Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

API(4) Move HelAmps to the latest MemoryAccess classes #322

Merged
merged 100 commits into from
Jan 10, 2022
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
100 commits
Select commit Hold shift + click to select a range
ae643fb
[apihel] add __host__ to __device__ functions for MEs
valassi Dec 23, 2021
ab8db15
[apihel] improve namespaces in testxxx.cc, WIP on new MemoryAccess, b…
valassi Dec 23, 2021
2c0baf2
[apihel] first successful proof of concept for new MemoryAccess in He…
valassi Dec 23, 2021
34dd838
[apihel] WIP on testxxx, link fails
valassi Dec 24, 2021
8771755
[apihel] move imzxxx template definition to the header - builds ok bu…
valassi Dec 24, 2021
fd68ee5
[apihel] temporarely add back also the old imzxxx to debug
valassi Dec 24, 2021
8e598db
[apihel] add optional debug printouts to MemoryAccessMomenta
valassi Dec 24, 2021
f14912e
[apihel] improve debug printouts in testxxx
valassi Dec 24, 2021
b49e602
[apihel] more debug printouts for testxxx, fix some comments in Memor…
valassi Dec 24, 2021
5ad503c
[apihel] BUG FIX in testxxx - a call to ieventAccessRecordConst was m…
valassi Dec 24, 2021
04995dc
[apihel] cleanup - disable all debug printouts, runTest.exe is ok
valassi Dec 24, 2021
e1e3398
[apihel] WIP on fptye_sv in MemeoryAccess: first of all, imzxxx now t…
valassi Dec 24, 2021
d25a0a4
[apihel] comment out debug printouts instead of using "#if 0"
valassi Dec 24, 2021
6cc2ec3
[apihel] add partial hack for fptype_sv in MemoryAccessMomenta kernel…
valassi Dec 24, 2021
0b0a5d5
[apihel] add fptypevFromAlignedArray to wrap reinterpret cast
valassi Dec 24, 2021
0a0fadb
[apihel] return by value in kernelAccessIp4IparConst (prepare to supp…
valassi Dec 24, 2021
650bcd0
[apihel] WIP on unaligned/arbitrary access - testxxx now fails (no se…
valassi Dec 24, 2021
073d877
[apihel] use HostBufferMomenta in testxxx.cc - runTest still fails
valassi Dec 24, 2021
03ecc91
[apihel] reenable a "if constexpr" that I had forgotten - check.exe s…
valassi Dec 24, 2021
83fdd80
[apihel] BUG FIX in testxxx: with vector new MemoryAccess, locate ipa…
valassi Dec 24, 2021
2e9e26e
[apihel] first successful throughput test of imzxxx prototype - tests…
valassi Dec 24, 2021
13bd625
[apihel] remove unused code in HelAmps_sm.cc and rerun first test in …
valassi Jan 5, 2022
8fcc6a7
[apihel] test disabling the main changes of apihel over apirambo - re…
valassi Jan 5, 2022
3664b52
[apihel] reenable the memory access changes and lose performance again
valassi Jan 5, 2022
662bdda
[apihel] try out oxzxxx: logical bug, imzxxx worked by chance as all …
valassi Jan 5, 2022
97fa331
[apihel] WIP on fixing memory access - still MEs are 0
valassi Jan 5, 2022
37cabbb
[apihel] go back to old implementation and usual performance
valassi Jan 5, 2022
a0e8c11
[apihel] use the new oxzxxx/imzxxx in the functional tests. there all…
valassi Jan 5, 2022
feb8586
[apihel] BUG FIX which also FIXES PERFORMANCE - now c++ even gains ~3…
valassi Jan 5, 2022
f19604d
[apihel] implement new ixzxxx - all ok, same performance (or possibly…
valassi Jan 5, 2022
8509bae
[apihel] implement new opzxxx - all ok, same performance (or possibly…
valassi Jan 5, 2022
debad76
[apihel] adopt the same INLINE/ALWAYS_INLINE as old code - effectivel…
valassi Jan 5, 2022
9758f07
[apihel] add ixxxxx and ipzxxx - all tests are ok
valassi Jan 5, 2022
8b4bb8d
[apihel] add vxxxxx and sxxxxx - all tests are ok
valassi Jan 5, 2022
7569cb4
[apihel] add oxxxxx - all tests are ok
valassi Jan 5, 2022
ac09f1d
[apihel] add omxxxx, all tests are ok - this completes the move to ne…
valassi Jan 5, 2022
977915d
[apihel] clean up CPPProcess.cc and testxxx.cc - REMOVE OLD XXX CALLS
valassi Jan 5, 2022
00424d5
[apihel] clean up HelAmps_sm.cc/h - REMOVE OLD XXX FUNCTION DECLARATI…
valassi Jan 5, 2022
1ecd46f
[apihel] try moving back XXX implementation to HelAmps.cc, build fail…
valassi Jan 5, 2022
4827e15
Revert "[apihel] try moving back XXX implementation to HelAmps.cc, bu…
valassi Jan 5, 2022
2f6772f
[apihel] cleanup - use namespaces mg5amcGpu/Cpu also for HelAmps XXX/…
valassi Jan 5, 2022
f4e97e2
[apihel] cleanup, COMMENT OUT unused code in MemoryAccess.h
valassi Jan 5, 2022
ceda88e
[apihel] cleanup, REMOVE unused code in MemoryAccess.h
valassi Jan 5, 2022
aff24b1
[apihel] cleanup, RENAME MemoryAccess.h as MemoryAccessVectors.h
valassi Jan 5, 2022
f1e868c
[apihel] clean up namespaces for MemoryAccessVectors.h
valassi Jan 5, 2022
a781f08
[apihel] cleanup comments about momenta in sigmakin/calculate_wavefun…
valassi Jan 5, 2022
5597195
[apihel] clean up ispoweroftwo checks (from lambda to global inline f…
valassi Jan 5, 2022
6ac3814
[apihel] bug fix in RamboSamplingKernelHost - add missing check nevt …
valassi Jan 5, 2022
a325247
[apihel] cleanup - prepare to move neppM from mgOnGpuConfig.h to Memo…
valassi Jan 5, 2022
85ad397
[apihel] cleanup - FINALLY MOVE neppM from mgOnGpuConfig.h to MemoryA…
valassi Jan 5, 2022
16085ea
[apihel] test performance with no alignment check, eemumu/512y is 5.09E6
valassi Jan 5, 2022
3e37968
[apihel] test performance with alignment check (DEFAULT), eemumu/512y…
valassi Jan 5, 2022
75257cc
[apihel] test performance with unaligned arrays, eemumu/512y is 5.05E6
valassi Jan 5, 2022
a6a4d6f
[apihel] update comments, go back to alignment check (DEFAULT), eemum…
valassi Jan 5, 2022
c960b2c
[apihel] WIP on arbitrary arrays
valassi Jan 5, 2022
aa0d44f
[apihel] complete implementation and test performance with arbitrary …
valassi Jan 5, 2022
66881b8
[apihel] test performance with AOS (and arbitrary arrays), eemumu/512…
valassi Jan 5, 2022
d3d8fd3
[apihel] END OF PERFORMANCE TESTS, update comments, go back to DEFAUL…
valassi Jan 5, 2022
c5df299
[apihel] clean up sanity checks on gputhreads/nevt that should be a m…
valassi Jan 5, 2022
027f737
[apihel] improve readability of arbitrary array decoding
valassi Jan 5, 2022
52cfe0e
[apihel] cleanup momenta printout in check_sa: use MemoryAccessMoment…
valassi Jan 5, 2022
b52eb32
[apihel] clean up np4 and npar in runTest.cc
valassi Jan 5, 2022
9551a37
[apihel] cleanup momenta decoding in runTest: use MemoryAccessMomenta…
valassi Jan 5, 2022
3d693d3
[apihel] clean up runTest, remove unused variables (note: compiler do…
valassi Jan 5, 2022
32f80e2
[apihel] cleanup momenta encoding in testxxx: use MemoryAccessMomenta…
valassi Jan 5, 2022
7adabab
[apihel] clean up neppV vs neppM in testxxx
valassi Jan 5, 2022
b03ac6e
[apihel] testxxx fails for neppV>1 in AOS after swapping neppM for neppV
valassi Jan 5, 2022
ee0d975
[apihel] bug fix in clean up of neppV vs neppM in testxxx - also adju…
valassi Jan 5, 2022
4a3d355
[apihel] add a sanity check about neppV too in testxxx.cc
valassi Jan 5, 2022
762f759
[apihel] NEPPM CLEANUP IS COMPLETE, now go back to default AOSOA/nepp…
valassi Jan 5, 2022
5ad3891
[apihel] bug fix in codegen, remove stale grambo.cu
valassi Jan 5, 2022
eef9451
[apihel] WIP on backport to code generation and regeneration of eemum…
valassi Jan 5, 2022
