Unverified Commit 0fdb3127 authored by zhyncs's avatar zhyncs Committed by GitHub
Browse files

feat: update bench serving (#685)

parent 5ad033a0
...@@ -619,8 +619,11 @@ async def benchmark( ...@@ -619,8 +619,11 @@ async def benchmark(
and metrics.output_throughput is not None and metrics.output_throughput is not None
): ):
result = { result = {
"backend": args.backend,
"dataset_name": args.dataset_name, "dataset_name": args.dataset_name,
"request_rate": request_rate, "request_rate": request_rate,
"total_input": metrics.total_input,
"total_output": metrics.total_output,
"median_ttft": metrics.median_ttft_ms, "median_ttft": metrics.median_ttft_ms,
"median_itl": metrics.mean_itl_ms, "median_itl": metrics.mean_itl_ms,
"output_token_throughput": metrics.output_throughput, "output_token_throughput": metrics.output_throughput,
...@@ -674,8 +677,11 @@ async def benchmark( ...@@ -674,8 +677,11 @@ async def benchmark(
def parse_request_rate_range(request_rate_range): def parse_request_rate_range(request_rate_range):
if len(request_rate_range.split(",")) == 3:
start, stop, step = map(int, request_rate_range.split(",")) start, stop, step = map(int, request_rate_range.split(","))
return list(range(start, stop, step)) return list(range(start, stop, step))
else:
return list(map(int, request_rate_range.split(",")))
def fire(args: argparse.Namespace): def fire(args: argparse.Namespace):
...@@ -895,7 +901,7 @@ if __name__ == "__main__": ...@@ -895,7 +901,7 @@ if __name__ == "__main__":
"--request-rate-range", "--request-rate-range",
type=str, type=str,
default="2,34,2", default="2,34,2",
help="Range of request rates in the format start,stop,step. Default is 2,34,2", help="Range of request rates in the format start,stop,step. Default is 2,34,2. It also supports a list of request rates, requiring the parameters to not equal three.",
) )
parser.add_argument("--output-file", type=str, help="Output JSONL file name.") parser.add_argument("--output-file", type=str, help="Output JSONL file name.")
parser.add_argument( parser.add_argument(
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment