…6 builds #904 (disabling OMP only for clang16; add -no-pie for fcheck_cpp.exe)

…7 builds #904 (disable OMP also for clang17)

…move link-time -no-pie, add compiler-time -fPIC to fortran

…nd.h (BSD license) to detect when running on valgrind #906

This is needed as part of the fixes for runTest.exe #903, preliminary to #896

Note: the header as-is is copied from /cvmfs/sft.cern.ch/lcg/releases/valgrind/3.23.0-24262/x86_64-el9-gcc11-opt/
(except for the inclusion of "clang-format off" directives)

See https://valgrind.org/docs/manual/manual-core-adv.html#manual-core-adv.clientreq

…rehensive fixes and debug printouts for bug #903 (recursive iteration, stack overflow, segfault etc)

…target (address sanitizer #207), but keep it commented out

…lerances when running on valgrind #906

Also allow tan(x)=-inf if ctan(x)=+inf and viceversa when running on valgrind #906

… (I had forgotten it enabled)

…st.cc, testxxx.cc: simplify gtest templates, remove cudaDeviceReset to fix #907, complete preparation of two-test infrastructure #896

More in detail:
- move to the simplest "TEST(" use case of Google tests in MadgraphTest.h and runTest.cc (remove unnecessary levels of templating)
- move gpuDeviceReset() to an atexit function of main in testxxx and comment it out anyway, to fix the segfaults #907
  (eventually it may be necessary to remove all CUDA API calls from destructors, if ever we need to put this back in)
- in runTest.cc, complete a proff of concept for adding two separate tests (without/with multichannel #896)

Fix some clang formatting issues with respect to the last gg_tt.mad

…ng PR #905, constexpr_math.h PR #908 and runTest/cudaDeviceReset PR #909

Add valgrind.h and its symlink in the repo for gg_tt.mad

The new runTest.cc template now has a (commented out) proof of concept for including two tests (with/without multichannel) #896, I will resume from there

After building bldall, the following succeeds
for bck in none sse4 avx2 512y 512z cuda; do echo $bck; ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done

This instead is crashing (again?) for some AVX values
for bck in none sse4 avx2 512y 512z cuda; do echo $bck; valgrind ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done
On closer inspection, this is because valgrind does not support AVX512, so this is ok

…th/without multichannel #896 into the latest regenerated with fixes

Revert "[june24] in gg_tt.mad, temporarely go back to the last code regeneration, removing the attempts to add two tests #896"
This reverts commit 7ef597f.
Fix conflicts: epochX/cudacpp/gg_tt.mad/SubProcesses/runTest.cc

OK! Now the test runs, but nomultichannel succeeds, while multichannel fails as the reference ME is wrong!
This is now back on track, must create a second reference file, then add the actual channelid filling of warps...

…channel and <file.txt2> as ref with multichannel

…unTest (use cuda/double as the reference platform)

CUDACPP_RUNTEST_DUMPEVENTS=1 ./runTest_cuda.exe

Rerunning all tests then succeeds (but the channelid array is constant in all values for the moment...)

for bck in none sse4 avx2 512y 512z cuda; do echo $bck; ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done

…th/without multichannel #896; use <file.txt> as ref without multichannel and <file.txt2> as ref with multichannel

…ate txt ref for runTest (use cuda/double as the reference platform)

CUDACPP_RUNTEST_DUMPEVENTS=1 ./runTest_cuda.exe
\cp ../../test/ref/dump* ../../../CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/test/ref/

Rerunning all tests then succeeds (but the channelid array is constant in all values for the moment...)

… tests #896

…channelid debugging)

…onstexpr

FIXME? #910 this is a third different expression for the number of diagrams, should sanity checks for internal consistency...

….cc, move the dumpSignallingFPEs() call to the base clas dtor, add debug printouts commented out

…tations from .h to .cc, move the dumpSignallingFPEs() call to the base clas dtor, add debug printouts commented out

…ebug printouts if the code is compiled with 'make MG5AMC_CHANNELID_DEBUG=1'

FIXME? Note that MEKDevice takes a device channelid array, it would be easier if this was always a host array and MEKD managed the copy?

…add channelid debug printouts if the code is compiled with 'make MG5AMC_CHANNELID_DEBUG=1'

FIXME? Note that MEKDevice takes a device channelid array, it would be easier if this was always a host array and MEKD managed the copy?

