Merge branch '2.x' into 2.x-dev

[ROCm/rocprofiler-compute commit: bd803a0883]
2023-12-01 14:50:08 -06:00
parent c134a12674 311014ddbd
commit 0f8d5fe131
6 changed files with 73 additions and 66 deletions
@@ -1,7 +1,7 @@
 name: mi100

 on:
-  pull_request:
+  push:
    branches:
      - '2.x'

@@ -17,33 +17,38 @@ jobs:
  build:
    runs-on: [mi100, rhel9]
    env:
-       TEST_DIR: ${RUNNER_TEMP}/omniperf
-       PYTHONPATH: /share/sw/omniperf/python-libs
+       PYTHONPATH: ${{ github.workspace }}/python-libs
    steps:
      - name: Checkout
        uses: actions/checkout@v3
        with:
          submodules: recursive
          token: ${{ secrets.GH_PAT }}
+      - name: pythonpath
+        run: echo ${PYTHONPATH}
+      - name: Install Python collateral (build and test)
+        run: |
+           pip3 install -t ${PYTHONPATH} -r requirements.txt
+           pip3 install -t ${PYTHONPATH} -r requirements-test.txt
      - name: Configure
        run: |
          mkdir build
          cd build
          ml cmake
          cmake -DENABLE_COVERAGE=ON ..
-      - name: Install Python testing collateral
-        run: pip3 install pytest pytest-cov mock
+      - name: Create HIP binary (vcopy)
+        run: hipcc -o tests/vcopy ./sample/vcopy.cpp
      - name: Run [profile] mode
        run: |
          cd build
          ml cmake
-          ctest --verbose -R test_profiler_general
+          srun -N 1 -J omniperf -p ci -t 00:10:00 ctest --verbose -R test_profiler_general
      - name: Run [analyze] mode
        if: '!cancelled()'
        run: |
          cd build
          ml cmake
-          srun -N 1 -J omniperf -p ci -t 00:25:00 ctest --verbose -R test_analyze_commands
+          srun -N 1 -J omniperf -p ci -t 00:55:00 ctest --verbose -R test_analyze_commands
      - name: Pytest coverage comment
        if: '!cancelled()'
        uses: MishaKav/pytest-coverage-comment@v1.1.48
@@ -72,27 +72,19 @@ if(DEFINED PYTHON_DEPS)
 endif()

 # Check required Python packages
-set(pythonDeps
-    "astunparse"
-    "colorlover"
-    "dash"
-    "matplotlib"
-    "numpy"
-    "pandas"
-    "pymongo"
-    "yaml"
-    "tabulate"
-    "tqdm"
-    "dash_svg"
-    "dash_bootstrap_components"
-    "kaleido")
+file(STRINGS "${CMAKE_CURRENT_SOURCE_DIR}/requirements.txt" pythonDeps)

 message(STATUS "Checking for required Python package dependencies...")
 set_property(GLOBAL PROPERTY pythonDepsFlag "groovy")

 function(checkPythonPackage [package])
+    # mapping for non-default package names
+    set(PACKAGE ${ARGV0})
+    if(${ARGV0} STREQUAL "pyyaml")
+        set(PACKAGE "yaml")
+    endif()
    execute_process(
-        COMMAND ${Python3_EXECUTABLE} -c "import ${ARGV0}"
+        COMMAND ${Python3_EXECUTABLE} -c "import ${PACKAGE}"
        OUTPUT_QUIET ERROR_QUIET
        RESULT_VARIABLE EXIT_CODE)
    if(${EXIT_CODE} EQUAL 0)
@@ -105,6 +97,9 @@ endfunction()

 list(APPEND CMAKE_MESSAGE_INDENT "  ")
 foreach(package IN LISTS pythonDeps)
+    # Filter out any version requirements from requirements.txt
+    string(REGEX REPLACE "[><=].*" "" package "${package}")
+    string(REPLACE "-" "_" package "${package}")
    checkpythonpackage(${package})
 endforeach()
 list(POP_BACK CMAKE_MESSAGE_INDENT)
@@ -1,5 +1,6 @@
 [![Ubuntu 20.04](https://github.com/AMDResearch/omniperf/actions/workflows/ubuntu-focal.yml/badge.svg)](https://github.com/AMDResearch/omniperf/actions/workflows/ubuntu-focal.yml)
 [![RHEL 8](https://github.com/AMDResearch/omniperf/actions/workflows/opensuse.yml/badge.svg)](https://github.com/AMDResearch/omniperf/actions/workflows/opensuse.yml)
+[![MI100](https://github.com/AMDResearch/omniperf/actions/workflows/mi100.yml/badge.svg?branch=2.x)](https://github.com/AMDResearch/omniperf/actions/workflows/mi100.yml)
 [![Docs](https://github.com/AMDResearch/omniperf/actions/workflows/docs.yml/badge.svg)](https://amdresearch.github.io/omniperf/)
 [![DOI](https://zenodo.org/badge/561919887.svg)](https://zenodo.org/badge/latestdoi/561919887)

@@ -18,7 +18,7 @@
 		"chai": "~4.2.0",
 		"gulp": "^4.0.2",
 		"gulp-babel": "^7.0.1",
-		"jsdom": "~16.2.2",
+		"jsdom": ">=16.5.0",
 		"prunk": "^1.3.1",
 		"q": "^1.5.1"
 	},
@@ -0,0 +1,3 @@
+mock
+pytest
+pytest-cov
@@ -11,13 +11,15 @@ import inspect

 omniperf = SourceFileLoader("omniperf", "src/omniperf").load_module()
 workload_1 = os.path.realpath("workload")
-# kernel_name_1 = "vecCopy(double*, double*, double*, int, int) [clone .kd]"
-kernel_name_1 = "void benchmark_func<__half2, 256, 8u, 11u>(__half2, __half2*) "
+kernel_name_1 = "vecCopy(double*, double*, double*, int, int) [clone .kd]"
+## kernel_name_1 = "void benchmark_func<__half2, 256, 8u, 11u>(__half2, __half2*) "

