Skip to content
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

Add more control arguments for benchmark_vllm_throughput #11291

Merged
merged 1 commit into from
Jun 12, 2024
Merged

Add more control arguments for benchmark_vllm_throughput #11291

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
15 changes: 12 additions & 3 deletions docker/llm/serving/xpu/docker/benchmark_vllm_throughput.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,6 +77,7 @@ def run_vllm(
gpu_memory_utilization: float = 0.9,
load_in_low_bit: str = "sym_int4",
max_num_batched_tokens: int = 5000,
max_num_seqs: int = 256,
) -> float:
from vllm import SamplingParams
from ipex_llm.vllm.xpu.engine import IPEXLLMClass as LLM
Expand All @@ -94,7 +95,8 @@ def run_vllm(
device=device,
enable_prefix_caching=enable_prefix_caching,
load_in_low_bit=load_in_low_bit,
max_num_batched_tokens=max_num_batched_tokens,)
max_num_batched_tokens=max_num_batched_tokens,
max_num_seqs=max_num_seqs,)


# Add the requests to the engine.
Expand Down Expand Up @@ -238,7 +240,8 @@ def main(args: argparse.Namespace):
args.tensor_parallel_size, args.seed, args.n, args.use_beam_search,
args.trust_remote_code, args.dtype, args.max_model_len,
args.enforce_eager, args.kv_cache_dtype, args.device,
args.enable_prefix_caching, args.gpu_memory_utilization, args.load_in_low_bit, args.max_num_batched_tokens)
args.enable_prefix_caching, args.gpu_memory_utilization, args.load_in_low_bit,
args.max_num_batched_tokens,args.max_num_seqs)
elif args.backend == "hf":
assert args.tensor_parallel_size == 1
elapsed_time = run_hf(requests, args.model, tokenizer, args.n,
Expand Down Expand Up @@ -348,9 +351,15 @@ def main(args: argparse.Namespace):

parser.add_argument('--max-num-batched-tokens',
type=int,
default=5000,
default=4096,
help='maximum number of batched tokens per iteration')

parser.add_argument('--max-num-seqs',
type=int,
default=256,
help='Maximum number of sequences per iteration.')


args = parser.parse_args()
if args.tokenizer is None:
args.tokenizer = args.model
Expand Down