Madgraph5 GPU development
Madgraph4gpu dev meeting Tue 19.03.2024
https://indico.cern.ch/event/1355148/
# Present: SR, SH, OM, AV (notes), CV, TC
## Round table
### SR
SR: 1, developed the channelid, backported to codegen.
All working except one thing, something strange executing fortran from fortran side, to be debugged.
2, opened the post, have a shortlist, doing interviews next week, maybe next week can give a name.
(SH: but not much to do with madgraph?
SR: madgraph but also related things, generic parallel programming)
(OM: how many candidates? 50, then 30 left, interviewing 5)
### SH
SH: want to know which machine I can destroy.
AV: keep 90, can remove 80 or 03, will have a look in the next few days.
SR: wider context is to upgrade the A100 machines.
### AV
AV: presents the slides, long discussion
OM: Generate g g > h > b b~
!! OM: actually these "BSM" parameters are probably also in SM processes
SR at some point we should define the content of the release, we are going beyond SM LO
AV/OM no important that we fix the "BSM" including 825/826
SR are you done with your scripts?
AV: essentially documented, just need to put that in a script
AV: but if you want to look at 825/826, let me know, and I check SMEFT instead
SR/OM ok may have a look
### OM
OM: nta
### NN
NN: can merge - AV will merge
NN: looking at kernel launchers, no equivalence in sycl
AV: is this what you said about cuda streams?
NN: no unrelated
### TC
TC: ntr
### CV
CV: ntr
## AOB
Next meeting: Thu 04 Apr (Tue 02 AV on a flight, SR not convenient either)