Migrate MLPerf inference unofficial results repo to MLCommons (#59)

* Use mlcommons repo for uploading unofficial results of nvidia/intel github actions * Fix format * Fix rgat download path, added libbz2 deps for draw-graph-from-json * Support windows for pull-git-repo * Fix libbz2-dev detect * Added separate installation options for libbz2-dev and bzip2 * Restrict libbz2-dev install only for ubuntu (install-python-src) * Update VERSION
mlcommons · Dec 21, 2024 · d2db3b4 · d2db3b4
1 parent f399c2c
commit d2db3b4
Show file tree

Hide file tree

Showing 13 changed files with 117 additions and 25 deletions.
diff --git a/.github/workflows/test-intel-mlperf-inference-implementations.yml b/.github/workflows/test-intel-mlperf-inference-implementations.yml
@@ -22,5 +22,5 @@ jobs:
           export CM_REPOS=$HOME/GH_CM
           pip install --upgrade cm4mlops
           pip install tabulate
-          cm run script --tags=run-mlperf,inference,_all-scenarios,_submission,_full,_r4.1-dev --preprocess_submission=yes --execution_mode=valid  --pull_changes=yes --pull_inference_changes=yes --model=${{ matrix.model }} --submitter="MLCommons" --hw_name=IntelSPR.24c  --implementation=intel    --backend=pytorch    --category=datacenter --division=open --scenario=Offline  --docker_dt=yes --docker_it=no --docker_cm_repo=gateoverflow@cm4mlops --adr.compiler.tags=gcc --device=cpu --use_dataset_from_host=yes --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --clean  --docker --quiet
-          cm run script --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/gateoverflow/mlperf_inference_unofficial_submissions_v5.0 --repo_branch=main --commit_message="Results from GH action on SPR.24c" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=IntelSPR.24c
+          cm run script --tags=run-mlperf,inference,_all-scenarios,_submission,_full,_r4.1-dev --preprocess_submission=yes --execution_mode=valid  --pull_changes=yes --pull_inference_changes=yes --model=${{ matrix.model }} --submitter="MLCommons" --hw_name=IntelSPR.24c  --implementation=intel    --backend=pytorch    --category=datacenter --division=open --scenario=Offline  --docker_dt=yes --docker_it=no --docker_cm_repo=mlcommons@mlperf-automations --docker_cm_repo_branch=dev --adr.compiler.tags=gcc --device=cpu --use_dataset_from_host=yes --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --clean  --docker --quiet
+          cm run script --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/mlcommons/mlperf_inference_unofficial_submissions_v5.0 --repo_branch=auto-update --commit_message="Results from GH action on SPR.24c" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=IntelSPR.24c
diff --git a/.github/workflows/test-nvidia-mlperf-inference-implementations.yml b/.github/workflows/test-nvidia-mlperf-inference-implementations.yml
@@ -39,6 +39,6 @@ jobs:
           pip install --upgrade cm4mlops
           cm pull repo
           
-          cm run script --tags=run-mlperf,inference,_all-scenarios,_submission,_full,_r4.1-dev --preprocess_submission=yes --pull_changes=yes --pull_inference_changes=yes --execution_mode=valid --gpu_name=rtx_4090 --pull_changes=yes --pull_inference_changes=yes --model=${{ matrix.model }} --submitter="MLCommons" --hw_name=$hw_name --implementation=nvidia --backend=tensorrt --category=datacenter,edge --division=closed  --docker_dt=yes --docker_it=no --docker_cm_repo=gateoverflow@cm4mlops --adr.compiler.tags=gcc --device=cuda --use_model_from_host=yes --use_dataset_from_host=yes --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --clean  --docker --quiet
+          cm run script --tags=run-mlperf,inference,_all-scenarios,_submission,_full,_r4.1-dev --preprocess_submission=yes --pull_changes=yes --pull_inference_changes=yes --execution_mode=valid --gpu_name=rtx_4090 --pull_changes=yes --pull_inference_changes=yes --model=${{ matrix.model }} --submitter="MLCommons" --hw_name=$hw_name --implementation=nvidia --backend=tensorrt --category=datacenter,edge --division=closed  --docker_dt=yes --docker_it=no --docker_cm_repo=mlcommons@mlperf-automations --docker_cm_repo_branch=dev --adr.compiler.tags=gcc --device=cuda --use_model_from_host=yes --use_dataset_from_host=yes --results_dir=$HOME/gh_action_results --submission_dir=$HOME/gh_action_submissions --clean  --docker --quiet
 
