LHE file mismatch fortran vs cudacpp in tlau (for multi backend builds, possibly for normal builds too) - also: "fail to reach target" #993

valassi · 2024-09-11T05:41:17Z

I just had a quick first look at lhe event files produced in multi backend tests.

In dy+3j just a few occasional events are the same, but most are different. The problem is also that the weight is completely different. This is from the 500 events in PR #948

diff events_fortran.lhe events_cppnone.lhe | more
450c450
< #  Integrated weight (pb)  :       142.38107118268414
---
> #  Integrated weight (pb)  :       3199.366020218372
455c455
< 1.423811e+02 6.637245e-01 1.423811e+02 0
---
> 3.199366e+03 1.661125e+01 3.199366e+03 0

Will need to understand this.

The text was updated successfully, but these errors were encountered:

…ith 500 events instead of 100 Note that the lhe events file are different madgraph5#993 CUDACPP_RUNTIME_DISABLEFPE=1 ./tlau/lauX.sh -ALL pp_dy3j.mad -fromgridpack

valassi · 2024-09-14T14:29:59Z

I added 'fail to reach target' in the title, I saw this in dy+4j: what is this? And is this related to what I observed above in dd+3j?

In dy+4jets

[avalassi@itscrd90 gcc11/usr] /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/tlau/fromgridpacks> ls -l pp_dy4j.mad/*/*
-rw-r--r--. 1 avalassi zg  19851 Sep 14 14:36 pp_dy4j.mad/cpp512y/events.lhe.gz
-rw-r--r--. 1 avalassi zg 114963 Sep 14 14:36 pp_dy4j.mad/cpp512y/output.txt
-rw-r--r--. 1 avalassi zg  19851 Sep 14 15:06 pp_dy4j.mad/cpp512z/events.lhe.gz
-rw-r--r--. 1 avalassi zg 114963 Sep 14 15:06 pp_dy4j.mad/cpp512z/output.txt
-rw-r--r--. 1 avalassi zg  19851 Sep 14 14:07 pp_dy4j.mad/cppavx2/events.lhe.gz
-rw-r--r--. 1 avalassi zg 114963 Sep 14 14:07 pp_dy4j.mad/cppavx2/output.txt
-rw-r--r--. 1 avalassi zg  19851 Sep 14 12:42 pp_dy4j.mad/cppnone/events.lhe.gz
-rw-r--r--. 1 avalassi zg 114986 Sep 14 12:42 pp_dy4j.mad/cppnone/output.txt
-rw-r--r--. 1 avalassi zg  19851 Sep 14 13:37 pp_dy4j.mad/cppsse4/events.lhe.gz
-rw-r--r--. 1 avalassi zg 114978 Sep 14 13:37 pp_dy4j.mad/cppsse4/output.txt
-rw-r--r--. 1 avalassi zg  19851 Sep 14 11:00 pp_dy4j.mad/cuda/events.lhe.gz
-rw-r--r--. 1 avalassi zg 115208 Sep 14 11:00 pp_dy4j.mad/cuda/output.txt
-rw-r--r--. 1 avalassi zg  35494 Sep 14 10:46 pp_dy4j.mad/fortran/events.lhe.gz
-rw-r--r--. 1 avalassi zg 104982 Sep 14 10:46 pp_dy4j.mad/fortran/output.txt

egrep '(events :|fail)' pp_dy4j.mad/*/*txt
pp_dy4j.mad/cpp512y/output.txt:INFO: fail to reach target 100 
pp_dy4j.mad/cpp512y/output.txt:     Nb of events :  44
pp_dy4j.mad/cpp512z/output.txt:INFO: fail to reach target 100 
pp_dy4j.mad/cpp512z/output.txt:     Nb of events :  44
pp_dy4j.mad/cppavx2/output.txt:INFO: fail to reach target 100 
pp_dy4j.mad/cppavx2/output.txt:     Nb of events :  44
pp_dy4j.mad/cppnone/output.txt:INFO: fail to reach target 100 
pp_dy4j.mad/cppnone/output.txt:     Nb of events :  44
pp_dy4j.mad/cppsse4/output.txt:INFO: fail to reach target 100 
pp_dy4j.mad/cppsse4/output.txt:     Nb of events :  44
pp_dy4j.mad/cuda/output.txt:INFO: fail to reach target 100 
pp_dy4j.mad/cuda/output.txt:     Nb of events :  44
pp_dy4j.mad/fortran/output.txt:     Nb of events :  100

