add HEFT gg_bb to the repo (with additional fixes for underflow FPEs) by valassi · Pull Request #832 · madgraph5/madgraph4gpu

valassi · 2024-04-05T14:29:17Z

This is a WIP PR that logically follows the SUSY and SMEFT PRs. In particular it includes #632 which includes #824.

The main change is that I added heft_gg_bb to the repo (and am removing heft_gg_h).

Concerning heft_gg_bb, this now includes 4 diagrams (3 SM and also 1 HEFT) after fixing #828. This essentially means adding one diagram even if it is suppressed.

Now, the fact that the contribution is suppressed is probably why I start getting FPE underflows #831. I have now also fixed this (essentially by flushing to 0 all values of jamp_sv below E-18, so that their square is still above E-36 and fits in a float.

This is still WIP as I am running a few more tests.

…=1" to generate heft_gg_bb with 4 diagrams (3 SM + 1 HEFT) instead of only 3

…eft_gg_bb CUDACPP_RUNTEST_DUMPEVENTS=1 ./runTest.exe cp ../../test/ref/dump_CPUTest.Sigma_heft_gg_bbx.txt ../../../CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/test/ref/ git add ../../test/ref/dump_CPUTest.Sigma_heft_gg_bbx.txt ../../../CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/test/ref/

…issues for f ./tput/teeThroughputX.sh -flt -hrd -makej -heftggbb -makeclean STARTED AT Thu Apr 4 06:15:25 PM CEST 2024 ENDED AT Thu Apr 4 06:23:19 PM CEST 2024 The problem is again a FPE madgraph5#831 cmpExe /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/heft_gg_bb.mad/SubProcesses/P1_gg_bbx/build.none_f_inl0_hrd0/check.exe --common -p 2 64 2 cmpExe /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/heft_gg_bb.mad/SubProcesses/P1_gg_bbx/build.none_f_inl0_hrd0/fcheck.exe 2 64 2 Floating Point Exception (CPU) Avg ME (C++/C++) = Avg ME (F77/C++) = 4.3135739049175754 File "<string>", line 1 me1=; me2=4.3135739049175754; reldif=abs((me2-me1)/me1); print('Relative difference =', reldif); ok = reldif <= 5E-3; print ( '%s (relative difference %s 5E-3)' % ( ('OK','<=') if ok else ('ERROR','>') ) ) ^ SyntaxError: invalid syntax

…this is needed to simplify FPE handling in runTest.exe for madgraph5#831) (NB: the reason why main is added to testxxx.cc instead of runTest.cc is that it is easier to define all FPE signal handlers in testxxx.cc)

…n() function, for all tests in runTest.exe As expected, now runTest.exe throws an FPE also in MadgraphTest madgraph5#831 [----------] 1 test from SIGMA_HEFT_GG_BBX_CPU/MadgraphTest [ RUN ] SIGMA_HEFT_GG_BBX_CPU/MadgraphTest.CompareMomentaAndME/0 INFO: Opening reference file ../../test/ref/dump_CPUTest.Sigma_heft_gg_bbx.txt INFO: The application is built for nehalem (SSE4.2) and the host supports it Thread 1 "runTest.exe" received signal SIGFPE, Arithmetic exception. 0x0000000000410807 in mg5amcCpu::calculate_wavefunctions(int, float const*, float const*, float*, float __vector(4)*, int) (ihel=0, allmomenta=0x67e480, allcouplings=0x1492ec0, allMEs=0x683700, jamp2_sv=0x0, ievt00=4) at CPPProcess.cc:434 434 fptype2_sv deltaMEs2 = ( jampRi_sv * ztempR_sv + jampIi_sv * ztempI_sv ); Missing separate debuginfos, use: dnf debuginfo-install glibc-2.34-60.el9.x86_64 libgcc-11.3.1-4.3.el9.alma.x86_64 libgomp-11.3.1-4.3.el9.alma.x86_64 libstdc++-11.3.1-4.3.el9.alma.x86_64 nvidia-driver-cuda-libs-530.30.02-1.el9.x86_64 (gdb) where 0 0x0000000000410807 in mg5amcCpu::calculate_wavefunctions(int, float const*, float const*, float*, float __vector(4)*, int) (ihel=0, allmomenta=0x67e480, allcouplings=0x1492ec0, allMEs=0x683700, jamp2_sv=0x0, ievt00=4) at CPPProcess.cc:434 1 0x0000000000410f38 in mg5amcCpu::sigmaKin_getGoodHel (allmomenta=0x67e480, allcouplings=0x1492ec0, allMEs=0x683700, isGoodHel=0x1494740, nevt=256) at CPPProcess.cc:787 2 0x000000000041e979 in mg5amcCpu::MatrixElementKernelHost::computeGoodHelicities (this=0x7fffffffcbe0) at MatrixElementKernels.cc:70 3 0x0000000000456cb6 in CPUTest::runSigmaKin (this=0x67c170, iiter=0) at runTest.cc:103