…els 1,2,3,1,2,3... for different events (previously it was 1 for all events)

NB1: the cuda test now fails, the reference file must be recreated
NB2: I expect the SIMD tests to fail using the CUDA reference, due to the different bugs in the current channelId implementation
NB3: eventually #898 the implementation should enforce that all events in a warp use the same channelid

…channel test #896 to use channels 1,2,3,1,2,3... for different events (previously it was 1 for all events)

NB1: the cuda test now fails, the reference file must be recreated
NB2: I expect the SIMD tests to fail using the CUDA reference, due to the different bugs in the current channelId implementation
NB3: eventually #898 the implementation should enforce that all events in a warp use the same channelid

…ate txt ref for runTest (use cuda/double as the reference platform)

CUDACPP_RUNTEST_DUMPEVENTS=1 ./runTest_cuda.exe
\cp ../../test/ref/dump* ../../../CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/test/ref/

NB: the CUDA test succeeds with the new reference files, but the C++ multichannel test #896 fails due to bugs #894 and #899

… bug #899

make -j bldall -f cudacpp.mk
for bck in none sse4 avx2 512y 512z cuda; do echo BACKEND=$bck; ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done | egrep '(BACKEND|channelids_sv)'

BACKEND=none
channelids_sv 0 1
channelids_sv 1 1
channelids_sv 2 1
channelids_sv 3 1
channelids_sv 4 1
channelids_sv 5 1
channelids_sv 6 1
channelids_sv 7 1
channelids_sv 8 1
channelids_sv 9 1
channelids_sv 10 1
channelids_sv 11 1
channelids_sv 12 1
channelids_sv 13 1
channelids_sv 14 1
channelids_sv 15 1
BACKEND=sse4
channelids_sv 0 1 2
channelids_sv 2 1 2
channelids_sv 4 1 2
channelids_sv 6 1 2
channelids_sv 8 1 2
channelids_sv 10 1 2
channelids_sv 12 1 2
channelids_sv 14 1 2
BACKEND=avx2
channelids_sv 0 1 2 3 1
channelids_sv 4 1 2 3 1
channelids_sv 8 1 2 3 1
channelids_sv 12 1 2 3 1
BACKEND=512y
channelids_sv 0 1 2 3 1
channelids_sv 4 1 2 3 1
channelids_sv 8 1 2 3 1
channelids_sv 12 1 2 3 1
BACKEND=512z
channelids_sv 0 1 2 3 1 2 3 1 2
channelids_sv 8 1 2 3 1 2 3 1 2
BACKEND=cuda

…nly separating allChannelIds and channelIds and adding a missing ieventAccessRecordConst call

Also add the missing ieventAccessRecordConst function in MemoryAccessChannelIds.h and comment out unused non-const function kernelAccess

The debug printouts show now that the issue is solved

for bck in none sse4 avx2 512y 512z cuda; do echo BACKEND=$bck; ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done | egrep '(BACKEND|channelids_sv)'
BACKEND=none
channelids_sv 0 1
channelids_sv 1 2
channelids_sv 2 3
channelids_sv 3 1
channelids_sv 4 2
channelids_sv 5 3
channelids_sv 6 1
channelids_sv 7 2
channelids_sv 8 3
channelids_sv 9 1
channelids_sv 10 2
channelids_sv 11 3
channelids_sv 12 1
channelids_sv 13 2
channelids_sv 14 3
channelids_sv 15 1
BACKEND=sse4
channelids_sv 0 1 2
channelids_sv 2 3 1
channelids_sv 4 2 3
channelids_sv 6 1 2
channelids_sv 8 3 1
channelids_sv 10 2 3
channelids_sv 12 1 2
channelids_sv 14 3 1
BACKEND=avx2
channelids_sv 0 1 2 3 1
channelids_sv 4 2 3 1 2
channelids_sv 8 3 1 2 3
channelids_sv 12 1 2 3 1
BACKEND=512y
channelids_sv 0 1 2 3 1
channelids_sv 4 2 3 1 2
channelids_sv 8 3 1 2 3
channelids_sv 12 1 2 3 1
BACKEND=512z
channelids_sv 0 1 2 3 1 2 3 1 2
channelids_sv 8 3 1 2 3 1 2 3 1
BACKEND=cuda

