Conversation
bc275a5 to
9b4288c
Compare
| "hardware": "cpu", | ||
| "tuned": "no", | ||
| "comment": "", | ||
| "tags": ["column-oriented", "lukewarm-cold-run"], |
There was a problem hiding this comment.
It's still lukewarm since the Python process is kept around between query runs. The tag should be kept as is. Am I missing something?
|
@rschu1ze thanks for handling the issue. Are you going to rerun the benchmarks as well? The results are likely to change after your modifications. |
|
I triggered the automation but all jobs fell victim to GitHub's outage today :-( Trying again now. |
| "proprietary": "no", | ||
| "hardware": "cpu", | ||
| "tuned": "no", | ||
| "tags": ["column-oriented","dataframe","lukewarm-cold-run"], |
There was a problem hiding this comment.
Same as with polars, the lukewarm-cold-run tag should be kept. Worth checking other changed results too, e.g. it looks like duckdb-dataframe also keeps Python process around.
GH seems running normally now. Could you retry? |
|
Sure (however this PR is currently low priority for me, I work on it on the side, probably tomorrow or on the weekend). |
Sorry, I just wanted to check if this PR is still on your radar. No need to rush, I understand the circumstances. |
|
Hey team, quick question. I thought that the tag was for the systems that do not clear OS cache and do not restart with every query. Am I wrong? Forcing to all the systems restarting for every query is that in the roadmap yet? Thank you |
|
Hey @xe-nvdk, my understanding is that lukewarm runs only clear OS page cache and do not require restarts. True cold runs require database restarts to ensure the internal database caches are also cleared. Many results have not been migrated to 'true cold' runs yet, I think @rschu1ze wanted some help crowd-sourcing the refresh. Citation: |
|
I somehow pushed the wrong branch to the wrong location, GitHub closed this PR and won't let me open it again. |
I guess you could create a new branch and open a new PR. |
Resolves #774