-
Notifications
You must be signed in to change notification settings - Fork 5.9k
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.
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
make newFirstChunk
more effective
#9396
Comments
What's your workload? |
There are many ways to implement a column pool. For example,
There is still some problem to solve in the column pool implementation, I think we can leave this optimization in the future. For now, let's focused on chunk size control to improve the query latency. |
There are some pool implementations and the benchmark results in this gist: https://gist.github.com/zz-jason/c98d942dddbb32c134c9cadffd983f61 This is a simple lock free list implementation: https://github.com/zz-jason/lockfree, the benchmark can be gained by running |
I have tested |
Can you describe more on this part? |
We can give |
I used |
General Question
Here is a call graph about TiDB's CPU usage in attached file.
We can see
newFirstChunk
utilizes about 10% CPU of total.I have tried to use a
sync.Pool
to cache*column
s to improve it, but the result is not better. With the patchsync.Pool
itself becomes the new bottle neck.So please try to find a good wait to cache
*column
s to improve it.The text was updated successfully, but these errors were encountered: