Skip to content

Add dataprovider support #33

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

Open
Rohan138 wants to merge 13 commits into
base: develop
Choose a base branch
from
Open

Add dataprovider support #33

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
13 commits
Select commit Hold shift + click to select a range
0946333
Add local data support
Rohan138 Mar 21, 2025
9b1aa77
Remove data field
Rohan138 Mar 21, 2025
93612da
Merge branch 'develop' into ropotdar/add_nas_support
Rohan138 Mar 21, 2025
f16a6b5
Merge branch 'develop' into ropotdar/add_nas_support
Rohan138 Apr 1, 2025
e3fb802
Merge branch 'ROCm:develop' into ropotdar/add_nas_support
Rohan138 Apr 7, 2025
1922826
Add echo
Rohan138 Apr 7, 2025
ab7de7b
Add echo
Rohan138 Apr 7, 2025
9c1d328
Add echo
Rohan138 Apr 7, 2025
99cf154
Add llama2 7b and 70b nas for testing
Rohan138 Apr 28, 2025
beea1be
Merge branch 'develop' into ropotdar/add_nas_support
gargrahul May 1, 2025
1dc357f
Merge branch 'develop' into ropotdar/add_nas_support
Rohan138 May 8, 2025
b861623
Update Llama and Mistral data
Rohan138 May 8, 2025
7c9d80f
Add entries for llama FP8
Rohan138 May 9, 2025
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
3 changes: 0 additions & 3 deletions data.json
View file
Open in desktop

This file was deleted.