…(flush-to-zero jamp values whose square is below FLT_MIN) However now runTest.exe may still fail as follows (should avoid using relDelta!) [----------] 1 test from SIGMA_HEFT_GG_BBX_CPU/MadgraphTest [ RUN ] SIGMA_HEFT_GG_BBX_CPU/MadgraphTest.CompareMomentaAndME/0 INFO: Opening reference file ../../test/ref/dump_CPUTest.Sigma_heft_gg_bbx.txt INFO: The application is built for nehalem (SSE4.2) and the host supports it [New Thread 0x7fffedeec000 (LWP 2405183)] [New Thread 0x7fffed6eb000 (LWP 2405184)] [New Thread 0x7fffeceea000 (LWP 2405185)] Thread 1 "runTest.exe" received signal SIGFPE, Arithmetic exception. 0x0000000000453c82 in MadgraphTest_CompareMomentaAndME_Test::TestBody (this=0x148f850) at /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/heft_gg_bb.sa/SubProcesses/P1_Sigma_heft_gg_bbx/MadgraphTest.h:298 298 const fptype relDelta = fabs( ( pMadg - pOrig ) / pOrig ); Missing separate debuginfos, use: dnf debuginfo-install glibc-2.34-60.el9.x86_64 libgcc-11.3.1-4.3.el9.alma.x86_64 libgomp-11.3.1-4.3.el9.alma.x86_64 libstdc++-11.3.1-4.3.el9.alma.x86_64 nvidia-driver-cuda-libs-530.30.02-1.el9.x86_64 (gdb) where 0 0x0000000000453c82 in MadgraphTest_CompareMomentaAndME_Test::TestBody (this=0x148f850) at /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/heft_gg_bb.sa/SubProcesses/P1_Sigma_heft_gg_bbx/MadgraphTest.h:298

… reference, instead of comparing relative delta to tolerance (computing relDelta may lead to FPEs)

…emove -gtest_main (this is needed to simplify FPE handling in runTest.exe for madgraph5#831) (NB: the reason why main is added to testxxx.cc instead of runTest.cc is that it is easier to define all FPE signal handlers in testxxx.cc)

…ndling, via the main() function, for all tests in runTest.exe As expected, now heft_gg_bb runTest.exe throws an FPE also in MadgraphTest madgraph5#831

…w FPE madgraph5#831 (flush-to-zero jamp values whose square is below FLT_MIN) However now runTest.exe may still fail with FPEs in relDelta (should avoid using relDelta!)

… to tolerance times reference, instead of comparing relative delta to tolerance (computing relDelta may lead to FPEs)

