Can we move PixelTracks used for HI reco to patatrack PixelTracks, using GPU?

[clacaputo]

This issue is intended to initiate a discussion about whether the pixeltracks used for HI reco can be moved to patatrack pixel tracks that also run on GPU.

This would be an advantage also for 2022 data taking since the HI reco happens during the Christmas period at CERN also using the HLT farm that has gpus.

[cmsbuild]

A new Issue was created by @clacaputo Claudio Caputo.

@Dr15Jones, @perrotta, @dpiparo, @makortel, @smuzaffar, @qliphy can you please review it and eventually sign/assign? Thanks.

cms-bot commands are listed here

[clacaputo]

assign reconstruction

[cmsbuild]

New categories assigned: reconstruction

@jpata,@slava77,@clacaputo you have been requested to review this Pull request/Issue and eventually sign? Thanks

[clacaputo]

assign heterogeneous

[cmsbuild]

New categories assigned: heterogeneous

@fwyzard,@makortel you have been requested to review this Pull request/Issue and eventually sign? Thanks

[clacaputo]

Tagging HI reco contact: @mandrenguyen

[mandrenguyen]

To give some context, we have been running stand-alone pixel tracking in the offline heavy-ion reconstruction for a while.
We recently moved to the fit model used by patatrack #36558
We are running the patatrack code online, as is done in pp. I understand that this code is visible offline, as well.
I don't know of any good reason not to simply run the patatrack code.
I propose to organize a chat to discuss this.
On the heavy-ion side the tracking is manged by @CesarBernardes and @abaty
I've already been contacted by @AdrianoDee from the patatrack team.
I'll put the reco conveners in the loop, and we can add in any other relevant people from there.

[VinInn]

I suggest you just try and eventually report issues.

[jpata]

type trk

[jpata]

@mandrenguyen could you summarize the conclusions of the meeting? Is it planned to be implemented by HI, or should we close this issue?

[mandrenguyen]

At the time of the meeting that you refer to, @CesarBernardes did some comparisons, but it appeared that certain selections would need to be optimized in patatrack to match our offline pixel tracking. I believe there was some follow-up after the meeting by @CesarBernardes and @AdrianoDee Perhaps they can report the latest status here.

[AdrianoDee]

Hi @jpata and @mandrenguyen yes, with Cesar (not being able to tag him) we have been working on that and tackled down why we were loosing efficiency. Now we would need to test the gpu pixel tracking to better asses the performace. My idea is to anyway introduce a wf that runs the HIon pixel tracking on gpu so that the HI people may have a better look at it.

[mmusich]

type tracking

[mmusich]

@AdrianoDee @cms-sw/reconstruction-l2 is this issue still active after the merge of #41632 ?

[AdrianoDee]

Could be closed indeed.

[iarspider]

@cmsbuild please close