…crd90 (with 100 events) CUDACPP_RUNTIME_DISABLEFPE=1 ./tlau/lauX.sh -ALL pp_dy4j.mad -fromgridpack Note that the lhe event files are different madgraph5#993 -rw-r--r--. 1 avalassi zg 19851 Sep 14 14:36 pp_dy4j.mad/cpp512y/events.lhe.gz -rw-r--r--. 1 avalassi zg 114963 Sep 14 14:36 pp_dy4j.mad/cpp512y/output.txt -rw-r--r--. 1 avalassi zg 19851 Sep 14 15:06 pp_dy4j.mad/cpp512z/events.lhe.gz -rw-r--r--. 1 avalassi zg 114963 Sep 14 15:06 pp_dy4j.mad/cpp512z/output.txt -rw-r--r--. 1 avalassi zg 19851 Sep 14 14:07 pp_dy4j.mad/cppavx2/events.lhe.gz -rw-r--r--. 1 avalassi zg 114963 Sep 14 14:07 pp_dy4j.mad/cppavx2/output.txt -rw-r--r--. 1 avalassi zg 19851 Sep 14 12:42 pp_dy4j.mad/cppnone/events.lhe.gz -rw-r--r--. 1 avalassi zg 114986 Sep 14 12:42 pp_dy4j.mad/cppnone/output.txt -rw-r--r--. 1 avalassi zg 19851 Sep 14 13:37 pp_dy4j.mad/cppsse4/events.lhe.gz -rw-r--r--. 1 avalassi zg 114978 Sep 14 13:37 pp_dy4j.mad/cppsse4/output.txt -rw-r--r--. 1 avalassi zg 19851 Sep 14 11:00 pp_dy4j.mad/cuda/events.lhe.gz -rw-r--r--. 1 avalassi zg 115208 Sep 14 11:00 pp_dy4j.mad/cuda/output.txt -rw-r--r--. 1 avalassi zg 35494 Sep 14 10:46 pp_dy4j.mad/fortran/events.lhe.gz -rw-r--r--. 1 avalassi zg 104982 Sep 14 10:46 pp_dy4j.mad/fortran/output.txt egrep '(events :|fail)' pp_dy4j.mad/*/*txt pp_dy4j.mad/cpp512y/output.txt:INFO: fail to reach target 100 pp_dy4j.mad/cpp512y/output.txt: Nb of events : 44 pp_dy4j.mad/cpp512z/output.txt:INFO: fail to reach target 100 pp_dy4j.mad/cpp512z/output.txt: Nb of events : 44 pp_dy4j.mad/cppavx2/output.txt:INFO: fail to reach target 100 pp_dy4j.mad/cppavx2/output.txt: Nb of events : 44 pp_dy4j.mad/cppnone/output.txt:INFO: fail to reach target 100 pp_dy4j.mad/cppnone/output.txt: Nb of events : 44 pp_dy4j.mad/cppsse4/output.txt:INFO: fail to reach target 100 pp_dy4j.mad/cppsse4/output.txt: Nb of events : 44 pp_dy4j.mad/cuda/output.txt:INFO: fail to reach target 100 pp_dy4j.mad/cuda/output.txt: Nb of events : 44 pp_dy4j.mad/fortran/output.txt: Nb of events : 100

…he number of events (showing madgraph5#993) ./parseGridpackLogs.sh pp_dy4j.mad | tee pp_dy4j.mad/summary.txt