: all ok now ./tput/teeThroughputX.sh -flt -hrd -makej -heftggbb -makeclean

…but not for f ./teeMadX.sh -heftggbb -mix -d +10x 24 /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/tmad/logs_heftggbb_mad/log_heftggbb_mad_d_inl0_hrd0.txt 1 /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/tmad/logs_heftggbb_mad/log_heftggbb_mad_f_inl0_hrd0.txt 24 /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/tmad/logs_heftggbb_mad/log_heftggbb_mad_m_inl0_hrd0.txt The problem is a mismatch of the LHE file between fortran and cpp madgraph5#833 *** (2-none) Compare MADEVENT_CPP x1 events.lhe to MADEVENT_FORTRAN events.lhe reference (including colors and helicities) *** ERROR! events.lhe.cpp.1 and events.lhe.ref.1 differ! diff /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/heft_gg_bb.mad/SubProcesses/P1_gg_bbx/events.lhe.cpp.1 /data/avalassi/GPU2023/madgraph4gpuX/epochX/cudacpp/heft_gg_bb.mad/SubProcesses/P1_gg_bbx/events.lhe.ref.1 | head -20 6206,6207c6206,6207 < 21 -1 0 0 502 503 -0.00000000000E+00 -0.00000000000E+00 -0.59936081260E+01 0.59936081260E+01 0.00000000000E+00 0. -1. < 5 1 1 2 501 0 0.45273385612E+02 -0.31131305296E+02 0.47763304676E+03 0.48080583916E+03 0.47000000000E+01 0. 1. --- > 21 -1 0 0 502 503 -0.00000000000E+00 -0.00000000000E+00 -0.59936081260E+01 0.59936081260E+01 0.00000000000E+00 0. 1. > 5 1 1 2 501 0 0.45273385612E+02 -0.31131305296E+02 0.47763304676E+03 0.48080583916E+03 0.47000000000E+01 0. -1. 8306,8307c8306,8307 < 21 -1 0 0 502 503 -0.00000000000E+00 -0.00000000000E+00 -0.23857997239E+02 0.23857997239E+02 0.00000000000E+00 0. 1. < 5 1 1 2 501 0 -0.34843521722E+02 0.35239303629E+02 0.13219496682E+02 0.51504607743E+02 0.47000000000E+01 0. -1. --- > 21 -1 0 0 502 503 -0.00000000000E+00 -0.00000000000E+00 -0.23857997239E+02 0.23857997239E+02 0.00000000000E+00 0. -1. > 5 1 1 2 501 0 -0.34843521722E+02 0.35239303629E+02 0.13219496682E+02 0.51504607743E+02 0.47000000000E+01 0. 1. 9606,9619d9605 < 4 1 1E-03 0.1250139E+03 0.7546771E-02 0.1235066E+00 < 21 -1 0 0 503 502 0.00000000000E+00 0.00000000000E+00 0.94948250004E+03 0.94948250004E+03 0.00000000000E+00 0. 1. < 21 -1 0 0 502 503 -0.00000000000E+00 -0.00000000000E+00 -0.41149990002E+01 0.41149990002E+01 0.00000000000E+00 0. -1. < 5 1 1 2 501 0 -0.96459450317E+01 -0.34409175043E+02 0.83136584965E+02 0.90613560477E+02 0.47000000000E+01 0. -1. < -5 1 1 2 0 501 0.96459450317E+01 0.34409175043E+02 0.86223091608E+03 0.86298393857E+03 0.47000000000E+01 0. 1. < <mgrwt> < <rscale> 0 0.12501391E+03</rscale>

…s a=0 but returns a failure status code!

…d underflow FPE madgraph5#831... note that now FPTYPE=f is ok but FPTYPE=m is not

…t only d/f (now 102 logs total, were 94)

…fter initial patch for FPEs madgraph5#831: mix fptype fails, as expected ./tput/teeThroughputX.sh -mix -hrd -makej -heftggbb -makeclean

…m scalar (still fails for vector)

…m vector too

…sues in the patch to avoid underflow FPE madgraph5#831 (fix handling of FPTYPE=f and add FPTYPE=m)

STARTED AT Sat Apr 13 08:57:22 PM CEST 2024 ./tput/teeThroughputX.sh -mix -hrd -makej -eemumu -ggtt -ggttg -ggttgg -gqttq -ggttggg -makeclean ENDED(1) AT Sat Apr 13 09:23:53 PM CEST 2024 [Status=0] ./tput/teeThroughputX.sh -flt -hrd -makej -eemumu -ggtt -ggttgg -inlonly -makeclean ENDED(2) AT Sat Apr 13 09:34:23 PM CEST 2024 [Status=0] ./tput/teeThroughputX.sh -makej -eemumu -ggtt -ggttg -gqttq -ggttgg -ggttggg -flt -bridge -makeclean ENDED(3) AT Sat Apr 13 09:42:52 PM CEST 2024 [Status=0] ./tput/teeThroughputX.sh -eemumu -ggtt -ggttgg -flt -rmbhst ENDED(4) AT Sat Apr 13 09:45:01 PM CEST 2024 [Status=0] ./tput/teeThroughputX.sh -eemumu -ggtt -ggttgg -flt -curhst ENDED(5) AT Sat Apr 13 09:46:08 PM CEST 2024 [Status=0] ./tput/teeThroughputX.sh -eemumu -ggtt -ggttgg -flt -common ENDED(6) AT Sat Apr 13 09:48:20 PM CEST 2024 [Status=0] ./tput/teeThroughputX.sh -mix -hrd -makej -susyggtt -susyggt1t1 -smeftggtttt -heftggbb -makeclean ENDED(7) AT Sat Apr 13 10:00:27 PM CEST 2024 [Status=0]

