What do I mean by slow?
At least on emacs, edits are taking too long, when I expand edit tool, the streaming is very slow.
Same issue when I expand thinking block too. This sometimes is triggering timeouts and LLM is stuck at the same step.
This happens with every model and every provider, I tried GitHub copilot, Azure openAI and ChatGPT.
I reverted back to 0.128.2, which is working for me and performing better with Emacs. Since it is working as expected with latest Melpa release and older version of ECA, I do not think it is a Emacs issue.
I do not see anything unusual in the logs, How do I debug to find the blocking process which is causing this issue in latest versions? 0.130 to 1.33 might be susceptible to this behavior, but I have not tested every version after 0.128.2, I will when I have time.
Is there a way to benchmark performance for different versions?
What do I mean by slow?
At least on emacs, edits are taking too long, when I expand edit tool, the streaming is very slow.
Same issue when I expand
thinkingblock too. This sometimes is triggering timeouts and LLM is stuck at the same step.This happens with every model and every provider, I tried GitHub copilot, Azure openAI and ChatGPT.
I reverted back to 0.128.2, which is working for me and performing better with Emacs. Since it is working as expected with latest Melpa release and older version of ECA, I do not think it is a Emacs issue.
I do not see anything unusual in the logs, How do I debug to find the blocking process which is causing this issue in latest versions? 0.130 to 1.33 might be susceptible to this behavior, but I have not tested every version after 0.128.2, I will when I have time.
Is there a way to benchmark performance for different versions?