Code

merge changes from opticks v0.3.1#25

Code

Merged

plexoos merged 106 commits intoesiBNLNPPS/eic-opticks:esifrom esi-rcBNLNPPS/eic-opticks:esi-rcCopy head branch name to clipboardFeb 5, 2025

+23,179-4,875

Commits

Commits on Nov 12, 2024

• remove old GDMLPath, start reviving local validation script G4CXTest_GEOM.sh and associated chi2 checker sseq_index_test.sh

  simoncblyth

  committed
  dc961d5

  Copy full SHA for dc961d5

• update progress notes

  simoncblyth

  committed
  ada41a4

  Copy full SHA for ada41a4

• lack of RE on B side fixed by putting process order back how it was, prior to UseGivenVelocity kludge investigation

  simoncblyth

  committed
  4544b1a

  Copy full SHA for 4544b1a

• revived simtrace running as look into deviant simple history associated with the sticks notes/issues/G4CXTest_GEOM_deviant_simple_history_TO_BT_BT_SA_at_1000_per_1M_level.rst

  simoncblyth

  committed
  8d63987

  Copy full SHA for 8d63987

Commits on Nov 13, 2024

• debug notes

  simoncblyth

  committed
  153ccb4

  Copy full SHA for 153ccb4

Commits on Nov 14, 2024

• debug scripts, notes

  simoncblyth

  committed
  0c7ed8a

  Copy full SHA for 0c7ed8a

• putting the Fastener within AdditionAcrylic rather than soup volumes brings chi2 back under control once get the transform flip correct

  simoncblyth

  committed
  b213dd1

  Copy full SHA for b213dd1

Commits on Nov 15, 2024

• add an objective item for ray tracing the general spherical shell with theta and phi segments

  simoncblyth

  committed
  a4198b7

  Copy full SHA for a4198b7

Commits on Nov 22, 2024

• update advice on version sets to use

  simoncblyth

  committed
  e5b094a

  Copy full SHA for e5b094a

Commits on Nov 26, 2024

• add material debug control to U4Tree.h used for thin Water layer arounf fastener shakedown, notes on the chi2 result following shakedown, switch cxr_min.sh geometry resolve order

  simoncblyth

  committed
  3a82694

  Copy full SHA for 3a82694

• add SIMG::writeNPY

  simoncblyth

  committed
  cf41a35

  Copy full SHA for cf41a35

• add SGLM::FillZProjection while considering how to revive OpenGL depth compositing of rasterized event pixels and ray traced geometry pixels

  simoncblyth

  committed
  282fd15

  Copy full SHA for 282fd15

• start investigating how to composite the ray traced geometry with rasterized pixels such as event reps or highlighting bounding boxes, see notes/issues/impl_composited_rendering_in_7plus_workflow.rst

  simoncblyth

  committed
  8cd256c

  Copy full SHA for 8cd256c

Commits on Nov 28, 2024

• investigate projection matrix elements needed for pixel depth calculation

  simoncblyth

  committed
  8854047

  Copy full SHA for 8854047

• zdepth for all ray trace pixels starting to work, toggle rendertype with X key

  simoncblyth

  committed
  c9642c3

  Copy full SHA for c9642c3

Commits on Nov 29, 2024

• start on a more flexible SCurandState.h aiming to replace SCurandState.{hh,cc} for finer granularity and more dynamic maxphoton control

  simoncblyth

  committed
  7682021

  Copy full SHA for 7682021

• start on SCurandState.h using chunked states for more flexible maxphoton without repetition

  simoncblyth

  committed
  c748c89

  Copy full SHA for c748c89

Commits on Nov 30, 2024

• start on chunked approach to curandState setup in SCurandState.h SCurandChunk.h aiming to allow flexible runtime maxphoton config

  simoncblyth

  committed
  fca21ca

  Copy full SHA for fca21ca

• chunked curand state creation working, next implement concat chunk loading with partial load of the final chunk to meet the dynamic maxphoton requirement

  simoncblyth

  committed
  8633a3f

  Copy full SHA for 8633a3f

Commits on Dec 1, 2024

• adopted chunked SCurandState.h in QRng, making OPTICKS_MAX_PHOTON runtime dynamic via partial chunk loading in new QRng::LoadAndUpload

  simoncblyth

  committed
  e47e091

  Copy full SHA for e47e091

• prepare QRngTest.sh to compare generation between the old monolithic curandState and the new dynamic chunked curandState

  simoncblyth

  committed
  1f6157c

  Copy full SHA for 1f6157c

Commits on Dec 2, 2024

• rename the old monolithic impl accordingly SCurandStateMonolithic.hh QCurandStateMonolithic.hh QCurandStateMonolithicTest.cc etc

  simoncblyth

  committed
  f250141

  Copy full SHA for f250141

• remove the underscores from the names of the new chunked impl types SCurandState and QCurandState

  simoncblyth

  committed
  ecb371c

  Copy full SHA for ecb371c

• tidy up QRng QRngTest.sh

  simoncblyth

  committed
  45f0966

  Copy full SHA for 45f0966

• comparing generated randoms from old and new curandState impls matched perfectly at 3M level, curandState load bytes digest

  simoncblyth

  committed
  ac7defd

  Copy full SHA for ac7defd

• fix typo causing test fails

  simoncblyth

  committed
  1b27899

  Copy full SHA for 1b27899

• QRngTest matching between chunked and monolithic curandState at M100 by reducing NV to 4 to avoid 2GB truncation from somewhere

  simoncblyth

  committed
  cdfc68f

  Copy full SHA for cdfc68f

Commits on Dec 3, 2024

• fix the NP.hh 2GB truncation issue by switching all NP.hh NPX.h NPU.hh size/indice int to NP::INT typedeffed to std::int64_t : revive ~/np/tests/go.sh auto-testing

  simoncblyth

  committed
  286ea6e

  Copy full SHA for 286ea6e

Commits on Dec 4, 2024

• fixed 5/56 ~/np/tests/go.sh test fails following major change to NP.hh to fix the 2GB truncation issue

  simoncblyth

  committed
  bb47af7

  Copy full SHA for bb47af7

• QRngTest.sh for M200 perfect match between chunked and monolithic curandState, first impl setting heuristically determined max photon based on detected VRAM in SEventConfig::SetDevice when config O…

  Show description for 8a4ebb0
  simoncblyth

  committed
  8a4ebb0

  Copy full SHA for 8a4ebb0

Commits on Dec 5, 2024

• SEventConfig rationalize Initialize methods, move scontext salloc instanciation down to SEventConfig::Initialize_Meta from CSGOptiX, add indep SEventConfig::MaxCurand for QRng control

  simoncblyth

  committed
  c250059

  Copy full SHA for c250059

• start implementing split launches for events with more photons than VRAM fits with SGenstep::GetGenstepSlices

  simoncblyth

  committed
  873be6a

  Copy full SHA for 873be6a

• rename qcurandstate.h to qcurandwrap.h to avoid potential for name clash with QCurandState.h that only differs by case

  simoncblyth

  committed
  eeaf2be

  Copy full SHA for eeaf2be

• correct OptiX version in README.rst

  simoncblyth

  committed
  048ddc3

  Copy full SHA for 048ddc3

Commits on Dec 6, 2024

• rename to SGenstep.h from .hh following convention for header only impl

  simoncblyth

  committed
  57be265

  Copy full SHA for 57be265

• changing to SEventConfig::MaxSlot to explicitly convey what is actually VRAM constrained, namely the launch slots

  simoncblyth

  committed
  42202a9

  Copy full SHA for 42202a9

• add NPFold::concat that NP::Concatenates common subfold arrays into the top level NPFold, for multi-launch sub-array combination

  simoncblyth

  committed
  9d75338

  Copy full SHA for 9d75338

• handle the trivial case of a single subfold with NPFold::concat

  simoncblyth

  committed
  c802d48

  Copy full SHA for c802d48

Commits on Dec 7, 2024

• fix SGenstep::GetGenstepSlices missing last slice in some cases

  simoncblyth

  committed
  a7b852b

  Copy full SHA for a7b852b

• revive QEvent tests and add sliced genstep uploading for multi-launch support

  simoncblyth

  committed
  838ca38

  Copy full SHA for 838ca38

• generalize SEvt NPFold array handling with topfold and fold

  simoncblyth

  committed
  2da07c2

  Copy full SHA for 2da07c2

• try QSim multi-launch using sliced genstep uploads and gathering into SEvt sub-fold then NPFold::concat into SEvt topfold

  simoncblyth

  committed
  7e6f5a2

  Copy full SHA for 7e6f5a2

• enable fabrication of multi-genstep torch genstep arrays with SEvent::MakeTorchGenstep

  simoncblyth

  committed
  5ac6c46

  Copy full SHA for 5ac6c46

• verbose NPFold tracing of added and cleared arrays : for finding double frees

  simoncblyth

  committed
  66ee3b8

  Copy full SHA for 66ee3b8

• testing QSim::simulate multi-launch with TEST=ref10_multilaunch ~/o/cxs_min.sh

  simoncblyth

  committed
  fd13d5e

  Copy full SHA for fd13d5e

• debug double free on clear, reveals cause is duplicated gensteps array in each subfold : switch in QEvent::gatherComponent_ to QEvent::gatherGenstepFromDevice avoids the problem

  simoncblyth

  committed
  7349b82

  Copy full SHA for 7349b82

Commits on Dec 8, 2024

• overhaul NPFold copy, adding subfold copy and changing API for clarity with shallowcopy deepcopy added, use shallowcopy from SEvt::save

  simoncblyth

  committed
  48780ac

  Copy full SHA for 48780ac

• add maxdepth arg to NPFold::find_subfold_with_prefix to avoid sreport fail when multi-launch subfold are kept

  simoncblyth

  committed
  de7f794

  Copy full SHA for de7f794

• avoided the NPFold change parent assert with SPMT using deepcopy, after excursion from NPFold::CopyArrays bug introduced for a while

  simoncblyth

  committed
  3d9d19e

  Copy full SHA for 3d9d19e

• look around for RNG with less state that curandStateXORWOW counter-based-prng cbprng looks worth a try : can eliminate the state entirely, just need counters

  simoncblyth

  committed
  fcd65bd

  Copy full SHA for fcd65bd

Commits on Dec 9, 2024

• deep dive into curand impl, added curandStatePhilox4_32_10_OpticksLite.h stripping state to get down from 64 to 32 bytes of state

  simoncblyth

  committed
  6d5355c

  Copy full SHA for 6d5355c

Commits on Dec 10, 2024

• add handling for geometry with duplicated surface names

  simoncblyth

  committed
  afb825a

  Copy full SHA for afb825a

• make OPTICKS_MAX_CURAND=0 load and upload all available SCurandChunk files

  simoncblyth

  committed
  223f94e

  Copy full SHA for 223f94e

• rejig to facilitate switching curandState impl eg to Philox rather than XORWOW, controlled from qrng.h but as will need other changes like no state loading will also need eg WITH_PHILOX_LITE WITH_X…

  Show description for afe47f6
  simoncblyth

  committed
  afe47f6

  Copy full SHA for afe47f6

• rename srng.h to srngcpu.h for clarity

  simoncblyth

  committed
  f26ef46

  Copy full SHA for f26ef46

• get photon_slot_offset into Params on device for multi-launch genstep slices after the first, so multi-launch should be able to match single-launch

  simoncblyth

  committed
  05c53ef

  Copy full SHA for 05c53ef

Commits on Dec 11, 2024

• fix three QEvent test fails from ordinary single launch usage without genstep slices

  simoncblyth

  committed
  8f14268

  Copy full SHA for 8f14268

• simplify SEvt path handling making use of SEventConfig, plus change reldir default to incorporate TEST token SEventConfig::_EventReldirDefault = "ALL${VERSION:-0}_${TEST:-none}"

  simoncblyth

  committed
  3a83d31

  Copy full SHA for 3a83d31

• comparison of single-launch and multi-launch showing exact matching of hits except for the 31-bit photon idx in (3,2) : try to adjust to make that match too

  simoncblyth

  committed
  530661d

  Copy full SHA for 530661d

• arrange curand state usage to almost always use RNG, remove old s_mock_curand.h, instead adopt : using RNG = curandStateWhatever : approach

  simoncblyth

  committed
  166ef01

  Copy full SHA for 166ef01

Commits on Dec 12, 2024

• compare curand_init and curand_uniform between curand state impl XORWOW and Philox both with and without state loading for XORWOW, Philox is winner

  simoncblyth

  committed
  cffde33

  Copy full SHA for cffde33

• try switching qrng.h to Philox = curandStatePhilox4_32_10 : currently giving two fails from QSimTest QSimWithEventTest fails not present with XORWOW

  simoncblyth

  committed
  203a64d

  Copy full SHA for 203a64d

Commits on Dec 13, 2024

• RNG flexibility continued using qrng.h template specializations and removing rng_states array assumption now that only available for XORWOW not Philox or PhiloxLite, fix a few QSimTest fails

  simoncblyth

  committed
  b291c2d

  Copy full SHA for b291c2d

• fix a few stale tests from QSimTest_ALL.sh : down to 9/29 FAILs

  simoncblyth

  committed
  d8cfb70

  Copy full SHA for d8cfb70

Commits on Dec 14, 2024

• avoid some more QSimTest_ALL.sh FAILs, left with the four deliberate asserts noting need to reimplement

  simoncblyth

  committed
  705f7f3

  Copy full SHA for 705f7f3

Commits on Dec 15, 2024

• remove DEBUG_TAG compile definitions from all proj other than sysrap

  simoncblyth

  committed
  b92d6d7

  Copy full SHA for b92d6d7

• control RNG from macros, add sbuild.h for querying build settings and matching test strings with the settings of current build

  simoncblyth

  committed
  eeea775

  Copy full SHA for eeea775

• add OPTICKS_EVENT_NAME which when defined is constrained to contain strings matching important build settings eg Debug_Philox or Debug_XORWOW and is used within the reldir of saved SEvt

  simoncblyth

  committed
  a9b845c

  Copy full SHA for a9b845c

• get build constrained OPTICKS_EVENT_NAME to work for cxs_min.sh

  simoncblyth

  committed
  09a5046

  Copy full SHA for 09a5046

• TEST=large_evt ~/o/cxs_min.sh shakedown : change SEventConfig max defaults for Philox

  simoncblyth

  committed
  d3a4684

  Copy full SHA for d3a4684

• avoid QRngTest using rngmax for num_items as thats a billion for Philox

  simoncblyth

  committed
  a2e0b24

  Copy full SHA for a2e0b24

• shakedown testing with max_slot 0 for VRAM maximal launches

  simoncblyth

  committed
  11ac4e8

  Copy full SHA for 11ac4e8

Commits on Dec 16, 2024

• Fixed QSimTest:fake_propagate OOM by MaxSlot reduction for debug array running from QSimTest::EventConfig, improved OOM reporting.

  simoncblyth

  committed
  5bf7d5a

  Copy full SHA for 5bf7d5a

• prep for cxs_min.sh TEST=vlarge_evt

  simoncblyth

  committed
  fe0c238

  Copy full SHA for fe0c238

• Ada didnt break a sweat, push to a billion photons with vvlarge_evt

  simoncblyth

  committed
  0e7793a

  Copy full SHA for 0e7793a

• fix multi-launch kernel time reporting in QSim::simulate before get into profile info

  simoncblyth

  committed
  bac63bb

  Copy full SHA for bac63bb

• standardize NPFold DUMP-ing for control/clarity of sreport output

  simoncblyth

  committed
  158136f

  Copy full SHA for 158136f

• start generalizing NP/sreport metadata/timestamp handling to work with multi-launch running

  simoncblyth

  committed
  8fc26fe

  Copy full SHA for 8fc26fe

Commits on Dec 17, 2024

• generalize metadata handling to cope better with repeated keys, as happens with multi-launch running

  simoncblyth

  committed
  45f5123

  Copy full SHA for 45f5123

• improve metadata handling clarity of output and impl

  simoncblyth

  committed
  c53e85f

  Copy full SHA for c53e85f

• adjust QSim::simulate profile stamps for multi-launch running

  simoncblyth

  committed
  de4533c

  Copy full SHA for de4533c

Commits on Dec 18, 2024

• repeated ranges need not be contiguous, generalize identifying them

  simoncblyth

  committed
  6189db2

  Copy full SHA for 6189db2

• add x-multiplier handling to sstr::ParseIntSpecList eg for use with opticks_num_genstep

  simoncblyth

  committed
  12f399b

  Copy full SHA for 12f399b

• need cxs_min.sh ctx eg Debug_Philox on macOS without building opticks for plotting

  simoncblyth

  committed
  3fd37c9

  Copy full SHA for 3fd37c9

• update sreport_ab.sh plotting for new medium scan

  simoncblyth

  committed
  8ea371b

  Copy full SHA for 8ea371b

Commits on Jan 5, 2025

• annotations while work on proceedings and presentation

  simoncblyth

  committed
  485be5d

  Copy full SHA for 485be5d

Commits on Jan 8, 2025

• Prepare to addtag.sh v0.3.0 OPTICKS_VERSION_NUMBER 30

  simoncblyth

  committed
  8ad2028

  Copy full SHA for 8ad2028

• add QU::_cudaMalloc_OOM_NOTES regarding max_slot default based on VRAM being too high for debug running

  simoncblyth

  committed
  39d9f44

  Copy full SHA for 39d9f44

• NPFold::concat and NPFold::clear_subfold for the B side U4Recorder::EndOfEventAction_, will be done trivially as no multi-launch for Geant4 running

  simoncblyth

  committed
  71a422a

  Copy full SHA for 71a422a

• remove stray return 0 from qsim::propagate_at_boundary causing BT/BR to become zero in seq history ~/j/issues/jok-tds-missing-BR-BT-on-A-side.rst

  simoncblyth

  committed
  dc6cc80

  Copy full SHA for dc6cc80

Commits on Jan 9, 2025

• fix reversion of no hits array on B side following multi-launch impl as was trying to get photons from topfold before the gather, fix by getting from fold

  simoncblyth

  committed
  c3ce2ef

  Copy full SHA for c3ce2ef

• investigate ~/j/issues/jok-tds-mu-running-NOT-A-SENSOR-warnings.rst from qsim::propagate_at_surface_CustomART confirmed them to be from hits onto WP PMTs : so comment the warning for now

  simoncblyth

  committed
  4934664

  Copy full SHA for 4934664

Commits on Jan 10, 2025

• try genstep plotting

  simoncblyth

  committed
  bfc0970

  Copy full SHA for bfc0970

• add grab command to sgenstep_test.sh

  simoncblyth

  committed
  ff6982b

  Copy full SHA for ff6982b

• avoid case degeneracy

  simoncblyth

  committed
  8867651

  Copy full SHA for 8867651

• improve pvplt_lines for use from sgenstep__test.sh genstep plotting

  simoncblyth

  committed
  cd416a2

  Copy full SHA for cd416a2

Commits on Jan 11, 2025

• better error handling

  simoncblyth

  committed
  8c873a4

  Copy full SHA for 8c873a4

• Prepare to addtag.sh v0.3.1 OPTICKS_VERSION_NUMBER 31

  simoncblyth

  committed
  5a3aa77

  Copy full SHA for 5a3aa77

Commits on Feb 3, 2025

• Merge tag 'v0.3.1' into esi-rc

  Show description for bf3d0fa
  plexoos

  committed
  bf3d0fa

  Copy full SHA for bf3d0fa

• revert: do not convert compiler warnings to errors

  plexoos

  committed
  e7257ef

  Copy full SHA for e7257ef

Commits on Feb 4, 2025

• Merge branch 'esi' into esi-rc

  plexoos

  authored
  9e3bbf6

  Copy full SHA for 9e3bbf6

Commits on Feb 5, 2025

• remove failing test QCurandStateTest.cc

  Show description for 72c0440
  plexoos

  committed
  72c0440

  Copy full SHA for 72c0440

• use QCurandStateMonolithicTest instead of QCurandStateTest

  plexoos

  committed
  8fcaab8

  Copy full SHA for 8fcaab8

• rename random seed files for SysRapTest.SRngSpecTest

  plexoos

  committed
  508547c

  Copy full SHA for 508547c

• revert last change in sysrap/tests/SSimTest.cc

  Show description for 8727ff1
  plexoos

  committed
  8727ff1

  Copy full SHA for 8727ff1

• disable QEvent_Lifecycle_Test

  plexoos

  committed
  4ea970c

  Copy full SHA for 4ea970c