STARTED AT Sat Apr 13 10:00:27 PM CEST 2024 (SM tests) ENDED(1) AT Sun Apr 14 01:29:37 AM CEST 2024 [Status=0] (BSM tests) ENDED(1) AT Sun Apr 14 01:36:02 AM CEST 2024 [Status=0] 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_eemumu_mad/log_eemumu_mad_d_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_eemumu_mad/log_eemumu_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_eemumu_mad/log_eemumu_mad_m_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttggg_mad/log_ggttggg_mad_d_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttggg_mad/log_ggttggg_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttggg_mad/log_ggttggg_mad_m_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttgg_mad/log_ggttgg_mad_d_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttgg_mad/log_ggttgg_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttgg_mad/log_ggttgg_mad_m_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttg_mad/log_ggttg_mad_d_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttg_mad/log_ggttg_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggttg_mad/log_ggttg_mad_m_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggtt_mad/log_ggtt_mad_d_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggtt_mad/log_ggtt_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_ggtt_mad/log_ggtt_mad_m_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_gqttq_mad/log_gqttq_mad_d_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_gqttq_mad/log_gqttq_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_gqttq_mad/log_gqttq_mad_m_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_heftggbb_mad/log_heftggbb_mad_d_inl0_hrd0.txt 1 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_heftggbb_mad/log_heftggbb_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_heftggbb_mad/log_heftggbb_mad_m_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_smeftggtttt_mad/log_smeftggtttt_mad_d_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_smeftggtttt_mad/log_smeftggtttt_mad_f_inl0_hrd0.txt 20 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_smeftggtttt_mad/log_smeftggtttt_mad_m_inl0_hrd0.txt 0 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_susyggt1t1_mad/log_susyggt1t1_mad_d_inl0_hrd0.txt 0 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_susyggt1t1_mad/log_susyggt1t1_mad_f_inl0_hrd0.txt 0 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_susyggt1t1_mad/log_susyggt1t1_mad_m_inl0_hrd0.txt 0 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_susyggtt_mad/log_susyggtt_mad_d_inl0_hrd0.txt 0 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_susyggtt_mad/log_susyggtt_mad_f_inl0_hrd0.txt 0 /data/avalassi/GPU2024/madgraph4gpuX/epochX/cudacpp/tmad/logs_susyggtt_mad/log_susyggtt_mad_m_inl0_hrd0.txt

Revert "[heft] rerun 30 tmad tests on itgold91 - all as expected" This reverts commit f0e9980. Revert "[heft] rerun 102 tput tests on itgold91 - all ok" This reverts commit 42c3797.

(Did not rerun tests on AMD GPUs - LUMI in maintenance for one week)

valassi · 2024-04-15T08:35:12Z

Hi @oliviermattelaer I confirm I have finished. This is good to merge for me after your review. Thanks

oliviermattelaer

For me this is good to go,
maybe one detail would be to remove the special case for
'mdl_G__exp__2' and similar but that's ok if you want to keep it.

…handling of mdl_G__exp__2/3 as suggested by Olivier ./CODEGEN/generateAndCompare.sh -q heft_gg_bb ./CODEGEN/generateAndCompare.sh -q susy_gg_tt ./CODEGEN/generateAndCompare.sh -q susy_gg_t1t1 ./CODEGEN/generateAndCompare.sh -q smeft_gg_tttt

…p__2/3 as suggested by Olivier

…ut in SUSY mdl_G__exp__2 is complex again? (mg5amcnlo#89 is not fixed?) ./CODEGEN/generateAndCompare.sh -q heft_gg_bb ./CODEGEN/generateAndCompare.sh -q susy_gg_tt ./CODEGEN/generateAndCompare.sh -q susy_gg_t1t1 ./CODEGEN/generateAndCompare.sh -q smeft_gg_tttt