7252577
[apihel] clean up codegen, remove dead code in model_handling
valassi Jan 5, 2022
73b1004
[apihel] more WIP on backport to code generation and regeneration of …
valassi Jan 5, 2022
a570168
[apihel] try again to move XXX function implementation to cc file - t…
valassi Jan 6, 2022
f594600
[apihel] first fix to move XXX function implementation to cc file - t…
valassi Jan 6, 2022
4f9089e
[apihel] attempt another fix, define INLINE as inline in all cases - …
valassi Jan 6, 2022
63ca571
[apihel] experiment with noinline keyword - gives build warnings, per…
valassi Jan 6, 2022
46e8c26
Revert the last four changes - put back XXX implementations in HelAmps.h
valassi Jan 6, 2022
ecb2ca8
[apihel] more WIP on backport to code generation and regeneration of …
valassi Jan 6, 2022
e54f638
[apihel] more WIP on backport to code generation and regeneration of …
valassi Jan 6, 2022
f37bce3
[apihel] COMPLETE BACKPORT of HelAmps.h/cc - eemumu manual and auto a…
valassi Jan 6, 2022
79b3922
[apihel] rerun performance test for eemumu manual (double) - all ok
valassi Jan 6, 2022
b276b35
[apihel] rerun performance test for eemumu manual (float) - all ok, a…
valassi Jan 6, 2022
294a7a6
[apihel] rerun performance test for eemumu manual (double, common/cur…
valassi Jan 6, 2022
1e2d2d3
[apihel] rerun performance test for eemumu manual (double, rmbhst ram…
valassi Jan 6, 2022
6490968
[apihel] rerun performance test for eemumu manual (double, hardcoded …
valassi Jan 6, 2022
a0809e6
[apihel] regenerate ggtt auto
valassi Jan 6, 2022
de07ca7
[apihel] bug fix in syncManu script
valassi Jan 6, 2022
9b0cf4d
[apihel] resync ggtt manual to ggtt auto
valassi Jan 6, 2022
54e09f4
[apihel] manually delete MemoryAccess.h in ggtt manual (not handled b…
valassi Jan 6, 2022
30f70f0
[apihel] test performance of new ggtt double - all ok (slightly worse…
valassi Jan 6, 2022
5dcb821
[apihel] rerun a second time ggtt - confirms previous performance
valassi Jan 6, 2022
6b10827
[apihel] regenerate ggttgg auto and sync to manu including new/delete…
valassi Jan 6, 2022
897ffec
[apihel] test also float performance of ggtt - as before, none/sse4 s…
valassi Jan 6, 2022
579e7bb
[apihel] test also hrd1 performance of ggtt - none/sse4 slightly fast…
valassi Jan 6, 2022
77c3d4a
[apihel] test performance of new ggttgg double - all ok (unlike ggtt,…
valassi Jan 6, 2022
af6ee70
[apihel] test performance of new ggttgg float - all ok, ~same perf, n…
valassi Jan 6, 2022
0218277
[apihel] (** COMPLETE APIHEL **) test performance of ggttgg double wi…
valassi Jan 6, 2022
5738378
Merge remote-tracking branch 'upstream/master' into apihel
valassi Jan 10, 2022
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view

Large diffs are not rendered by default.

Large diffs are not rendered by default.

Original file line number Diff line number Diff line change
Expand Up @@ -98,6 +98,7 @@ ifneq ($(wildcard $(CUDA_HOME)/bin/nvcc),)
CULIBFLAGS = -L$(CUDA_HOME)/lib64/ -lcurand # NB: -lcuda is not needed here!
CUOPTFLAGS = -lineinfo
CUFLAGS = $(OPTFLAGS) $(CUOPTFLAGS) $(INCFLAGS) $(CUINC) $(USE_NVTX) $(CUARCHFLAGS) -use_fast_math
###CUFLAGS += -Xcompiler -Wall -Xcompiler -Wextra -Xcompiler -Wshadow
NVCC_VERSION = $(shell $(NVCC) --version | grep 'Cuda compilation tools' | cut -d' ' -f5 | cut -d, -f1)
# TEMPORARY! For as long as cuda 11.0/11.1 are supported (issues #282 and #292)
# (Eventually, use only c++17; previously, c++14 in cuda110/111 and c++17 above; now c++14 everywhere)
Expand Down

This file was deleted.

Loading