add OMP_NUMNER_THREADS in cpu benchmark; increase request to get gpu … (

#2092) * add OMP_NUMNER_THREADS in cpu benchmark; increase request to get gpu report for mnist * increase number of requests in benchmark --------- Co-authored-by: Ankith Gunapal <agunapal@meta.com>
pytorch · Jan 31, 2023 · 05e80be · 05e80be
1 parent d264163
commit 05e80be
Show file tree

Hide file tree

Showing 7 changed files with 11 additions and 10 deletions.
diff --git a/.github/workflows/benchmark_nightly_cpu.yml b/.github/workflows/benchmark_nightly_cpu.yml
@@ -35,6 +35,7 @@ jobs:
             sudo apt-get update -y
             sudo apt-get install -y apache2-utils
             pip install -r benchmarks/requirements-ab.txt
+            export omp_num_threads=1
       - name: Benchmark cpu nightly
         run: python benchmarks/auto_benchmark.py --input benchmarks/benchmark_config_cpu.yaml --skip false
       - name: Save benchmark artifacts

diff --git a/benchmarks/models_config/bert_multi_gpu.yaml b/benchmarks/models_config/bert_multi_gpu.yaml
@@ -12,9 +12,9 @@ bert:
             - 4
             - 8
         input: "./examples/Huggingface_Transformers/Seq_classification_artifacts/sample_text_captum_input.txt"
-        requests: 10000
+        requests: 50000
         concurrency: 100
         backend_profiling: False
         exec_env: "local"
         processors:
-            - "gpus": "all"
+            - "gpus": "all"
diff --git a/benchmarks/models_config/bert_multi_gpu_better_transformer.yaml b/benchmarks/models_config/bert_multi_gpu_better_transformer.yaml
@@ -15,7 +15,7 @@ bert_bt:
             - 32
             - 64
         input: "./examples/Huggingface_Transformers/Seq_classification_artifacts/sample_text_captum_input.txt"
-        requests: 10000
+        requests: 50000
         concurrency: 100
         backend_profiling: False
         exec_env: "local"

diff --git a/benchmarks/models_config/bert_multi_gpu_no_better_transformer.yaml b/benchmarks/models_config/bert_multi_gpu_no_better_transformer.yaml
@@ -15,7 +15,7 @@ bert_no_bt:
             - 32
             - 64
         input: "./examples/Huggingface_Transformers/Seq_classification_artifacts/sample_text_captum_input.txt"
-        requests: 10000
+        requests: 50000
         concurrency: 100
         backend_profiling: False
         exec_env: "local"

diff --git a/benchmarks/models_config/mnist.yaml b/benchmarks/models_config/mnist.yaml
@@ -11,11 +11,11 @@ mnist:
             - 2
             - 4
             - 8
-        requests: 10000
+        requests: 500000
         concurrency: 10
         input: "./examples/image_classifier/mnist/test_data/0.png"
         backend_profiling: False
         exec_env: "local"
         processors:
             - "cpu"
-            - "gpus": "all"
+            - "gpus": "all"
diff --git a/benchmarks/models_config/wf_dog_breed.yaml b/benchmarks/models_config/wf_dog_breed.yaml
@@ -10,11 +10,11 @@ dog_breed:
         batch_size:
             - 1
             - 2
-        requests: 10000
+        requests: 50000
         concurrency: 10
         input: "./docs/images/kitten_small.jpg"
         backend_profiling: False
         exec_env: "local"
         processors:
             - "cpu"
-            - "gpus": "all"
+            - "gpus": "all"
diff --git a/benchmarks/models_config/wf_nmt_dualtranslation.yaml b/benchmarks/models_config/wf_nmt_dualtranslation.yaml
@@ -12,11 +12,11 @@ nmt_dual_translation:
             - 2
             - 4
             - 8
-        requests: 10000
+        requests: 50000
         concurrency: 10
         input: "./examples/Workflows/nmt_transformers_pipeline/model_input/sample.txt"
         backend_profiling: False
         exec_env: "local"
         processors:
             - "cpu"
-            - "gpus": "all"
+            - "gpus": "all"