NB: after fixing bug #899, the SIMD tests still fail because of bug #894

for bck in none sse4 avx2 512y 512z cuda; do echo BACKEND=$bck; ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done | egrep '(BACKEND| ME |r.ME |In comparing)'
BACKEND=none
BACKEND=sse4
MadgraphTest.h:254: In comparing event 0 from iteration 0
  ME  1.094026373218036e-01
r.ME  8.613813520483170e-02
BACKEND=avx2
MadgraphTest.h:254: In comparing event 0 from iteration 0
  ME  1.914754413491214e-01
r.ME  8.613813520483170e-02
BACKEND=512y
MadgraphTest.h:254: In comparing event 0 from iteration 0
  ME  1.914754413491214e-01
r.ME  8.613813520483170e-02
BACKEND=512z
MadgraphTest.h:254: In comparing event 0 from iteration 0
  ME  1.972915668783644e-01
r.ME  8.613813520483170e-02
BACKEND=cuda

…s in MemoryAccessChannelIds.h after fixing #899

…ts for bug #899 after fixing the bug

…n channelIds and allChannelIds also in sigmaKin (fix #899; wip on #911)

…s SIMD handling of numertaors and denominators

Now the tests #898 succeed when using channels123123123...

for bck in none sse4 avx2 512y 512z cuda; do echo BACKEND=$bck; ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done | egrep '(BACKEND| ME |r.ME>
BACKEND=none
BACKEND=sse4
BACKEND=avx2
BACKEND=512y
BACKEND=512z
BACKEND=cuda

Note: the debug printout gives
DEBUG: MEKB processed 512 events across 3 channels { 1 : 172, 2 : 170, 3 : 170 }

…cks #898 that channelids are the same inside each warp

As expected, the tests #898 using 123123123 etc fail the assertions

for bck in none sse4 avx2 512y 512z cuda; do echo BACKEND=$bck; ./build.${bck}_d_inl0_hrd0/runTest_*.exe; done |& egrep '(BACKEND|Assert)'
BACKEND=none
BACKEND=sse4
runTest_cpp.exe: CPPProcess.cc:335: void mg5amcCpu::calculate_wavefunctions(int, const fptype*, const fptype*, mgOnGpu::fptype*, const unsigned int*, mgOnGpu::fptype*, mgOnGpu::fptype*, mg5amcCpu::fptype_sv*, int): Assertion `channelId == channelIds_sv[i]' failed.
BACKEND=avx2
runTest_cpp.exe: CPPProcess.cc:335: void mg5amcCpu::calculate_wavefunctions(int, const fptype*, const fptype*, mgOnGpu::fptype*, const unsigned int*, mgOnGpu::fptype*, mgOnGpu::fptype*, mg5amcCpu::fptype_sv*, int): Assertion `channelId == channelIds_sv[i]' failed.
BACKEND=512y
runTest_cpp.exe: CPPProcess.cc:335: void mg5amcCpu::calculate_wavefunctions(int, const fptype*, const fptype*, mgOnGpu::fptype*, const unsigned int*, mgOnGpu::fptype*, mgOnGpu::fptype*, mg5amcCpu::fptype_sv*, int): Assertion `channelId == channelIds_sv[i]' failed.
BACKEND=512z
runTest_cpp.exe: CPPProcess.cc:335: void mg5amcCpu::calculate_wavefunctions(int, const fptype*, const fptype*, mgOnGpu::fptype*, const unsigned int*, mgOnGpu::fptype*, mgOnGpu::fptype*, mg5amcCpu::fptype_sv*, int): Assertion `channelId == channelIds_sv[i]' failed.
BACKEND=cuda

… test #896 to use channels 1,2,3,1,2,3... for different WARPS of 32 events

NB: the cuda test now fails, the reference file must be recreated

…ate txt ref for runTest (use cuda/double as the reference platform)

CUDACPP_RUNTEST_DUMPEVENTS=1 ./runTest_cuda.exe
\cp ../../test/ref/dump* ../../../CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/test/ref/

Note: the debug printout is now the following, all multiples of 32
DEBUG: MEKB processed 512 events across 3 channels { 1 : 192, 2 : 160, 3 : 160 }

…check all is ok

…the fact #898 that channelids are the same inside each warp

This also fixes all pending CID_ACCESS::ieventAccessRecordConst calls #899 and #911

Note: CPPProcess.cc is now again very close to upstream/master (most of master_june24 changes from #830 have been undone)

#898, all ok

…nelids - add new files

for d in $(git ls-tree --name-only HEAD *.mad *.sa); do \
  git add $d/SubProcesses/MemoryAccessChannelIds.h $d/SubProcesses/P*/MemoryAccessChannelIds.h \
          $d/SubProcesses/valgrind.h $d/SubProcesses/P*/valgrind.h; done

…2 reference logs for all processes (from cuda)

…r all processes (using the CODEGEN/recreateRefs.sh script)

Note: some processes fail (some of the fixes in upstream/master may be needed...)

[ RUN      ] SIGMA_SM_GG_TTXGG_GPU_MULTICHANNEL.compareMomAndME
DEBUG: MatrixElementKernelDevice::computeMatrixElements 0x1420650 T 256
runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.

[ RUN      ] SIGMA_SM_GG_TTXGGG_GPU_MULTICHANNEL.compareMomAndME
DEBUG: MatrixElementKernelDevice::computeMatrixElements 0x3a1fe00 T 256
runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.

[----------] 1 test from SIGMA_SM_GG_TTXGG_GPU_MULTICHANNEL
[ RUN      ] SIGMA_SM_GG_TTXGG_GPU_MULTICHANNEL.compareMomAndME
DEBUG: MatrixElementKernelDevice::computeMatrixElements 0x15e77f0 T 256
runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.

…ne24) to 6c7fda883 (current valassi_gpucpp_june24 including merge of current gpucpp)

Note: what remains to be done is to include the changes in gpucpp_june24 that are not yet used by master_june24 (see #886)

…e_data, txt ref files also changes in .sa processes (and add .txt2 in .sa even if not strcictly needed)

…e to merge the valgrind branch, checkout all .mad but gg_tt.mad

for d in $(git ls-tree --name-only HEAD *.mad | grep -v ^gg_tt.mad); do git checkout valgrind $d; done

Fix conflicts:
	MG5aMC/mg5amcnlo (git add ../../MG5aMC/mg5amcnlo)
	epochX/cudacpp/gg_tt.mad/SubProcesses/reweight.f (use Olivier's latest MG5AMC fix)

…, no changes

Fix conflicts (use current june24, which was synced wfrom upstream/master and is more recent: git checkout june24 ../../.github)
	.github/workflows/testsuite_oneprocess.sh
	.github/workflows/testsuite_oneprocess.yml

No need to regenerate, this only changed tmad and .github

…e to merge the tmad branch, checkout CODEGEN and tmad logs from tmad

for d in $(git ls-tree --name-only HEAD cudacpp/*/CODEGEN*txt cudacpp/tmad); do git checkout tmad $d; done

Fix conflicts:
	.github/workflows/testsuite_oneprocess.sh (keep june24 version: git checkout june24 ../../.github/)
	MG5aMC/mg5amcnlo (keep june24 version: git add ../../MG5aMC/mg5amcnlo)

… changes

…e to merge the color2 branch, checkout all .sa (and gg_tt.mad CODEGEN log) from color2

for d in $(git ls-tree --name-only HEAD *.sa); do git checkout color2 $d; done
for d in $(git ls-tree --name-only HEAD gg_tt.mad/CODEGEN*); do git checkout color2 $d; done

Fix conflicts:
        MG5aMC/mg5amcnlo (keep june24 version: git add ../../MG5aMC/mg5amcnlo)
	epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/process_sigmaKin_function.inc
	epochX/cudacpp/gg_tt.mad/SubProcesses/P1_gg_ttx/CPPProcess.cc
	epochX/cudacpp/gg_tt.mad/SubProcesses/P1_gg_ttx/coloramps.h

…no changes

Fix conflicts (use june24 versions in all cases: ignore omp/clang lines in cudacpp.mk, these will be merged from omp/clang branches later on)
	epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/cudacpp.mk
	epochX/cudacpp/gg_tt.mad/CODEGEN_mad_gg_tt_log.txt
	epochX/cudacpp/gg_tt.mad/SubProcesses/cudacpp.mk