-          cm run script --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/gateoverflow/mlperf_inference_unofficial_submissions_v5.0 --repo_branch=main --commit_message="Results from GH action on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name
+          cm run script --tags=push,github,mlperf,inference,submission --repo_url=https://github.com/mlcommons/mlperf_inference_unofficial_submissions_v5.0 --repo_branch=auto-update --commit_message="Results from GH action on NVIDIA_$hw_name" --quiet --submission_dir=$HOME/gh_action_submissions --hw_name=$hw_name
diff --git a/VERSION b/VERSION
@@ -1 +1 @@
-0.6.14
+0.6.15
diff --git a/automation/script/module.py b/automation/script/module.py
@@ -1798,9 +1798,16 @@ def _run(self, i):
 
             tmp_curdir = os.getcwd()
             if env.get('CM_OUTDIRNAME', '') != '':
-                if not os.path.exists(env['CM_OUTDIRNAME']):
-                    os.makedirs(env['CM_OUTDIRNAME'])
-                os.chdir(env['CM_OUTDIRNAME'])
+                if os.path.isabs(env['CM_OUTDIRNAME']) or recursion:
+                    c_outdirname = env['CM_OUTDIRNAME']
+                else:
+                    c_outdirname = os.path.join(
+                        env['CM_TMP_CURRENT_PATH'], env['CM_OUTDIRNAME'])
+                    env['CM_OUTDIRNAME'] = c_outdirname
+
+                if not os.path.exists(c_outdirname):
+                    os.makedirs(c_outdirname)
+                os.chdir(c_outdirname)
 
             # Check if pre-process and detect
             if 'preprocess' in dir(customize_code) and not fake_run:
@@ -5860,7 +5867,10 @@ def convert_env_to_script(env, os_info, start_script=None):
                 key = key[1:]
 
             # Append the existing environment variable to the new value
-            env_value = f"{env_separator.join(env_value)}{env_separator}{os_info['env_var'].replace('env_var', key)}"
+            env_value = f"""{
+                env_separator.join(env_value)}{env_separator}{
+                os_info['env_var'].replace(
+                    'env_var', key)}"""
 
