1
0
mirror of https://github.com/paboyle/Grid.git synced 2024-11-16 10:45:36 +00:00
Commit Graph

7961 Commits

Author SHA1 Message Date
Peter Boyle
5791021dcd Speed up Cshift more with coalesced 2020-09-01 15:28:15 -04:00
Peter Boyle
c273fb051c Peek poke laattice 2020-09-01 15:27:59 -04:00
Peter Boyle
c545530170 little worry large Nbasis doesnt compile GPU 2020-09-01 00:14:33 -04:00
Peter Boyle
d982a5b6d5 Fix coaarsened 2020-09-01 00:14:04 -04:00
Peter Boyle
15ca8637f3 No norms in HermOp 2020-09-01 00:13:32 -04:00
Peter Boyle
cbc995b74c Made better interface 2020-09-01 00:12:54 -04:00
Peter Boyle
8b74174d74 Eigen tensor serialisatiino happy undeer GPU. Regret agreeing to let us couple Eigen types to Grid IO 2020-09-01 00:03:26 -04:00
Peter Boyle
e21fef17df real and imag part not in ET 2020-08-31 23:56:26 -04:00
Peter Boyle
3d27708f07 Basic where test 2020-08-31 23:55:49 -04:00
Peter Boyle
b918744184 Prettificatoin 2020-08-31 23:54:46 -04:00
Peter Boyle
7d14a3c086 Where working 2020-08-31 23:53:46 -04:00
Peter Boyle
e14a84317d GPU math unary calls 2020-08-31 23:50:49 -04:00
Peter Boyle
6c31b99f1f I knew coupling Eigen Tensor to Grid serialisation was a bad iddea.
Now the complex is different on GPU creates probblems
2020-08-31 23:49:19 -04:00
Peter Boyle
9522dcd611 Remove dead commented ouot coode 2020-08-31 23:40:29 -04:00
Peter Boyle
ed469898dc coalesced ET expressions 2020-08-31 23:38:40 -04:00
Peter Boyle
1eee94a809 Sorting real/im in read coalesced GPU ET 2020-08-31 23:36:49 -04:00
Bartosz Kostrzewa
54523369a3 do not use backspace in Coordinate output stream operator 2020-08-31 19:39:36 +02:00
Peter Boyle
a98c91c2a5
Merge pull request #309 from kostrzewa/format_benchmark_wilson_sweep
Format benchmark wilson sweep
2020-08-31 12:43:46 -04:00
Bartosz Kostrzewa
a9b92867a8 use tabulator 2020-08-31 18:41:17 +02:00
Bartosz Kostrzewa
65920faeba correct formatting of Benchmark_wilson_sweep output 2020-08-31 18:39:27 +02:00
Christoph Lehner
249e2db87d
Merge pull request #14 from DanielRichtmann/feature/gpt-coarsenedmatrix
Expose more functions in CMat
2020-08-27 15:18:56 +02:00
Daniel Richtmann
cf3535d16e
Expose more functions in CMat 2020-08-27 14:06:48 +02:00
Christoph Lehner
d61ee817f4
Merge pull request #13 from DanielRichtmann/feature/gpt-coarsenedmatrix
Changes needed for GPT MG
2020-08-27 12:11:06 +02:00
Peter Boyle
3448b7387c Almost there to coalesced ET 2020-08-26 17:04:49 -04:00
Peter Boyle
47b89d2739 Pragma protection improvementt 2020-08-26 17:04:27 -04:00
Christoph Lehner
2a75516330 state MPI/SLURM message only on world_rank zero 2020-08-26 12:34:17 -04:00
Daniel Richtmann
b2087f14c4
Fix CoarsenedMatrix regarding illegal memory accesses
Need a reference to geom since the lambda copies the this pointer which points to host memory, see
- https://docs.nvidia.com/cuda/cuda-c-programming-guide/#star-this-capture
- https://devblogs.nvidia.com/new-compiler-features-cuda-8/
2020-08-24 17:46:47 +02:00
Daniel Richtmann
dd1ba266b2
Fix mapping between dir + disp and point in CMat 2020-08-24 17:46:46 +02:00
Daniel Richtmann
1292d59563
Add a typedef + broaden interface of CMat 2020-08-24 17:46:45 +02:00
Christoph Lehner
9877ed9bf8
Merge pull request #12 from paboyle/develop
Sync
2020-08-22 16:35:35 +02:00
Christoph Lehner
f0dc0f3621 fix compile issue on Qpace3 2020-08-22 13:57:33 +02:00
Peter Boyle
1efe30d6cc SLurm stop nodes using same GPU 2020-08-21 02:02:53 +02:00
Peter Boyle
0b787e9fe0 Avoid namespaec collision to make gcc happy 2020-08-20 22:23:29 +02:00
Peter Boyle
37ec4b241c Default thread count sensible 2020-08-20 22:12:31 +02:00
Christoph Lehner
63b0a19f37
Merge pull request #11 from paboyle/develop
Sync
2020-08-20 20:53:39 +02:00
Peter Boyle
90ea7dfa99 Accelerator loops for device resident comms buf 2020-08-19 22:40:44 +02:00
Peter Boyle
f866d7c33e
Merge pull request #307 from lehner/feature/gpt
Merged Nils's A64FX and minor fixes (MemoryManager::InitMessage, Tensor_index zeroit, ...)
2020-08-18 23:27:21 -04:00
Christoph Lehner
542bdef198 cleanup comments 2020-08-14 18:39:44 +02:00
Christoph Lehner
06007db3d9 true shm_none implementation with GPUs that disables the use of device shared memory for the stencils 2020-08-14 18:37:00 +02:00
Christoph Lehner
12e6059a70 Merge branch 'feature/gpt' of https://github.com/lehner/Grid into feature/gpt 2020-08-13 16:16:52 +02:00
Christoph Lehner
dbaa24ebf6 further GPU memory access fixes (with this GPT passes all single-rank tests on non-summit GPUs) 2020-08-13 16:14:15 +02:00
Peter Boyle
3276aa67dc Update 2020-08-12 14:15:53 -04:00
Christoph Lehner
3b30b9f0c0 Merge branch 'feature/gpt' of https://github.com/lehner/Grid into feature/gpt 2020-08-06 16:59:17 +02:00
Christoph Lehner
69db4816f7 fix variable capture in Scatter_plane_merge on accelerators 2020-08-06 16:57:16 +02:00
Christoph Lehner
3abe09025a when using SHM_NONE allow multiple ranks per node but without using shared memory 2020-08-06 14:42:38 +02:00
Christoph Lehner
e33878e0de
Trigger re-run of CI 2020-08-06 11:50:24 +02:00
Christoph Lehner
27b4fbf3f0 assert for forbidden code path and fix check for faster CPU codepath in basisRotate 2020-08-03 07:57:33 -04:00
Christoph Lehner
968a90633a Zero -> zeroit in Tensor_index 2020-07-31 02:07:17 -04:00
Christoph Lehner
6365a89ba3 create separate InitMessage for MemoryManager that can be called after communicator setup 2020-07-30 07:25:05 -04:00
Christoph Lehner
ddbb008694
Merge pull request #10 from lehner/feature/gpt-sycl
Feature/gpt sycl
2020-07-30 13:12:09 +02:00