Upgrade to CMSSW 14_1_0_pre3

[ariostas]

CMSSW 14_1_0_pre3 was released recently and it has some nice additions, including CUDA 12.4 (which fully fixes the printf issue from a while back), and fixes the ROCm compilation so that is can now compile LSTCore.

This PR goes along with the new CMSSW_14_1_0_pre3_LST_X branch of the CMSSW repo.

This works fine on cgpu-1, but I have to make sure that it also works on the Cornell machines since they have a slightly older nvidia driver.

[ariostas]

With the current setup, the ROCm version fails to link on the Cornell machines. There's an issue with incompatible versions of libstdc++ and I haven't been able to track down why it tries to use different versions.

For lnx4555 we can simply not use Apptainer and change the setup.sh script so that it uses the el9_amd64_gcc12 platform.

For lnx7188, I have tried switching the platform to slc7_amd64_gcc12 (without Apptainer) and also tried using various different container images, but nothing works. The CUDA and CPU versions work fine, though.

So do you guys think it would be fine if I modify the setup.sh so that it uses el9 for lnx4555 and el8 for lnx7188 and cgpu-1, and we stop using Apptainer for lnx4555, and just accept that the ROCm version doesn't compile on lnx7188 (which we can't run anyway)? Maybe @GNiendorf should be the one to comment on this.

[GNiendorf]

@ariostas That sounds fine to me.

[ariostas]

/run standalone
/run cmssw CMSSW_14_1_0_pre3_LST_X

[github-actions]

The PR was built and ran successfully in standalone mode. Here are some of the comparison plots.

The full set of validation and comparison plots can be found here.

Here is a timing comparison:

   Evt    Hits       MD       LS      T3       T5       pLS       pT5      pT3      TC       Reset    Event     Short             Rate
   avg     42.9    321.8    121.5     69.6     93.3    543.6    130.2    156.8    103.6      1.9    1585.2     998.7+/- 268.8     432.4   explicit_cache[s=4] (master)
   avg     43.0    324.4    123.0     70.7     93.6    550.4    129.5    158.1    104.3      1.9    1599.1    1005.6+/- 264.9     438.4   explicit_cache[s=4] (this PR)

[GNiendorf]

Looks good, will merge after CI finishes.

[github-actions]

The PR was built and ran successfully with CMSSW. Here are some plots.

OOTB All Tracks

The full set of validation and comparison plots can be found here.