…without the fix for mg5amcnlo#89 (G^2 is real or complex) in commit 619f5db Will try to regenerate the code again and see what happens

…h for mg5amcnlo#89 - only change is mdl_G__exp__3 becmes complex in SMEFT Running summary below Without Olivier's patch for mg5amcnlo#89 Without special handling of mdl_G__exp__2 and mdl_G__exp__3 in cudacpp: - mdl_G__exp__2 is double in HEFT and SMEFT (correct) - mdl_G__exp__2 is complex in SUSY (incorrect) - mdl_G__exp__3 is complex in SMEFT (incorrect)

…ix for mg5amcnlo#89 (G^2 is real or complex) in commit 619f5db

…or mg5amcnlo#89 - fixes mdl_G__exp__3 in SMEFT (but SUSY is wrong here, WITHOUT The additional cudacpp patch) Running summary below Without Olivier's patch for mg5amcnlo#89 Without special handling of mdl_G__exp__2 and mdl_G__exp__3 in cudacpp: - mdl_G__exp__2 is double in HEFT and SMEFT (correct) - mdl_G__exp__2 is complex in SUSY (incorrect) <--- FIXME - mdl_G__exp__3 is complex in SMEFT (incorrect) <--- FIXME WITH Olivier's patch for mg5amcnlo#89 Without special handling of mdl_G__exp__2 and mdl_G__exp__3 in cudacpp: - mdl_G__exp__2 is double in HEFT and SMEFT (correct) - mdl_G__exp__2 is complex in SUSY (incorrect) <--- FIXME - mdl_G__exp__3 is double in SMEFT (correct) <=== NOW FIXED

…ne for mdl_G__exp__3, now fixed by Olivier)

…amcnlo#89, WITH cudacpp patch for mdl_G__exp__2, WITHOUT cudacpp patch for mdl_G__exp__3 Now all is ok again for both SUSY and SMEFT/HEFT Running summary below Without Olivier's patch for mg5amcnlo#89 Without special handling of mdl_G__exp__2 and mdl_G__exp__3 in cudacpp: - mdl_G__exp__2 is double in HEFT and SMEFT (correct) - mdl_G__exp__2 is complex in SUSY (incorrect) <--- FIXME - mdl_G__exp__3 is complex in SMEFT (incorrect) <--- FIXME WITH Olivier's patch for mg5amcnlo#89 Without special handling of mdl_G__exp__2 and mdl_G__exp__3 in cudacpp: - mdl_G__exp__2 is double in HEFT and SMEFT (correct) - mdl_G__exp__2 is complex in SUSY (incorrect) <--- FIXME - mdl_G__exp__3 is double in SMEFT (correct) <=== NOW FIXED WITH Olivier's patch for mg5amcnlo#89 WITH special handling of mdl_G__exp__2 in cudacpp: Without special handling of mdl_G__exp__3 in cudacpp: - mdl_G__exp__2 is double in HEFT and SMEFT (correct) - mdl_G__exp__2 is double in SUSY (correct) <=== NOW FIXED - mdl_G__exp__3 is double in SMEFT (correct) <=== NOW FIXED

…except for cosmetics in G^3 of smeft_gg_tttt.mad)

valassi · 2024-05-14T07:40:46Z

For me this is good to go, maybe one detail would be to remove the special case for 'mdl_G__exp__2' and similar but that's ok if you want to keep it.

Hi Olivier, thanks a lot.

I had a look at the 'mdl_G__exp__2' and 'mdl_G__exp__3' issue. It seems that your patch fixed 'mdl_G__exp__3' but NOT 'mdl_G__exp__2'. I will add a comment in mg5amcnlo/mg5amcnlo#89.

So anyway here I have removed my cudacpp patch for G^3 but kept it for G^2. I regenerated all code and am realunching the CI. I will merge when the CI is ok.

I will reopen mg5amcnlo/mg5amcnlo#89 so that you can have a look. If/when you create a new patch I will test it in a separate madgraph4gpu PR.

