Merge branch 'main' into yc-0829-move-tracing-before-sanitizer

Author: yingcong-wu

CMakeLists.txt

@@ -4,7 +4,7 @@
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
 cmake_minimum_required(VERSION 3.20.0 FATAL_ERROR)
-project(unified-runtime VERSION 0.10.0)
+project(unified-runtime VERSION 0.11.0)
 
 # Check if unified runtime is built as a standalone project.
 if(CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_SOURCE_DIR OR UR_STANDALONE_BUILD)

include/ur_api.h

@@ -7,7 +7,7 @@
  * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
  *
  * @file ur_api.h
- * @version v0.10-r0
+ * @version v0.11-r0
  *
  */
 #ifndef UR_API_H_INCLUDED
@@ -1133,7 +1133,8 @@ typedef enum ur_api_version_t {
     UR_API_VERSION_0_8 = UR_MAKE_VERSION(0, 8),      ///< version 0.8
     UR_API_VERSION_0_9 = UR_MAKE_VERSION(0, 9),      ///< version 0.9
     UR_API_VERSION_0_10 = UR_MAKE_VERSION(0, 10),    ///< version 0.10
-    UR_API_VERSION_CURRENT = UR_MAKE_VERSION(0, 10), ///< latest known version
+    UR_API_VERSION_0_11 = UR_MAKE_VERSION(0, 11),    ///< version 0.11
+    UR_API_VERSION_CURRENT = UR_MAKE_VERSION(0, 11), ///< latest known version
     /// @cond
     UR_API_VERSION_FORCE_UINT32 = 0x7fffffff
     /// @endcond

include/ur_ddi.h

@@ -7,7 +7,7 @@
  * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
  *
  * @file ur_ddi.h
- * @version v0.10-r0
+ * @version v0.11-r0
  *
  */
 #ifndef UR_DDI_H_INCLUDED

include/ur_print.hpp

@@ -7,7 +7,7 @@
  * SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
  *
  * @file ur_print.hpp
- * @version v0.10-r0
+ * @version v0.11-r0
  *
  */
 #ifndef UR_PRINT_HPP

scripts/Doxyfile

@@ -38,7 +38,7 @@ PROJECT_NAME           = "Intel One API Unified Runtime API"
 # could be handy for archiving the generated documentation or if some version
 # control system is used.
 
-PROJECT_NUMBER         = v0.10
+PROJECT_NUMBER         = v0.11
 
 # Using the PROJECT_BRIEF tag one can provide an optional one line description
 # for a project that appears at the top of each page and should give viewer a

scripts/benchmarks/benches/compute.py

@@ -82,7 +82,7 @@ def run(self, env_vars) -> Result:
 
         result = self.run_bench(command, env_vars)
         (label, mean) = self.parse_output(result)
-        return Result(label=label, value=mean, command=command, env=env_vars, stdout=result)
+        return Result(label=label, value=mean, command=command, env=env_vars, stdout=result, lower_is_better=self.lower_is_better())
 
     def parse_output(self, output):
         csv_file = io.StringIO(output)

scripts/benchmarks/benches/velocity.py

@@ -61,7 +61,7 @@ def run(self, env_vars) -> Result:
 
         result = self.run_bench(command, env_vars)
 
-        return Result(label=self.bench_name, value=self.parse_output(result), command=command, env=env_vars, stdout=result)
+        return Result(label=self.bench_name, value=self.parse_output(result), command=command, env=env_vars, stdout=result, lower_is_better=self.lower_is_better())
 
     def teardown(self):
         return

scripts/benchmarks/main.py

@@ -52,34 +52,46 @@ def main(directory, additional_env_vars, save_name, compare_names, filter):
         benchmarks = [benchmark for benchmark in benchmarks if filter.search(benchmark.name())]
 
     for benchmark in benchmarks:
-        print(f"setting up {benchmark.name()}... ", end='', flush=True)
-        benchmark.setup()
-        print("complete.")
+        try:
+            print(f"setting up {benchmark.name()}... ", end='', flush=True)
+            benchmark.setup()
+            print("complete.")
+        except Exception as e:
+            if options.exit_on_failure:
+                raise e
+            else:
+                print(f"failed: {e}")
 