         # Replace placeholders in the platform-specific environment command
         env_command = os_info['set_env'].replace(

diff --git a/...et-dataset-mlperf-inference-igbh/_cm.yaml → script/get-dataset-igbh/_cm.yaml b/...et-dataset-mlperf-inference-igbh/_cm.yaml → script/get-dataset-igbh/_cm.yaml
@@ -1,4 +1,4 @@
-alias: get-dataset-mlperf-inference-igbh
+alias: get-dataset-igbh
 automation_alias: script
 automation_uid: 5b4e0237da074764
 cache: true
@@ -37,6 +37,8 @@ prehook_deps:
       CM_DOWNLOAD_FILENAME: node_feat.npy
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper/
     extra_cache_tags: dataset,igbh,paper,node_feat
+    force_env_keys:
+    - CM_OUTDIRNAME
     force_cache: true
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
@@ -54,6 +56,8 @@ prehook_deps:
       CM_DOWNLOAD_FILENAME: node_label_19.npy
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper/
     extra_cache_tags: dataset,igbh,paper,node_label_19
+    force_env_keys:
+    - CM_OUTDIRNAME
     force_cache: true
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
@@ -72,6 +76,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper/
     extra_cache_tags: dataset,igbh,paper,node_label_2K
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -89,6 +95,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper/
     extra_cache_tags: dataset,igbh,paper,paper_id_index_mapping
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -107,6 +115,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper__cites__paper/
     extra_cache_tags: dataset,igbh,paper_cites_paper,edge_index
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -125,6 +135,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/author/
     extra_cache_tags: dataset,igbh,author,author_id_index_mapping
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -142,6 +154,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/author/
     extra_cache_tags: dataset,igbh,author,node_feat
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -160,6 +174,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/conference/
     extra_cache_tags: dataset,igbh,conference,conference_id_index_mapping
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -177,6 +193,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/conference/
     extra_cache_tags: dataset,igbh,conference,node_feat
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -195,6 +213,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/institute/
     extra_cache_tags: dataset,igbh,institute,institute_id_index_mapping
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -212,6 +232,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/institute/
     extra_cache_tags: dataset,igbh,institute,node_feat
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -230,6 +252,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/journal/
     extra_cache_tags: dataset,igbh,journal,journal_id_index_mapping
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -247,6 +271,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/journal/
     extra_cache_tags: dataset,igbh,journal,node_feat
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -265,6 +291,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/fos/
     extra_cache_tags: dataset,igbh,fos,fos_id_index_mapping
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -282,6 +310,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/fos/
     extra_cache_tags: dataset,igbh,fos,node_feat
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -300,6 +330,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/author__affiliated_to__institute/
     extra_cache_tags: dataset,igbh,author_affiliated_to_institute,edge_index
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -318,6 +350,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper__published__journal/
     extra_cache_tags: dataset,igbh,paper_published_journal,edge_index
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -336,6 +370,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper__topic__fos/
     extra_cache_tags: dataset,igbh,paper_topic_fos,edge_index
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -354,6 +390,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper__venue__conference/
     extra_cache_tags: dataset,igbh,paper_venue_conference,edge_index
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'
@@ -372,6 +410,8 @@ prehook_deps:
       CM_DOWNLOAD_PATH: <<<CM_DATASET_IGBH_DOWNLOAD_LOCATION>>>/full/processed/paper__written_by__author/
     extra_cache_tags: dataset,igbh,paper_written_by_author,edge_index
     force_cache: true
+    force_env_keys:
+    - CM_OUTDIRNAME
     enable_if_env:
       CM_DATASET_IGBH_TYPE: 
       - 'full'

diff --git a/...ataset-mlperf-inference-igbh/customize.py → script/get-dataset-igbh/customize.py b/...ataset-mlperf-inference-igbh/customize.py → script/get-dataset-igbh/customize.py
diff --git a/.../get-dataset-mlperf-inference-igbh/run.sh → script/get-dataset-igbh/run.sh b/.../get-dataset-mlperf-inference-igbh/run.sh → script/get-dataset-igbh/run.sh
diff --git a/script/get-generic-sys-util/_cm.yaml b/script/get-generic-sys-util/_cm.yaml
@@ -212,21 +212,33 @@ variations:
         brew: ''
         dnf: boost-devel
         yum: boost-devel
-  libbz2-dev:
+  bzip2:
     env:
-      CM_SYS_UTIL_NAME: libbz2_dev
+      CM_SYS_UTIL_NAME: bzip2
       CM_SYS_UTIL_VERSION_CMD_OVERRIDE: bzcat --version 2>&1 | grep bzip > tmp-ver.out
       CM_SYS_UTIL_VERSION_RE: ([0-9]+(\.[0-9]+)+)
       CM_TMP_VERSION_DETECT_GROUP_NUMBER: 1
     new_env_keys:
+    - CM_BZIP2_VERSION
+    state:
+      bzip2:
+        apt: bzip2
+        brew: bzip2
+        dnf: bzip2
+        yum: bzip2
+  libbz2-dev:
+    env:
+      CM_SYS_UTIL_NAME: libbz2_dev
+      CM_SYS_UTIL_VERSION_CMD: dpkg -s libbz2-dev | grep 'Version'
+      CM_SYS_UTIL_VERSION_RE: ([0-9]+(\.[0-9]+)+)
+      CM_TMP_VERSION_DETECT_GROUP_NUMBER: 0
+    new_env_keys:
     - CM_LIBBZ2_DEV_VERSION
     state:
       libbz2_dev:
         apt: libbz2-dev
-        brew: bzip2
         dnf: libbzip2-devel
         yum: libbzip2-devel
-        zlib-devel: libbz2-devel
   libev-dev:
     env:
       CM_SYS_UTIL_NAME: libev_dev

