Merge pull request #180 from mlcommons/dev

20240816: regular sync of mlperf-inference & dev with main

.github/workflows/test-mlperf-inference-abtf-poc.yml

@@ -37,7 +37,7 @@ jobs:
         cm pull repo mlcommons@cm4abtf --branch=poc
     - name: Test MLPerf Inference ABTF POC using ${{ matrix.backend }} on docker
       run: |
-        cm run script --tags=run-abtf,inference,_poc-demo --adr.compiler.tags=gcc --quiet  -v
+        cm run script --tags=run-abtf,inference,_poc-demo --test_query_count=5 --adr.compiler.tags=gcc --quiet  -v
 
   build2:
     runs-on: ${{ matrix.os }}
@@ -48,9 +48,6 @@ jobs:
         python-version: [ "3.8", "3.12" ]
         backend: [ "pytorch" ]
         implementation: [ "python" ]
-        exclude:
-        - os: ubuntu-24.04
-          python-version: "3.8"
 
     steps:
     - uses: actions/checkout@v3
@@ -92,4 +89,4 @@ jobs:
         cm pull repo mlcommons@cm4abtf --branch=poc
     - name: Test MLPerf Inference ABTF POC using ${{ matrix.backend }} on ${{ matrix.os }}
       run: |
-        cm run script --tags=run-abtf,inference,_poc-demo --quiet --env.CM_MLPERF_LOADGEN_BUILD_FROM_SRC=off  -v
+        cm run script --tags=run-abtf,inference,_poc-demo --quiet --env.CM_MLPERF_LOADGEN_BUILD_FROM_SRC=off  -v

.github/workflows/test-mlperf-inference-bert-deepsparse-tf-onnxruntime-pytorch.yml

@@ -38,7 +38,6 @@ jobs:
       run: |
         python3 -m pip install cmind
         cm pull repo --url=${{ github.event.pull_request.head.repo.html_url }} --checkout=${{ github.event.pull_request.head.ref }}
-        cm run script --quiet --tags=get,sys-utils-cm
     - name: Test MLPerf Inference Bert (DeepSparse, TF, ONNX, PyTorch)
       run: |
         cm run script --tags=run,mlperf,inference,generate-run-cmds,_submission,_short --submitter="cTuning" --model=bert-99 --backend=${{ matrix.backend }} --device=cpu --scenario=Offline --test_query_count=5 --precision=${{ matrix.precision }} --target_qps=1 -v --quiet

.github/workflows/test-mlperf-inference-gptj.yml

@@ -32,7 +32,6 @@ jobs:
       run: |
         python3 -m pip install cmind
         cm pull repo --url=${{ github.event.pull_request.head.repo.html_url }} --checkout=${{ github.event.pull_request.head.ref }}
-        cm run script --quiet --tags=get,sys-utils-cm
     - name: Test MLPerf Inference GPTJ
       run: |
         cm run script --tags=run,mlperf,inference,generate-run-cmds,_submission,_short --submitter="cTuning" --model=gptj --backend=${{ matrix.backend }} --device=cpu --scenario=Offline --test_query_count=1 --precision=${{ matrix.precision }} --target_qps=1 --quiet

.github/workflows/test-mlperf-inference-resnet50.yml

@@ -42,7 +42,6 @@ jobs:
       run: |
         python3 -m pip install cmind
         cm pull repo --url=${{ github.event.pull_request.head.repo.html_url }} --checkout=${{ github.event.pull_request.head.ref }}
-        cm run script --quiet --tags=get,sys-utils-cm
     - name: Test MLPerf Inference ResNet50
       run: |
         cm run script --tags=run,mlperf,inference,generate-run-cmds,_submission,_short --submitter="cTuning" --hw_name=default --model=resnet50 --implementation=${{ matrix.implementation }} --backend=${{ matrix.backend }} --device=cpu --scenario=Offline --test_query_count=500 --target_qps=1 -v --quiet 

.github/workflows/test-mlperf-inference-retinanet.yml

@@ -35,7 +35,6 @@ jobs:
       run: |
         python3 -m pip install cmind
         cm pull repo --url=${{ github.event.pull_request.head.repo.html_url }} --checkout=${{ github.event.pull_request.head.ref }}
-        cm run script --quiet --tags=get,sys-utils-cm
     - name: Test MLPerf Inference Retinanet using ${{ matrix.backend }}
       run: |
         cm run script --tags=run,mlperf,inference,generate-run-cmds,_submission,_short --submitter="cTuning" --hw_name=default --model=retinanet --implementation=${{ matrix.implementation }} --backend=${{ matrix.backend }} --device=cpu --scenario=Offline --test_query_count=5 --adr.compiler.tags=gcc --quiet  -v --target_qps=1

automation/script/module.py

@@ -1329,6 +1329,8 @@ def _run(self, i):
                         if "add_deps_recursive" in versions_meta:
                             self._merge_dicts_with_tags(add_deps_recursive, versions_meta['add_deps_recursive'])
 
+            env['CM_TMP_CURRENT_SCRIPT_PATH'] = path
+
             # Run chain of docker dependencies if current run cmd is from inside a docker container
             docker_deps = []
             if i.get('docker_run_deps'):
@@ -1446,6 +1448,7 @@ def _run(self, i):
             if pip_version_string != '':
                 logging.debug(recursion_spaces+'    # potential PIP version string (if needed): '+pip_version_string)
 
