.inOrderQueue} measureCompletion {self.measureCompletionTime}"
 
     def display_name(self) -> str:
-        return f"{self.runtime.value.upper()} SubmitGraph {self.ioq_str}{self.measure_str}{self.use_events_str}, {self.numKernels} kernels"
+        return f"{self.runtime.value.upper()} SubmitGraph {self.ioq_str}{self.measure_str}{self.use_events_str}{self.host_tasks_str}, {self.numKernels} kernels"
 
     def get_tags(self):
         return [
@@ -888,6 +905,7 @@ def bin_args(self) -> list[str]:
             "--KernelExecutionTime=1",
             f"--UseEvents={self.useEvents}",
             "--UseExplicit=0",
+            f"--UseHostTasks={self.useHostTasks}",
         ]
 
 

devops/scripts/benchmarks/compare.py

@@ -112,13 +112,9 @@ def get_result_paths() -> list[str]:
         def validate_benchmark_result(result: BenchmarkRun) -> bool:
             """
             Returns True if result file:
-            - Was ran on the target machine/hostname specified
-            - Sanity check: ensure metadata are all expected values:
               - Date is truly before cutoff timestamp
               - Name truly matches up with specified result_name
             """
-            if result.hostname != hostname:
-                return False
             if result.name != result_name:
                 log.warning(
                     f"Result file {result_path} does not match specified result name {result.name}."

llvm/docs/requirements-hashed.txt

@@ -118,9 +118,9 @@ commonmark==0.9.1 \
     --hash=sha256:452f9dc859be7f06631ddcb328b6919c67984aca654e5fefb3914d54691aed60 \
     --hash=sha256:da2f38c92590f83de410ba1a3cbceafbc74fee9def35f9251ba9a971d6d66fd9
     # via recommonmark
-docutils==0.22 \
-    --hash=sha256:4ed966a0e96a0477d852f7af31bdcb3adc049fbb35ccba358c2ea8a03287615e \
-    --hash=sha256:ba9d57750e92331ebe7c08a1bbf7a7f8143b86c476acd51528b042216a6aad0f
+docutils==0.21.2 \
+    --hash=sha256:3a6b18732edf182daa3cd12775bbb338cf5691468f91eeeb109deff6ebfa986f \
+    --hash=sha256:dafca5b9e384f0e419294eb4d2ff9fa826435bf15f15b7bd45723e8ad76811b2
     # via
     #   -r requirements.txt
     #   myst-parser

llvm/docs/requirements.txt

@@ -1,5 +1,5 @@
 sphinx==8.1.3
-docutils==0.22
+docutils==0.21.2
 sphinx-markdown-tables==0.0.17
 recommonmark==0.7.1
 sphinx-automodapi==0.20.0

sycl/doc/extensions/experimental/sycl_ext_oneapi_device_image_backend_content.asciidoc

@@ -224,24 +224,24 @@ namespace syclexp = sycl::ext::oneapi::experimental;
 
 SYCL_EXT_ONEAPI_FUNCTION_PROPERTY((syclexp::nd_range_kernel<1>))
 void iota(float start, float *ptr) {
-  size_t id = syclext::this_work_item::get_nd_item().get_global_linear_id();
+  size_t id = syclext::this_work_item::get_nd_item<1>().get_global_linear_id();
   ptr[id] = start + static_cast<float>(id);
 }
 
-void main() {
+int main() {
   sycl::device d;
   sycl::queue q{d};
   sycl::context ctxt = q.get_context();
 
   // Get a kernel bundle that contains the kernel "iota".
-  sycl::kernel_id iota = syclexp::get_kernel_id<iota>();
+  sycl::kernel_id iota_id = syclexp::get_kernel_id<iota>();
   auto exe_bndl =
-    sycl::get_kernel_bundle<sycl::bundle_state::executable>(ctxt, {iota});
+    sycl::get_kernel_bundle<sycl::bundle_state::executable>(ctxt, {iota_id});
 
   std::vector<std::byte> bytes;
-  for (auto& img: bundle) {
+  for (auto& img: exe_bndl) {
     // Search for the device image that contains "iota" for this device.
-    if (img.has_kernel(iota, dev)) {
+    if (img.has_kernel(iota_id, d)) {
       bytes = img.ext_oneapi_get_backend_content();
       break;
     }