Skip to content

Fixes for nvidia-mlperf-inference #77

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Dec 24, 2024
Merged

Fixes for nvidia-mlperf-inference #77

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
10f1481
Fixed stable-diffusion-xl name in SUT configs
arjunsuresh Dec 24, 2024
a10c616
Fix tensorrt detect on aarch64
arjunsuresh Dec 24, 2024
83d9d2d
Added torch deps for get-ml-model-gptj-nvidia
arjunsuresh Dec 24, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions script/get-ml-model-gptj/_cm.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -161,6 +161,10 @@ variations:
- python3
tags: get,python3
- tags: get,generic-python-lib,_package.safetensors
- tags: get,generic-python-lib,_torch
names:
- torch
- pytorch
rclone:
add_deps_recursive:
dae:
Expand Down
2 changes: 1 addition & 1 deletion script/get-ml-model-gptj/run-nvidia.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,5 +17,5 @@ export DOCKER_RUN_ARGS=" -v ${CM_NVIDIA_MLPERF_SCRATCH_PATH}:/mnt"
make -C docker run LOCAL_USER=1
test $? -eq 0 || exit $?

${CM_PYTHON_BIN_WITH_PATH} ${CM_MLPERF_INFERENCE_NVIDIA_CODE_PATH}/code/gptj/tensorrt/onnx_tune.py --fp8-scalers-path=${CM_NVIDIA_MLPERF_SCRATCH_PATH}/models/GPTJ-6B/fp8-quantized-ammo/GPTJ-FP8-quantized/rank0.safetensors --scaler 1.005 --index 15
PYTHONPATH='' ${CM_PYTHON_BIN_WITH_PATH} ${CM_MLPERF_INFERENCE_NVIDIA_CODE_PATH}/code/gptj/tensorrt/onnx_tune.py --fp8-scalers-path=${CM_NVIDIA_MLPERF_SCRATCH_PATH}/models/GPTJ-6B/fp8-quantized-ammo/GPTJ-FP8-quantized/rank0.safetensors --scaler 1.005 --index 15
test $? -eq 0 || exit $?
38 changes: 0 additions & 38 deletions ...onfigs/IntelSPR.24c/intel-implementation/cpu-device/pytorch-framework/default-config.yaml
View file
Open in desktop

This file was deleted.

30 changes: 0 additions & 30 deletions ...TX4090x1/nvidia_original-implementation/gpu-device/tensorrt-framework/default-config.yaml
View file
Open in desktop

This file was deleted.

5 changes: 2 additions & 3 deletions ...mplementation/gpu-device/tensorrt-framework/framework-version-default/default-config.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,9 +28,8 @@ retinanet:
target_qps: 850.0
Server:
target_qps: 630.0
sdxl:
stable-diffusion-xl:
Offline:
target_qps: 0.7
Server:
target_qps: 0.3

target_qps: 0.3
38 changes: 0 additions & 38 deletions ...TX4090x2/nvidia_original-implementation/gpu-device/tensorrt-framework/default-config.yaml
View file
Open in desktop

This file was deleted.

2 changes: 1 addition & 1 deletion ...mplementation/gpu-device/tensorrt-framework/framework-version-default/default-config.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@
target_qps: 8
Server:
target_qps: 7
sdxl:
stable-diffusion-xl:
Offline:
target_qps: 1.3
Server:
Expand Down
2 changes: 1 addition & 1 deletion script/get-tensorrt/customize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,7 +13,7 @@ def preprocess(i):

# Not enforcing dev requirement for now
if env.get('CM_TENSORRT_TAR_FILE_PATH', '') == '' and env.get(
'CM_TENSORRT_REQUIRE_DEV1', '') != 'yes' and env.get('CM_HOST_PLATFORM_FLAVOR', '') != 'aarch64':
'CM_TENSORRT_REQUIRE_DEV1', '') != 'yes' and env.get('CM_HOST_PLATFORM_FLAVOR_', '') != 'aarch64':

if os_info['platform'] == 'windows':
extra_pre = ''
Expand Down
Loading