…changes

Fix conflicts (keep the june24 version of MG5AMC as this already includes the merge of gpucpp into valassi_gpucpp_june24
	MG5aMC/mg5amcnlo (git add ../../MG5aMC/mg5amcnlo)
	epochX/cudacpp/gg_tt.mad/CODEGEN_mad_gg_tt_log.txt (git checkout june24 gg_tt.mad/CODEGEN_mad_gg_tt_log.txt)

…no changes

…o changes

Fix conflicts (keep the june24 of runTest.cc, including two tests without/with multichannel)
	epochX/cudacpp/CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/madgraph/iolibs/template_files/gpu/runTest.cc
	epochX/cudacpp/gg_tt.mad/SubProcesses/runTest.cc

…l is ok, no changes

… and test1 second to ensure that the channelid printout comes from test1 first

…e channelid debug printouts

…r into the modified master_june24

…t and gg_ttg, but this was not needed...

…p some debug printouts

…n sa builds

…GS by USEOPENMP (keep OMP disabled byut add the option to enable it)

…CHANNELID_DEBUG=1 to debug channelid in MatrixElementKernelBase

…le OpenMP by default unless USEOPENMP=1

./CODEGEN/generateAndCompare.sh gg_tt --mad --nopatch
git diff --no-ext-diff -R gg_tt.mad/Source/makefile gg_tt.mad/Source/dsample.f gg_tt.mad/Source/genps.inc gg_tt.mad/SubProcesses/makefile > CODEGEN/PLUGIN/CUDACPP_SA_OUTPUT/MG5aMC_patches/PROD/patch.common
git diff --no-ext-diff -R gg_tt.mad/bin/internal/banner.py gg_tt.mad/bin/internal/gen_ximprove.py gg_tt.mad/bin/internal/madevent_interface.py >> 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/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

…keep OMP disabled but add the option to enable it)

…ebug channelid in MatrixElementKernelBase

…ests

STARTED  AT Wed Jul 17 01:07:19 PM CEST 2024
./tput/teeThroughputX.sh -mix -hrd -makej -eemumu -ggtt -ggttg -ggttgg -gqttq -ggttggg -makeclean
ENDED(1) AT Wed Jul 17 02:54:54 PM CEST 2024 [Status=2]
./tput/teeThroughputX.sh -flt -hrd -makej -eemumu -ggtt -ggttgg -inlonly -makeclean
ENDED(2) AT Wed Jul 17 03:10:55 PM CEST 2024 [Status=2]
./tput/teeThroughputX.sh -makej -eemumu -ggtt -ggttg -gqttq -ggttgg -ggttggg -flt -bridge -makeclean
ENDED(3) AT Wed Jul 17 03:16:02 PM CEST 2024 [Status=2]
./tput/teeThroughputX.sh -eemumu -ggtt -ggttgg -flt -rmbhst
ENDED(4) AT Wed Jul 17 03:17:40 PM CEST 2024 [Status=2]
./tput/teeThroughputX.sh -eemumu -ggtt -ggttgg -flt -curhst
ENDED(5) AT Wed Jul 17 03:19:15 PM CEST 2024 [Status=2]
./tput/teeThroughputX.sh -eemumu -ggtt -ggttgg -flt -common
ENDED(6) AT Wed Jul 17 03:20:54 PM CEST 2024 [Status=2]
./tput/teeThroughputX.sh -mix -hrd -makej -susyggtt -susyggt1t1 -smeftggtttt -heftggbb -makeclean
ENDED(7) AT Wed Jul 17 03:49:01 PM CEST 2024 [Status=2]

./tput/logs_ggttggg_mad/log_ggttggg_mad_f_inl0_hrd0_bridge.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttggg_mad/log_ggttggg_mad_f_inl0_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttggg_mad/log_ggttggg_mad_f_inl0_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttggg_mad/log_ggttggg_mad_m_inl0_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttggg_mad/log_ggttggg_mad_d_inl0_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttggg_mad/log_ggttggg_mad_m_inl0_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttggg_mad/log_ggttggg_mad_d_inl0_hrd0_bridge.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttggg_mad/log_ggttggg_mad_d_inl0_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl0_hrd0_common.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl0_hrd0_curhst.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl1_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl0_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl1_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl1_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl0_hrd0_curhst.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl0_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl0_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_m_inl0_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl0_hrd0_bridge.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_m_inl0_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl1_hrd0.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl0_hrd1.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl0_hrd0_bridge.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_d_inl0_hrd0_rmbhst.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl0_hrd0_rmbhst.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.
./tput/logs_ggttgg_mad/log_ggttgg_mad_f_inl0_hrd0_common.txt:runTest_cuda.exe: GpuRuntime.h:26: void assertGpu(cudaError_t, const char*, int, bool): Assertion `code == gpuSuccess' failed.

