From 2de61f62b12d14f95d840b5f7bba9b5189415f11 Mon Sep 17 00:00:00 2001
From: jainapurva <apurvajain.kota@gmail.com>
Date: Sun, 16 Nov 2025 22:27:46 -0800
Subject: [PATCH 1/2] Test int8wo delay

---
 .github/workflows/run_microbenchmarks.yml                 | 2 +-
 .../dashboard/microbenchmark_quantization_config.yml      | 8 ++++----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/.github/workflows/run_microbenchmarks.yml b/.github/workflows/run_microbenchmarks.yml
index 67f977d051..2cac203532 100644
--- a/.github/workflows/run_microbenchmarks.yml
+++ b/.github/workflows/run_microbenchmarks.yml
@@ -22,7 +22,7 @@ jobs:
       - name: Setup miniconda
         uses: pytorch/test-infra/.github/actions/setup-miniconda@main
         with:
-          python-version: "3.9"
+          python-version: "3.10"
 
       - name: Run benchmark
         shell: bash
diff --git a/benchmarks/dashboard/microbenchmark_quantization_config.yml b/benchmarks/dashboard/microbenchmark_quantization_config.yml
index 8156422668..7a32f5e43d 100644
--- a/benchmarks/dashboard/microbenchmark_quantization_config.yml
+++ b/benchmarks/dashboard/microbenchmark_quantization_config.yml
@@ -2,10 +2,10 @@
 benchmark_mode: "inference"
 quantization_config_recipe_names: # Will run a baseline inference for model by default, without quantization for comparison
   - "int8wo"
-  - "int8dq"
-  - "float8dq-tensor"
-  - "float8dq-row"
-  - "float8wo"
+  # - "int8dq"
+  # - "float8dq-tensor"
+  # - "float8dq-row"
+  # - "float8wo"
 output_dir: "benchmarks/microbenchmarks/results"
 model_params:
   - name: "small_bf16_linear"

From 588019e2943c12128eeee5c5126685b438dcc547 Mon Sep 17 00:00:00 2001
From: Apurva Jain <apurvajain.kota@gmail.com>
Date: Mon, 17 Nov 2025 11:18:06 -0800
Subject: [PATCH 2/2] Update quantization config for small_bf16_linear model

---
 benchmarks/dashboard/microbenchmark_quantization_config.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/benchmarks/dashboard/microbenchmark_quantization_config.yml b/benchmarks/dashboard/microbenchmark_quantization_config.yml
index 7a32f5e43d..807a066f5f 100644
--- a/benchmarks/dashboard/microbenchmark_quantization_config.yml
+++ b/benchmarks/dashboard/microbenchmark_quantization_config.yml
@@ -12,7 +12,7 @@ model_params:
     matrix_shapes:
       - name: "small_sweep"
         min_power: 10
-        max_power: 15
+        max_power: 14
     high_precision_dtype: "torch.bfloat16"
     torch_compile_mode: "max-autotune"
     device: "cuda"