…with 500 events instead of 100) CUDACPP_RUNTIME_DISABLEFPE=1 ./tlau/lauX.sh -ALL pp_dy4j.mad -fromgridpack Note that the lhe event files are still different madgraph5#993 ls -l tlau/fromgridpacks/pp_dy4j.mad/*/* -rw-r--r--. 1 avalassi zg 62245 Sep 15 15:42 tlau/fromgridpacks/pp_dy4j.mad/cpp512y/events.lhe.gz -rw-r--r--. 1 avalassi zg 298970 Sep 15 15:42 tlau/fromgridpacks/pp_dy4j.mad/cpp512y/output.txt -rw-r--r--. 1 avalassi zg 62245 Sep 15 17:44 tlau/fromgridpacks/pp_dy4j.mad/cpp512z/events.lhe.gz -rw-r--r--. 1 avalassi zg 298970 Sep 15 17:44 tlau/fromgridpacks/pp_dy4j.mad/cpp512z/output.txt -rw-r--r--. 1 avalassi zg 62245 Sep 15 13:45 tlau/fromgridpacks/pp_dy4j.mad/cppavx2/events.lhe.gz -rw-r--r--. 1 avalassi zg 298973 Sep 15 13:45 tlau/fromgridpacks/pp_dy4j.mad/cppavx2/output.txt -rw-r--r--. 1 avalassi zg 62245 Sep 15 07:36 tlau/fromgridpacks/pp_dy4j.mad/cppnone/events.lhe.gz -rw-r--r--. 1 avalassi zg 299015 Sep 15 07:36 tlau/fromgridpacks/pp_dy4j.mad/cppnone/output.txt -rw-r--r--. 1 avalassi zg 62245 Sep 15 11:39 tlau/fromgridpacks/pp_dy4j.mad/cppsse4/events.lhe.gz -rw-r--r--. 1 avalassi zg 298996 Sep 15 11:39 tlau/fromgridpacks/pp_dy4j.mad/cppsse4/output.txt -rw-r--r--. 1 avalassi zg 62245 Sep 15 00:06 tlau/fromgridpacks/pp_dy4j.mad/cuda/events.lhe.gz -rw-r--r--. 1 avalassi zg 299215 Sep 15 00:06 tlau/fromgridpacks/pp_dy4j.mad/cuda/output.txt -rw-r--r--. 1 avalassi zg 30360 Sep 14 23:21 tlau/fromgridpacks/pp_dy4j.mad/fortran/events.lhe.gz -rw-r--r--. 1 avalassi zg 271132 Sep 14 23:21 tlau/fromgridpacks/pp_dy4j.mad/fortran/output.txt egrep '(events :|fail)' tlau/fromgridpacks/pp_dy4j.mad/*/* tlau/fromgridpacks/pp_dy4j.mad/cpp512y/output.txt:INFO: fail to reach target 500 tlau/fromgridpacks/pp_dy4j.mad/cpp512y/output.txt: Nb of events : 195 tlau/fromgridpacks/pp_dy4j.mad/cpp512z/output.txt:INFO: fail to reach target 500 tlau/fromgridpacks/pp_dy4j.mad/cpp512z/output.txt: Nb of events : 195 tlau/fromgridpacks/pp_dy4j.mad/cppavx2/output.txt:INFO: fail to reach target 500 tlau/fromgridpacks/pp_dy4j.mad/cppavx2/output.txt: Nb of events : 195 tlau/fromgridpacks/pp_dy4j.mad/cppnone/output.txt:INFO: fail to reach target 500 tlau/fromgridpacks/pp_dy4j.mad/cppnone/output.txt: Nb of events : 195 tlau/fromgridpacks/pp_dy4j.mad/cppsse4/output.txt:INFO: fail to reach target 500 tlau/fromgridpacks/pp_dy4j.mad/cppsse4/output.txt: Nb of events : 195 tlau/fromgridpacks/pp_dy4j.mad/cuda/output.txt:INFO: fail to reach target 500 tlau/fromgridpacks/pp_dy4j.mad/cuda/output.txt: Nb of events : 195 tlau/fromgridpacks/pp_dy4j.mad/fortran/output.txt:INFO: fail to reach target 500 tlau/fromgridpacks/pp_dy4j.mad/fortran/output.txt: Nb of events : 81

valassi self-assigned this Sep 11, 2024

valassi mentioned this issue Sep 11, 2024

WIP studies and fixes with generate_events #948

Draft

valassi changed the title ~~LHE file mismatch fortran vs cudacpp in tlau (for multi backend builds, possibly for normal builds too)~~ LHE file mismatch fortran vs cudacpp in tlau (for multi backend builds, possibly for normal builds too) - also: "fail to reach target" Sep 14, 2024

valassi added a commit to valassi/madgraph4gpu that referenced this issue Sep 14, 2024

[grid] in tlau/fromgridpacks, recreate pp_dy4j.mad/summary.txt with t…

cf12395

…he number of events (showing madgraph5#993) ./parseGridpackLogs.sh pp_dy4j.mad | tee pp_dy4j.mad/summary.txt

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

LHE file mismatch fortran vs cudacpp in tlau (for multi backend builds, possibly for normal builds too) - also: "fail to reach target" #993

LHE file mismatch fortran vs cudacpp in tlau (for multi backend builds, possibly for normal builds too) - also: "fail to reach target" #993

valassi commented Sep 11, 2024

valassi commented Sep 14, 2024

LHE file mismatch fortran vs cudacpp in tlau (for multi backend builds, possibly for normal builds too) - also: "fail to reach target" #993

LHE file mismatch fortran vs cudacpp in tlau (for multi backend builds, possibly for normal builds too) - also: "fail to reach target" #993

Comments

valassi commented Sep 11, 2024

valassi commented Sep 14, 2024