Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[prof] regenerate CODEGEN patch from gg_tt.mad including the madgraph…
…5#969 performance improvements in sample_get_x in dsample.f This includes - simplify the code for xbin_min and xbin_max (remove dead code) - cache xbin_min for xmin=0 and xbin_max for xmax=1 - comment out dead if/then branches (for warnings that were already commented out) - [NOT YET INCLUDED! I forgot this...] optionally skip xbin checks if CUDACPP_RUNTIME_SKIPXBINCHECKS is set The only files that still need to be patched are - 4 in patch.common: Source/makefile, Source/genps.inc, Source/dsample.f, SubProcesses/makefile - 4 in patch.P1: auto_dsig1.f, auto_dsig.f, driver.f, matrix1.f ./CODEGEN/generateAndCompare.sh gg_tt --mad --nopatch git diff --no-ext-diff -R gg_tt.mad/Source/makefile gg_tt.mad/Source/genps.inc gg_tt.mad/SubProcesses/makefile gg_tt.mad/Source/dsample.f > CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/MG5aMC_patches/PROD/patch.common git diff --no-ext-diff -R gg_tt.mad/SubProcesses/P1_gg_ttx/auto_dsig1.f gg_tt.mad/SubProcesses/P1_gg_ttx/auto_dsig.f gg_tt.mad/SubProcesses/P1_gg_ttx/driver.f gg_tt.mad/SubProcesses/P1_gg_ttx/matrix1.f > CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/MG5aMC_patches/PROD/patch.P1 git checkout gg_tt.mad (Later checked that regenerating gg_tt.mad is ok)
- Loading branch information