[GH-ISSUE #698] 2.2 rc1版本去掉了性能提示 #509

New issue

Closed

opened 2026-10-29 14:40:38 +00:00 by bsakaguchi · 3 comments

bsakaguchi commented

2026-10-29 14:40:38 +00:00

Owner

Originally created by @yeungtuzi on GitHub (Feb 26, 2025).
Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/698

之前的版本 make dev_install 会在每次请求结束后输出prefill和decode的速度和耗时，新版本不输出了，怎么设置可以让其继续输出？

Originally created by @yeungtuzi on GitHub (Feb 26, 2025). Original GitHub issue: https://github.com/kvcache-ai/ktransformers/issues/698 之前的版本 make dev_install 会在每次请求结束后输出prefill和decode的速度和耗时，新版本不输出了，怎么设置可以让其继续输出？

bsakaguchi closed this issue

2026-10-29 14:40:38 +00:00

bsakaguchi commented

2026-10-29 14:40:39 +00:00

Author

Owner

@Azure-Tang commented on GitHub (Feb 27, 2025):

local chat么？您是怎么跑的

@Azure-Tang commented on GitHub (Feb 27, 2025): local chat么？您是怎么跑的

bsakaguchi commented

2026-10-29 14:40:39 +00:00

Author

Owner

@yeungtuzi commented on GitHub (Feb 27, 2025):

HF_ENDPOINT=https://hf-mirror.com/
PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True
ktransformers
--gguf_path /root/DeepSeek-R1-GGUF/DeepSeek-R1-Q4_K_M/
--model_path deepseek-ai/DeepSeek-R1
--model_name DeepSeek-R1
--cpu_infer 60
--max_new_tokens 131072
--cache_lens 131072
--force_think
--cache_q4 TRUE
--max_response_tokens 131072
--use_cuda_graph
--host 0.0.0.0
--mode="long_context"
--log_level "debug"
--port 10087

@yeungtuzi commented on GitHub (Feb 27, 2025): HF_ENDPOINT=https://hf-mirror.com/ \ PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True \ ktransformers \ --gguf_path /root/DeepSeek-R1-GGUF/DeepSeek-R1-Q4_K_M/ \ --model_path deepseek-ai/DeepSeek-R1 \ --model_name DeepSeek-R1 \ --cpu_infer 60 \ --max_new_tokens 131072 \ --cache_lens 131072 \ --force_think \ --cache_q4 TRUE \ --max_response_tokens 131072 \ --use_cuda_graph \ --host 0.0.0.0 \ --mode="long_context" \ --log_level "debug" \ --port 10087

bsakaguchi commented

2026-10-29 14:40:39 +00:00

Author

Owner

@yeungtuzi commented on GitHub (Feb 27, 2025):

今天又更新了一下git，问题已经解决，又出性能报告了。

@yeungtuzi commented on GitHub (Feb 27, 2025): 今天又更新了一下git，问题已经解决，又出性能报告了。