Thanks again Andrea

…und for issue madgraph5#838)

….sha (debug madgraph5#838)

valassi · 2024-05-14T13:14:56Z

There are some issues on Mac see #838, I am trying to debug them and work around them

…for issue madgraph5#838)

…workaround for Mac segfault madgraph5#838

valassi · 2024-05-14T13:27:32Z

My workaround for #838 was ok on ee_mumu. I am applying it on CODEGEN and all processes and relaunching the CI

valassi · 2024-05-14T13:39:41Z

Ok the MacOS tests succeed after disabling testmisc (workaround for #838 - real fixes will be needed in #839 and #840).

All CI tests succeed. I am merging this. Thanks again @oliviermattelaer for the review and the approval.

(I still need to add a comment on mg5amcnlo/mg5amcnlo#89, I will do that now. I got distracted bythe MacOS issues).

…adgraph5#832) into jtmk Fix conflicts: epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/cudacpp.mk epochX/cudacpp/tput/teeThroughputX.sh epochX/cudacpp/tput/throughputX.sh

valassi added 17 commits April 4, 2024 17:57

[heft] in CODEGEN/generateAndCompare.sh, fix madgraph5#828: use "NIW<…

6fb3d10

…=1" to generate heft_gg_bb with 4 diagrams (3 SM + 1 HEFT) instead of only 3

[heft] add heft_gg_bb.sa to the repo

502e32d

[heft] add heft_gg_bb.mad to the repo

7dbabc5

[heft] add heft_gg_bb and remove heft_gg_h from tput scripts

870561a

[heft] in heft_gg_bb MadgraphTest.h, compare delta to tolerance times…

9cbdb94

… reference, instead of comparing relative delta to tolerance (computing relDelta may lead to FPEs)

[heft] in CODEGEN (backport heft_gg_bb) testxxx.cc, add FPE signal ha…

5d4700b

…ndling, via the main() function, for all tests in runTest.exe As expected, now heft_gg_bb runTest.exe throws an FPE also in MadgraphTest madgraph5#831

[heft] in CODEGEN (backport heft_gg_bb CPPProcess.cc), avoid underflo…

fb808aa

…w FPE madgraph5#831 (flush-to-zero jamp values whose square is below FLT_MIN) However now runTest.exe may still fail with FPEs in relDelta (should avoid using relDelta!)

[heft] in CODEGEN (backport heft_gg_bb) MadgraphTest.h, compare delta…

25bd6e7

… to tolerance times reference, instead of comparing relative delta to tolerance (computing relDelta may lead to FPEs)

[heft] in CODEGEN and regenerated heft_gg_bb.sa, fix clang formatting

90854af

[heft] regenerate heft_gg_bb.mad

1e1a2d9

[heft] rerun 4 tput tests for heft_gg_bb after fixing FPEs madgraph5#831

c3f841d

: all ok now ./tput/teeThroughputX.sh -flt -hrd -makej -heftggbb -makeclean

valassi self-assigned this Apr 5, 2024

This was linked to issues Apr 5, 2024

FFS2_0 missing in HEFT gg to bb? #828

Closed

Underflow Floating Point Exception for float and mixed FPTYPE in HEFT gg to bb #831

Closed

valassi added 2 commits April 5, 2024 17:29

[heft] add heft_gg_bb to tmad scripts

d8b635a

valassi mentioned this pull request Apr 5, 2024

LHE file mismatch between fortran and cpp in heft_gg_bb for FPTYPE=f #833

Open

valassi added 7 commits April 5, 2024 19:28

[heft] fix tmad/madX.sh for an unexpected bash issue: (( a = 0 )) set…

ff0c7e7

…s a=0 but returns a failure status code!

[heft] in heft_gg_bb CPPProcess.cc, fix an error in the patch to avoi…

6c648af

…d underflow FPE madgraph5#831... note that now FPTYPE=f is ok but FPTYPE=m is not

[heft] in tput/allTees.sh, run all four BSM processes in d/m/f and no…

de2d593

…t only d/f (now 102 logs total, were 94)

[heft] run 6 tput tests (now including mixed fptype) for heft_gg_bb a…

0047c87

…fter initial patch for FPEs madgraph5#831: mix fptype fails, as expected ./tput/teeThroughputX.sh -mix -hrd -makej -heftggbb -makeclean

[heft] in heft_gg_bb CPPProcess.cc, fix FPE madgraph5#831 for FPTYPE=…

9ac4225

…m scalar (still fails for vector)