     results = []
     for benchmark in benchmarks:
-        merged_env_vars = {**additional_env_vars}
-        iteration_results = []
-        for iter in range(options.iterations):
-            print(f"running {benchmark.name()}, iteration {iter}... ", end='', flush=True)
-            bench_results = benchmark.run(merged_env_vars)
-            if bench_results is not None:
-                print(f"complete ({bench_results.value} {benchmark.unit()}).")
-                iteration_results.append(bench_results)
+        try:
+            merged_env_vars = {**additional_env_vars}
+            iteration_results = []
+            for iter in range(options.iterations):
+                print(f"running {benchmark.name()}, iteration {iter}... ", end='', flush=True)
+                bench_results = benchmark.run(merged_env_vars)
+                if bench_results is not None:
+                    print(f"complete ({bench_results.value} {benchmark.unit()}).")
+                    iteration_results.append(bench_results)
+                else:
+                    print(f"did not finish.")
+
+            if len(iteration_results) == 0:
+                continue
+
+            iteration_results.sort(key=lambda res: res.value)
+            median_index = len(iteration_results) // 2
+            median_result = iteration_results[median_index]
+
+            median_result.unit = benchmark.unit()
+            median_result.name = benchmark.name()
+
+            results.append(median_result)
+        except Exception as e:
+            if options.exit_on_failure:
+                raise e
             else:
-                print(f"did not finish.")
-
-        if len(iteration_results) == 0:
-            continue
-
-        iteration_results.sort(key=lambda res: res.value)
-        median_index = len(iteration_results) // 2
-        median_result = iteration_results[median_index]
-
-        median_result.unit = benchmark.unit()
-        median_result.name = benchmark.name()
-
-        results.append(median_result)
+                print(f"failed: {e}")
 
     for benchmark in benchmarks:
         print(f"tearing down {benchmark.name()}... ", end='', flush=True)
@@ -126,6 +138,7 @@ def validate_and_parse_env_args(env_args):
     parser.add_argument("--timeout", type=int, help='Timeout for individual benchmarks in seconds.', default=600)
     parser.add_argument("--filter", type=str, help='Regex pattern to filter benchmarks by name.', default=None)
     parser.add_argument("--verbose", help='Print output of all the commands.', action="store_true")
+    parser.add_argument("--exit_on_failure", help='Exit on first failure.', action="store_true")
 
     args = parser.parse_args()
     additional_env_vars = validate_and_parse_env_args(args.env)
@@ -137,6 +150,7 @@ def validate_and_parse_env_args(env_args):
     options.timeout = args.timeout
     options.ur_dir = args.ur_dir
     options.ur_adapter_name = args.ur_adapter_name
+    options.exit_on_failure = args.exit_on_failure
 
     benchmark_filter = re.compile(args.filter) if args.filter else None
 

scripts/benchmarks/output.py

@@ -116,7 +116,7 @@ def generate_summary_table(chart_data: dict[str, list[Result]]):
             if key in results:
                 value = results[key].value
                 if key == best_key:
-                    row += f" `**{value}**` |"  # Highlight the best value
+                    row += f" <ins>{value}</ins> |"  # Highlight the best value
                 else:
                     row += f" {value} |"
             else:
@@ -132,6 +132,7 @@ def generate_markdown(chart_data: dict[str, list[Result]]):
 
     return f"""
 # Summary
+<ins>result</ins> is better\n
 {summary_table}
 # Charts
 {mermaid_script}

scripts/core/CONTRIB.rst

@@ -127,9 +127,9 @@ Adapter Change Process
 .. _intel/llvm:
    https://github.com/intel/llvm
 .. _UNIFIED_RUNTIME_REPO:
-   https://github.com/intel/llvm/blob/sycl/sycl/plugins/unified_runtime/CMakeLists.txt#L102
+   https://github.com/intel/llvm/blob/sycl/sycl/cmake/modules/FetchUnifiedRuntime.cmake#L119
 .. _UNIFIED_RUNTIME_TAG:
-   https://github.com/intel/llvm/blob/sycl/sycl/plugins/unified_runtime/CMakeLists.txt#L109
+   https://github.com/intel/llvm/blob/sycl/sycl/cmake/modules/FetchUnifiedRuntime.cmake#L126
 
 Build Environment
 =================