From ed307a40bfa9ce526700d36b5e4bf8afb8725cc5 Mon Sep 17 00:00:00 2001 From: fzyzcjy <5236035+fzyzcjy@users.noreply.github.com> Date: Tue, 6 Jan 2026 22:43:52 +0800 Subject: [PATCH] Tiny add --gsp-ordered (#16575) --- python/sglang/bench_serving.py | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/python/sglang/bench_serving.py b/python/sglang/bench_serving.py index 8eca2251c..880221138 100644 --- a/python/sglang/bench_serving.py +++ b/python/sglang/bench_serving.py @@ -1862,8 +1862,8 @@ def sample_generated_shared_prefix_requests( total_input_tokens += prompt_len total_output_tokens += output_len_val - # Shuffle questions - random.shuffle(input_requests) + if not getattr(args, "gsp_ordered", False): + random.shuffle(input_requests) # Print statistics print(f"\nGenerated shared prefix dataset statistics:") @@ -3192,6 +3192,11 @@ if __name__ == "__main__": default=1, help="Number of turns for multi-turn conversations. If > 1, each prompt becomes a list of questions sharing the same system prefix.", ) + group.add_argument( + "--gsp-ordered", + action="store_true", + help="Keep requests in order without shuffling. By default, requests are shuffled randomly.", + ) mooncake_group = parser.add_argument_group("mooncake dataset arguments") mooncake_group.add_argument( "--mooncake-slowdown-factor",