74 changes: 30 additions & 44 deletions models.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,119 +29,119 @@
"args": ""
},
{
"name": "pyt_vllm_llama-3.1-8b",
"name": "pyt_vllm_llama-2-7b",
"url": "",
"data": "meta-llama/Llama-2-7b-chat-hf",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
"multiple_results": "perf_Llama-3.1-8B-Instruct.csv",
"multiple_results": "perf_Llama-2-7b-chat-hf.csv",
"tags": [
"pyt",
"vllm"
],
"timeout": -1,
"args":
"--model_repo meta-llama/Llama-3.1-8B-Instruct --test_option latency,throughput --num_gpu 1 --datatype float16 --tunableop off"
"--model_repo meta-llama/Llama-2-7b-chat-hf --test_option latency,throughput --num_gpu 1 --datatype float16 --tunableop off"
},
{
"name": "pyt_vllm_llama-3.1-70b",
"name": "pyt_vllm_llama-2-70b",
"url": "",
"data": "meta-llama/Llama-2-70b-chat-hf",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
"multiple_results": "perf_Llama-3.1-70B-Instruct.csv",
"multiple_results": "perf_Llama-2-70b-chat-hf.csv",
"tags": [
"pyt",
"vllm"
],
"timeout": -1,
"args":
"--model_repo meta-llama/Llama-3.1-70B-Instruct --test_option latency,throughput --num_gpu 8 --datatype float16 --tunableop off"
"--model_repo meta-llama/Llama-2-70b-chat-hf --test_option latency,throughput --num_gpu 8 --datatype float16 --tunableop off"
},
{
"name": "pyt_vllm_llama-3.1-405b",
"name": "pyt_vllm_llama-3.1-8b",
"url": "",
"data": "meta-llama/Llama-3.1-8B-Instruct",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
"multiple_results": "perf_Llama-3.1-405B-Instruct.csv",
"multiple_results": "perf_Llama-3.1-8B-Instruct.csv",
"tags": [
"pyt",
"vllm"
],
"timeout": -1,
"args":
"--model_repo meta-llama/Llama-3.1-405B-Instruct --test_option latency,throughput --num_gpu 8 --datatype float16 --tunableop off"
"--model_repo meta-llama/Llama-3.1-8B-Instruct --test_option latency,throughput --num_gpu 1 --datatype float16 --tunableop off"
},
{
"name": "pyt_vllm_llama-3.2-11b-vision-instruct",
"name": "pyt_vllm_llama-3.1-70b",
"url": "",
"data": "meta-llama/Llama-3.1-70B-Instruct",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
"multiple_results": "perf_Llama-3.2-11B-Vision-Instruct.csv",
"multiple_results": "perf_Llama-3.1-70B-Instruct.csv",
"tags": [
"pyt",
"vllm"
],
"timeout": -1,
"args":
"--model_repo meta-llama/Llama-3.2-11B-Vision-Instruct --test_option latency,throughput --num_gpu 1 --datatype float16 --tunableop off"
"--model_repo meta-llama/Llama-3.1-70B-Instruct --test_option latency,throughput --num_gpu 8 --datatype float16 --tunableop off"
},
{
"name": "pyt_vllm_llama-2-7b",
"name": "pyt_vllm_llama-3.1-405b",
"url": "",
"data": "meta-llama/Llama-3.1-405B-Instruct",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
"multiple_results": "perf_Llama-2-7b-chat-hf.csv",
"multiple_results": "perf_Llama-3.1-405B-Instruct.csv",
"tags": [
"pyt",
"vllm"
],
"timeout": -1,
"args":
"--model_repo meta-llama/Llama-2-7b-chat-hf --test_option latency,throughput --num_gpu 1 --datatype float16 --tunableop off"
"--model_repo meta-llama/Llama-3.1-405B-Instruct --test_option latency,throughput --num_gpu 8 --datatype float16 --tunableop off"
},
{
"name": "pyt_vllm_llama-2-70b",
"name": "pyt_vllm_llama-3.2-11b-vision-instruct",
"url": "",
"data": "meta-llama/Llama-3.2-11B-Vision-Instruct",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
"multiple_results": "perf_Llama-2-70b-chat-hf.csv",
"multiple_results": "perf_Llama-3.2-11B-Vision-Instruct.csv",
"tags": [
"pyt",
"vllm"
],
"timeout": -1,
"args":
"--model_repo meta-llama/Llama-2-70b-chat-hf --test_option latency,throughput --num_gpu 8 --datatype float16 --tunableop off"
"--model_repo meta-llama/Llama-3.2-11B-Vision-Instruct --test_option latency,throughput --num_gpu 1 --datatype float16 --tunableop off"
},
{
"name": "pyt_vllm_mixtral-8x7b",
"url": "",
"data": "mistralai/Mixtral-8x7B-Instruct-v0.1",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -157,9 +157,9 @@
{
"name": "pyt_vllm_mixtral-8x22b",
"url": "",
"data": "mistralai/Mixtral-8x22B-Instruct-v0.1",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -175,9 +175,9 @@
{
"name": "pyt_vllm_mistral-7b",
"url": "",
"data": "mistralai/Mistral-7B-Instruct-v0.1",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -195,7 +195,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -213,7 +212,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -231,7 +229,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -249,7 +246,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -267,7 +263,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -285,7 +280,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -301,9 +295,9 @@
{
"name": "pyt_vllm_llama-3.1-8b_fp8",
"url": "",
"data": "amd/Llama-3.1-8B-Instruct-FP8-KV",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -319,9 +313,9 @@
{
"name": "pyt_vllm_llama-3.1-70b_fp8",
"url": "",
"data": "amd/Llama-3.1-70B-Instruct-FP8-KV",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -337,9 +331,9 @@
{
"name": "pyt_vllm_llama-3.1-405b_fp8",
"url": "",
"data": "amd/Llama-3.1-405B-Instruct-FP8-KV",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -357,7 +351,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -375,7 +368,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -393,7 +385,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -411,7 +402,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -429,7 +419,6 @@
"url": "",
"dockerfile": "docker/pyt_vllm",
"scripts": "scripts/vllm/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -447,7 +436,6 @@
"url": "",
"dockerfile": "docker/pytorch_train",
"scripts": "scripts/pytorch_train/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -464,7 +452,6 @@
"url": "",
"dockerfile": "docker/pytorch_train",
"scripts": "scripts/pytorch_train/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand All @@ -481,7 +468,6 @@
"url": "",
"dockerfile": "docker/pytorch_train",
"scripts": "scripts/pytorch_train/run.sh",
"data": "huggingface",
"n_gpus": "-1",
"owner": "mad.support@amd.com",
"training_precision": "",
Expand Down
6 changes: 6 additions & 0 deletions scripts/vllm/vllm_benchmark_report.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,6 +50,12 @@ model_org_name=(${model//// })
model_name=${model_org_name[1]}
tp=$numgpu

# Use local data if present
if [ -n "$MAD_DATAHOME" ]; then
echo "Using data from $MAD_DATAHOME"
model=$MAD_DATAHOME
fi

# perf configuration
export VLLM_USE_TRITON_FLASH_ATTN=0
export NCCL_MIN_NCHANNELS=112
Expand Down