 # change to directory where app is at
 # app_1 = ["./sample/vcopy", "1048576", "256"]
-app_1 = ["./mixbench/build_mi100/mixbench-hip"]
-
+# app_1 = ["./mixbench/build_mi100/mixbench-hip"]
+app_1 = ["./tests/vcopy", "1048576", "256"]
+num_kernels = 1
+dispatch_id = 0

 ALL_CSVS = [
    "SQ_IFETCH_LEVEL.csv",
@@ -254,8 +256,9 @@ def test_path():
            file_dict[file] = pd.read_csv(workload_1 + "/" + file, index_col=0)
            print("length is: ", len(file_dict[file].index))
            print(file_dict[file])
+            # TODO: verify contents: we know function evaluated
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    print(sorted(list(file_dict.keys())))
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
@@ -300,7 +303,7 @@ def test_kernel():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -344,7 +347,7 @@ def test_kernel_summaries():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -388,7 +391,7 @@ def test_ipblocks_SQ():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    print(sorted(list(file_dict.keys())))
    expected_csvs = [
        "SQ_IFETCH_LEVEL.csv",
@@ -474,7 +477,7 @@ def test_ipblocks_SQC():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    print(sorted(list(file_dict.keys())))
    expected_csvs = [
        "pmc_perf.csv",
@@ -527,7 +530,7 @@ def test_ipblocks_TA():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))

@@ -586,7 +589,7 @@ def test_ipblocks_TD():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))

@@ -649,7 +652,7 @@ def test_ipblocks_TCP():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    print(sorted(list(file_dict.keys())))

    expected_csvs = [
@@ -709,7 +712,7 @@ def test_ipblocks_TCC():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))
    expected_csvs = [
@@ -770,7 +773,7 @@ def test_ipblocks_SPI():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    print(sorted(list(file_dict.keys())))

    expected_csvs = [
@@ -829,7 +832,7 @@ def test_ipblocks_CPC():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    print(sorted(list(file_dict.keys())))
    expected_csvs = [
        "pmc_perf.csv",
@@ -883,7 +886,7 @@ def test_ipblocks_CPF():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    print(sorted(list(file_dict.keys())))
    expected_csvs = [
        "pmc_perf.csv",
@@ -936,7 +939,7 @@ def test_ipblocks_SQ_CPC():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))
    expected_csvs = [
@@ -1024,7 +1027,7 @@ def test_ipblocks_SQ_TA():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))
    expected_csvs = [
@@ -1111,7 +1114,7 @@ def test_ipblocks_SQ_SPI():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))
    expected_csvs = [
@@ -1200,7 +1203,7 @@ def test_ipblocks_SQ_SQC_TCP_CPC():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))
    expected_csvs = [
@@ -1291,7 +1294,7 @@ def test_ipblocks_SQ_SPI_TA_TCC_CPF():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels

    print(sorted(list(file_dict.keys())))
    expected_csvs = [
@@ -1423,7 +1426,7 @@ def test_dispatch_0_1():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file and not "roofline" in file:
-                assert len(file_dict[file].index) == 2
+                assert len(file_dict[file].index) == num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1449,7 +1452,7 @@ def test_dispatch_2():
                "--path",
                workload_1,
                "--dispatch",
-                "2",
+                dispatch_id,
                "--",
            ]
            + app_1,
@@ -1511,7 +1514,7 @@ def test_kernel_verbose_0():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1555,7 +1558,7 @@ def test_kernel_verbose_1():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1599,7 +1602,7 @@ def test_kernel_verbose_2():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1643,7 +1646,7 @@ def test_kernel_verbose_3():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1687,7 +1690,7 @@ def test_kernel_verbose_4():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1731,7 +1734,7 @@ def test_kernel_verbose_5():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1775,7 +1778,7 @@ def test_join_type_grid():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1819,7 +1822,7 @@ def test_join_type_kernel():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1866,7 +1869,7 @@ def test_device_0():
                if "roofline" in file:
                    assert len(file_dict[file].index)
                else:
-                    assert len(file_dict[file].index) > 3
+                    assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == ALL_CSVS_MI200
@@ -1909,7 +1912,7 @@ def test_no_roof():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
    if soc == "mi200":
        print(sorted(list(file_dict.keys())))
        assert sorted(list(file_dict.keys())) == [
@@ -1986,7 +1989,7 @@ def test_sort_dispatches():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2038,7 +2041,7 @@ def test_sort_kernels():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2091,7 +2094,7 @@ def test_mem_levels_HBM():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2144,7 +2147,7 @@ def test_mem_levels_L2():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2196,7 +2199,7 @@ def test_mem_levels_vL1D():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2248,7 +2251,7 @@ def test_mem_levels_LDS():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2301,7 +2304,7 @@ def test_mem_levels_HBM_LDS():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2354,7 +2357,7 @@ def test_mem_levels_vL1D_LDS():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2407,7 +2410,7 @@ def test_mem_levels_L2_vL1D_LDS():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":
@@ -2458,7 +2461,7 @@ def test_kernel_names():
        if file.endswith(".csv"):
            file_dict[file] = pd.read_csv(workload_1 + "/" + file)
            if not "sysinfo" in file:
-                assert len(file_dict[file].index) > 3
+                assert len(file_dict[file].index) >= num_kernels
        elif file.endswith(".pdf"):
            file_dict[file] = "pdf"
    if soc == "mi200":