…paces and too many empty lines)

…ments when inserting into auto_dsig1.f

…in sigmaKin...?

(This is from a cppnone build)
./madevent_cpp < /tmp/avalassi/input_ggtt_x1_cudacpp
...
DEBUG Bridge::cpu_sequence channelIds=0x47c1c4, nevt=64
DEBUG: channelIds[0..nevt]={0, 16, 1, 0, 0, 0, 0, 2, 2, 2, 2, 4294967295, 4294967295, 4294967295, 1, 4294967295, 4294967295, 4294967295, 4294967295, 4294967295, 4294967295, 1, 1, 4294967295, 4294967295, 1, 4294967295, 4294967295, 1, 4294967295, 1, 4294967295, 1, 4294967295, 4294967295, 4294967295, 1, 1, 1, 4294967295, 1, 1, 4294967295, 1, 4294967295, 4294967295, 1, 1, 4294967295, 4294967295, 4294967295, 1, 4294967295, 1, 1, 1, 4294967295, 1, 4294967295, 1, 1, 4294967295, 1, 1 }
INTERNAL ERROR! channelId == 0?
DEBUG: allChannelIds=0x63f0480, nevt=64
DEBUG: allChannelIds[0..nevt]={0, 16, 1, 0, 0, 0, 0, 2, 2, 2, 2, 4294967295, 4294967295, 4294967295, 1, 4294967295, 4294967295, 4294967295, 4294967295, 4294967295, 4294967295, 1, 1, 4294967295, 4294967295, 1, 4294967295, 4294967295, 1, 4294967295, 1, 4294967295, 1, 4294967295, 4294967295, 4294967295, 1, 1, 1, 4294967295, 1, 1, 4294967295, 1, 4294967295, 4294967295, 1, 1, 4294967295, 4294967295, 4294967295, 1, 4294967295, 1, 1, 1, 4294967295, 1, 4294967295, 1, 1, 4294967295, 1, 1 }
madevent_cpp: CPPProcess.cc:1096: void mg5amcCpu::sigmaKin(const fptype*, const fptype*, const fptype*, const fptype*, mgOnGpu::fptype*, const unsigned int*, mgOnGpu::fptype*, mgOnGpu::fptype*, int*, int*, int): Assertion `channelId > 0' failed.

…SEQUENCE_NOMULTICHANNEL and replace CHANNELS(1) by CHANNELS in FBRIDGESEQUENCE call

Compilation fails: there is a bug in fbridge.inc #916

gfortran -w -fPIC -O3 -ffast-math -fbounds-check -ffixed-line-length-132 -w -cpp -c -DMG5AMC_MEEXPORTER_CUDACPP auto_dsig1.f -I../../Source/  -o auto_dsig1_cudacpp.o
auto_dsig1.f:612:32:

  612 |      &      HEL_RAND, COL_RAND, CHANNELS, OUT2,
      |                                1
Error: Rank mismatch in argument ‘chanid’ at (1) (scalar and rank-1)

)

…e of the fbridgesequence call in Fortran to accept channelid arrays

In fbridge.cc, improve the comment and variable names (channelIds is a ponter like all others)

Now madevent completes successfullly and prints out a cross section
DEBUG: MEK 0x6dc99a0 processed 8320 events across 3 channels { 0 : 64, 1 : 8256 }

….cc and Bridge.h

Revert "[june24] in gg_tt.mad CPPProcess.cc and Bridge.h, debug channelID==0 in sigmaKin...?"
This reverts commit 19105b6.

…for MEK channelid processing (less verbose, add a tag)

…her issues

…_used #885 and #765

…to_dsig.f and auto_dsig1.f,