diff --git a/script/get-ml-model-rgat/_cm.yaml b/script/get-ml-model-rgat/_cm.yaml
@@ -21,12 +21,14 @@ prehook_deps:
     CM_TMP_REQUIRE_DOWNLOAD:
     - 'yes'
   env:
-    CM_DOWNLOAD_FINAL_ENV_NAME: CM_ML_MODEL_PATH
-  extra_cache_tags: rgat,gnn,model
+    CM_DOWNLOAD_FINAL_ENV_NAME: RGAT_DIR_PATH
+  extra_cache_tags: rgat,gnn,model,ml-model
   force_cache: true
   names:
-  - dae
-  tags: download-and-extract
+  - download-file
+  tags: download,file
+  force_env_keys:
+    - CM_OUTDIRNAME
   update_tags_from_env_with_prefix:
     _url.:
     - CM_DOWNLOAD_URL
@@ -55,7 +57,7 @@ variations:
     group: download-source
   rclone:
     adr:
-      dae:
+      download-file:
         tags: _rclone
     env:
       CM_DOWNLOAD_TOOL: rclone
@@ -65,3 +67,4 @@ variations:
     env:
       CM_ML_MODEL_STARTING_WEIGHTS_FILENAME: https://github.com/mlcommons/inference/tree/master/graph/R-GAT#download-model-using-rclone
       CM_DOWNLOAD_URL: mlc-inference:mlcommons-inference-wg-public/R-GAT/RGAT.pt
+      CM_DOWNLOAD_FILENAME: RGAT
diff --git a/script/get-ml-model-rgat/customize.py b/script/get-ml-model-rgat/customize.py
@@ -20,9 +20,10 @@ def postprocess(i):
     env = i['env']
 
     if env.get('CM_ML_MODEL_RGAT_CHECKPOINT_PATH', '') == '':
-        env['CM_ML_MODEL_RGAT_CHECKPOINT_PATH'] = os.path.join(
-            env['CM_ML_MODEL_PATH'], "RGAT.pt")
-    elif env.get('CM_ML_MODEL_PATH', '') == '':
+        env['CM_ML_MODEL_RGAT_CHECKPOINT_PATH'] = env.get(
+            'RGAT_CHECKPOINT_PATH', os.path.join(env['RGAT_DIR_PATH'], "RGAT.pt"))
+
+    if env.get('CM_ML_MODEL_PATH', '') == '':
         env['CM_ML_MODEL_PATH'] = env['CM_ML_MODEL_RGAT_CHECKPOINT_PATH']
 
     env['RGAT_CHECKPOINT_PATH'] = env['CM_ML_MODEL_RGAT_CHECKPOINT_PATH']

diff --git a/script/install-python-src/_cm.yaml b/script/install-python-src/_cm.yaml
@@ -16,6 +16,9 @@ deps:
 - tags: detect,cpu
 - tags: get,generic-sys-util,_libffi-dev
 - tags: get,generic-sys-util,_libbz2-dev
+  enable_if_env:
+    CM_HOST_OS_FLAVOR:
+    - ubuntu
 - tags: get,generic-sys-util,_libssl-dev
 - enable_if_env:
     CM_HOST_OS_FLAVOR:

diff --git a/script/pull-git-repo/customize.py b/script/pull-git-repo/customize.py
@@ -7,9 +7,6 @@ def preprocess(i):
 
     os_info = i['os_info']
 
-    if os_info['platform'] == 'windows':
-        return {'return': 1, 'error': 'Windows is not supported in this script yet'}
-
     env = i['env']
     meta = i['meta']
 

diff --git a/script/pull-git-repo/run.bat b/script/pull-git-repo/run.bat
@@ -0,0 +1,26 @@
+@echo off
+setlocal enabledelayedexpansion
+
+REM Save the current directory
+set "CUR_DIR=%CD%"
+set "SCRIPT_DIR=%CM_TMP_CURRENT_SCRIPT_PATH%"
+
+REM Change to the specified path
+set "path=%CM_GIT_CHECKOUT_PATH%"
+echo cd %path%
+
+cd /d "%path%"
+if errorlevel 1 (
+    echo Failed to change directory to %path%
+    exit /b %errorlevel%
+)
+
+REM Execute the Git pull command
+echo %CM_GIT_PULL_CMD%
+call %CM_GIT_PULL_CMD%
+REM Don't fail if there are local changes
+REM if errorlevel 1 exit /b %errorlevel%
+
+REM Return to the original directory
+cd /d "%CUR_DIR%"
+endlocal