+
             # Check if pre-process and detect
             if 'preprocess' in dir(customize_code) and not fake_run:
 
@@ -2915,6 +2918,10 @@ def _run_deps(self, deps, clean_env_keys_deps, env, state, const, const_state, a
                 if from_cache and not d.get("dynamic", None):
                     continue
 
+                if d.get('env'):
+                    r = update_env_with_values(d['env'], False, env) #to update env local to a dependency
+                    if r['return']>0: return r
+
                 update_tags_from_env_with_prefix = d.get("update_tags_from_env_with_prefix", {})
                 for t in update_tags_from_env_with_prefix:
                     for key in update_tags_from_env_with_prefix[t]:
@@ -3034,9 +3041,6 @@ def _run_deps(self, deps, clean_env_keys_deps, env, state, const, const_state, a
 
                     utils.merge_dicts({'dict1':ii, 'dict2':d, 'append_lists':True, 'append_unique':True})
 
-                    r = update_env_with_values(ii['env']) #to update env local to a dependency
-                    if r['return']>0: return r 
-
                     r = self.cmind.access(ii)
                     if r['return']>0: return r
 
@@ -4274,9 +4278,13 @@ def find_cached_script(i):
             dependent_cached_path = cached_script.meta.get('dependent_cached_path', '')
             if dependent_cached_path:
                 if not os.path.exists(dependent_cached_path):
-                    #Need to rm this cache entry
-                    skip_cached_script = True
-                    continue
+                    #TODO Need to restrict the below check to within container env
+                    i['tmp_dep_cached_path'] = dependent_cached_path
+                    r = utils.call_internal_module(self_obj, __file__, 'module_misc', 'get_container_path_script', i)
+                    if not os.path.exists(r['value_env']):
+                        #Need to rm this cache entry
+                        skip_cached_script = True
+                        continue
 
             if not skip_cached_script:
                 cached_script_version = cached_script.meta.get('version', '')
@@ -4351,7 +4359,7 @@ def any_enable_or_skip_script(meta, env):
     return False
 
 ############################################################################################################
-def update_env_with_values(env, fail_on_not_found=False):
+def update_env_with_values(env, fail_on_not_found=False, extra_env={}):
     """
     Update any env key used as part of values in meta
     """
@@ -4382,14 +4390,19 @@ def update_env_with_values(env, fail_on_not_found=False):
             continue
 
         for tmp_value in tmp_values:
-            if tmp_value not in env and fail_on_not_found:
+            if tmp_value not in env and tmp_value not in extra_env and fail_on_not_found:
                 return {'return':1, 'error':'variable {} is not in env'.format(tmp_value)}
+            found_env = {}
             if tmp_value in env:
+                found_env = env
+            elif tmp_value in extra_env:
+                found_env = extra_env
+            if found_env:
                 if type(value) == str:
-                    value = value.replace("<<<"+tmp_value+">>>", str(env[tmp_value]))
+                    value = value.replace("<<<"+tmp_value+">>>", str(found_env[tmp_value]))
                 elif type(value) == list:
                     for i,val in enumerate(value):
-                        value[i] = value[i].replace("<<<"+tmp_value+">>>", str(env[tmp_value]))
+                        value[i] = value[i].replace("<<<"+tmp_value+">>>", str(found_env[tmp_value]))
 
         env[key] = value
 
@@ -4618,7 +4631,7 @@ def prepare_and_run_script_with_postprocessing(i, postprocess="postprocess"):
                     repo_to_report = 'https://github.com/'+script_repo_alias.replace('@','/')+'/issues'
 
             if repo_to_report == '':
-                repo_to_report = 'https://github.com/mlcommons/ck/issues'
+                repo_to_report = 'https://github.com/mlcommons/cm4mlops/issues'
 
             note = '''
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

automation/script/module_misc.py

@@ -1422,6 +1422,7 @@ def dockerfile(i):
 
         variations = meta.get('variations', {})
         docker_settings = meta.get('docker', {})
+        docker_settings['dockerfile_env'] = dockerfile_env
         state['docker'] = docker_settings
         add_deps_recursive = i.get('add_deps_recursive', {})
 
@@ -1430,6 +1431,7 @@ def dockerfile(i):
             return r
 
         docker_settings = state['docker']
+        dockerfile_env = docker_settings['dockerfile_env']
 
         if not docker_settings.get('run', True) and not i.get('docker_run_override', False):
             print("docker.run set to False in _cm.json")
@@ -1598,6 +1600,11 @@ def get_host_path(value):
 
     return value
 
+def get_container_path_script(i):
+    tmp_dep_cached_path = i['tmp_dep_cached_path']
+    value_mnt,value_env = get_container_path(tmp_dep_cached_path)
+    return {'return': 0, 'value_mnt': value_mnt, 'value_env': value_env}
+
 def get_container_path(value):
     path_split = value.split(os.sep)
     if len(path_split) == 1:
@@ -1739,7 +1746,7 @@ def docker(i):
 
     image_repo = i.get('docker_image_repo','')
     if image_repo == '':
-        image_repo = 'cknowledge'
+        image_repo = 'local'
 
     # Host system needs to have docker
     r = self_module.cmind.access({'action':'run',