Ticket #11499 (closed: fixed)

Opened 6 years ago

Last modified 5 years ago

Parallel IntegrateEllipsoids

Reported by: Vickie Lynch Owned by: Vickie Lynch
Priority: major Milestone: Backlog
Component: Framework Keywords:
Cc: Blocked By:
Blocking: Tester: Owen Arnold

Description

OpenMP loop for qListFromEventWS and qListFromHistoWS. Also document IntegrateInHKL option better.

Change History

comment:1 Changed 6 years ago by Vickie Lynch

  • Status changed from new to inprogress

Refs #11499 parallel works for events

Changeset: c367e3e5495266efb57e7d9843c264b6b11a4eb0

comment:2 Changed 6 years ago by Vickie Lynch

Refs #11499 histo should not have 2 addEvents

Changeset: 862f38428dfb3c21a513d57af9c62981724bf39e

comment:3 Changed 6 years ago by Vickie Lynch

Refs #11499 add text to doc for IntegrateInHKL

Changeset: 0b357570a31d06b35da608ef35935c9da768b8de

comment:4 Changed 5 years ago by Vickie Lynch

Refs #11499 validate TOF and use ordered to get correct answer

Changeset: e0a732cb960029845049820c3dec6b93329d5531

comment:5 Changed 5 years ago by anonymous

  • Status changed from inprogress to verify
  • Resolution set to fixed

This is being verified as pull request #615.

comment:6 Changed 5 years ago by Owen Arnold

  • Status changed from verify to verifying
  • Tester set to Owen Arnold

comment:7 Changed 5 years ago by Vickie Lynch

Refs #11499 remove unused lines

Changeset: 5334e86efcb18673055f69c4bd9d223a48389651

comment:8 Changed 5 years ago by Vickie Lynch

Refs #11499 runs correctly in parallel

Changeset: 669bcc002ee1d5124917d0e892786c781c1fe0ff

comment:9 Changed 5 years ago by Vickie Lynch

Refs #11499 one peak is different on Mac

Changeset: bf6bd01c02e57690fc12fd0d75ee726306f2c050

comment:10 Changed 5 years ago by Owen Arnold

I have a few comments and questions, but generally, the parallel implementation has been put together how I would have expected it to be. The strange unit test failure on OSX is worrying, we've seen things like this before, and I think it's something we ought to get a better understanding of. Maybe Steven Hahn could help with this?

comment:11 Changed 5 years ago by Vickie Lynch

Refs #11499 changes suggested by Owen

Changeset: 8a967102351ac4825a4a98a8df31d22b080f7a32

comment:12 Changed 5 years ago by anonymous

I saved some output files from the test and compared my output with Steven's from the Mac. The 3rd peak was one row away on the Mac and it was centered on spectrum 5012 instead of 5013. Could it be a difference converting double to int on the Mac? 2 SEQN H K L COL ROW CHAN L2 2_THETA AZ WL D IPK INTI SIGI RFLG

3 4 -1 4 1 50.00 12.00 1875 112.446 0.47494 0.23554 0.666667 1.4170 0 0.00 0.00 310

---

3 4 -1 4 1 50.00 13.00 1875 112.557 0.47686 0.25437 0.666667 1.4114 0 0.00 0.00 310

Not sure why OSX failed this time. It was not the IntegrateEllipsoidTest.

See spectrum comparison: ![compare](https://cloud.githubusercontent.com/assets/1108354/7277646/2c06a204-e8df-11e4-9552-cc2d1891a296.png)

comment:13 Changed 5 years ago by Vickie Lynch

Refs #11499 try another peak on Mac

Changeset: 2796642e56fd8c6230c0babe0951face11726824

comment:14 Changed 5 years ago by Owen Arnold

  • Status changed from verifying to closed

Merge pull request #615 from mantidproject/11499_parallel_IntegrateEllipsoids

parallel loops for IntegrateEllipsoids

Full changeset: 040303026f14c6d9a37f4a859aac136550d5dae7

comment:15 Changed 5 years ago by Stuart Campbell

This ticket has been transferred to github issue 12337

Note: See TracTickets for help on using tickets.