[heft] in heft_gg_bb CPPProcess.cc, fix FPE madgraph5#831 for FPTYPE=…

474230c

…m vector too

[heft] in CODEGEN (backport heft_gg_bb CPPProcess.cc), fix various is…

8f8ac48

…sues in the patch to avoid underflow FPE madgraph5#831 (fix handling of FPTYPE=f and add FPTYPE=m)

valassi added 4 commits April 14, 2024 13:18

[heft] go back to itscrd90 logs

12e67c0

Revert "[heft] rerun 30 tmad tests on itgold91 - all as expected" This reverts commit f0e9980. Revert "[heft] rerun 102 tput tests on itgold91 - all ok" This reverts commit 42c3797.

[heft] ** COMPLETE HEFT ** in tput/allTees.sh, add -nomakeclean option

d698668

(Did not rerun tests on AMD GPUs - LUMI in maintenance for one week)

oliviermattelaer approved these changes Apr 30, 2024

View reviewed changes

valassi added 10 commits May 14, 2024 08:49

[heft] in CODEGEN, remove (comment out) special handling of mdl_G__ex…

91fba10

…p__2/3 as suggested by Olivier

[heft] temporarely move mg5amcnlo back to the previous gpucpp commit …

114cb84

…without the fix for mg5amcnlo#89 (G^2 is real or complex) in commit 619f5db Will try to regenerate the code again and see what happens

[heft] upgrade again mg5amcnlo to the latest gpucpp commit with the f…

c778ffd

…ix for mg5amcnlo#89 (G^2 is real or complex) in commit 619f5db

[heft] in CODEGEN, add back special handling of mdl_G__exp__2 (but no…

7e7ab27

…ne for mdl_G__exp__3, now fixed by Olivier)

[heft] ** COMPLETE HEFT AGAIN ** regenerate all processes (no change …

7a25b55

…except for cosmetics in G^3 of smeft_gg_tttt.mad)

valassi mentioned this pull request May 14, 2024

Segfault in testmisc test on MacOS/ARM (the same code succeds on MacOS/x86_64) - disable it as a workaround #838

Closed

valassi added 2 commits May 14, 2024 15:01

[heft] in ee_mumu.mad cudacpp.mk, disable testmisc on Darwin (workaro…

f5459bb

…und for issue madgraph5#838)

[heft] in CI c-cpp.yml, print date and github.event.pull_request.head…

12dba37

….sha (debug madgraph5#838)

valassi added 2 commits May 14, 2024 15:19

[heft] in CODEGEN cudacpp.mk, disable testmisc on Darwin (workaround …

089bc7a

…for issue madgraph5#838)

[heft] ** COMPLETE HEFT AGAIN2 ** regenerate all processes including …

1c23d24

…workaround for Mac segfault madgraph5#838

valassi linked an issue May 14, 2024 that may be closed by this pull request

Segfault in testmisc test on MacOS/ARM (the same code succeds on MacOS/x86_64) - disable it as a workaround #838

Closed

valassi merged commit cb68bce into madgraph5:master May 14, 2024

valassi mentioned this pull request May 14, 2024

G^2 is sometimes handle as complex and sometimes as real mg5amcnlo/mg5amcnlo#89

Closed

valassi mentioned this pull request May 14, 2024

Jorgen's makefiles with separate builds for CUDA, HIP and C++ #798

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add HEFT gg_bb to the repo (with additional fixes for underflow FPEs)#832

add HEFT gg_bb to the repo (with additional fixes for underflow FPEs)#832
valassi merged 160 commits intomadgraph5:masterfrom
valassi:heft

valassi commented Apr 5, 2024

Uh oh!

valassi commented Apr 15, 2024

Uh oh!

oliviermattelaer left a comment

Uh oh!

valassi commented May 14, 2024

Uh oh!

valassi commented May 14, 2024

Uh oh!

valassi commented May 14, 2024

Uh oh!

valassi commented May 14, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

valassi commented Apr 5, 2024

Uh oh!

valassi commented Apr 15, 2024

Uh oh!

oliviermattelaer left a comment

Choose a reason for hiding this comment

Uh oh!

valassi commented May 14, 2024

Uh oh!

valassi commented May 14, 2024

Uh oh!

valassi commented May 14, 2024

Uh oh!

valassi commented May 14, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants