1db58a8acc
Precision change improvements
...
Added a new, much faster implementation of precision change that uses (optionally) a precomputed workspace containing pointer offsets that is device resident, such that all lattice copying occurs only on the device and no host<->device transfer is required, other than the pointer table. It also avoids the need to unpack and repack the fields using explicit lane copying. When this new precisionChange is called without a workspace, one will be computed on-the-fly; however it is still considerably faster than the original implementation.
In the special case of using double2 and when the Grids are the same, calls to the new precisionChange will automatically use precisionChangeFast, such that there is a single API call for all precision changes.
Reliable update and mixed-prec multishift have been modified to precompute precision change workspaces
Renamed the original precisionChange as precisionChangeOrig
Fixed incorrect pointer offset bug in copyLane
Added a test and a benchmark for precisionChange
Added a test for reliable update CG
2023-02-21 10:52:42 -05:00
rhodgson
920a51438d
Added batched Mixed precision CG
2023-02-14 17:04:13 +00:00
rhodgson
be528b6d27
Add batched block project/promote functions
2023-02-14 14:37:10 +00:00
ccd21f96ff
Plaquette agreeing and moving to final form (slowly) need to optimise
2023-02-01 22:57:44 -05:00
4b90cb8888
First cut passes combining padded cell with general stencil towards fast plaquette and staggered force
2023-02-01 22:14:10 -05:00
796abfad80
Merge pull request #422 from fjosw/fix/NVCC_DIAG_PRAGMA_SUPPORT
...
Disable diagnostic pragma warnings for CUDA 12+
2023-01-17 09:34:49 -05:00
fjosw
ad0270ac8c
fix: diagnostic pragma warnings fixed for CUDA 12+
2023-01-12 12:36:30 +00:00
7d62f1d6d2
Populate the Cshift_table in the GPU
...
Cshift is allocated in Unified memory and used
in the LambdaApply kernels but also populated
from the host. This creates a lot of Unified HtoD
and DtoH mem operations and has a negative effect
in performance. With this commit we populate the
Cshift table in the device with the
populate_Cshift_table() kernel.
2023-01-11 21:26:25 +00:00
458c943987
merged upstream
2022-12-31 11:16:21 +02:00
88015b0858
Split sum in rankSum and GlobalSum
2022-12-26 10:01:32 +01:00
4ca1bf7cca
Added gauge invariance test
2022-12-21 07:23:16 -05:00
2ff868f7a5
CPU open doesn't need to free space
2022-12-20 05:10:23 -05:00
ede02b6883
Memory manager debug Felix case
2022-12-20 05:10:23 -05:00
1822ced302
Bug fix
2022-12-20 05:10:23 -05:00
37ba32776f
More logging
2022-12-20 05:10:23 -05:00
99b3697b03
More loggin
2022-12-20 05:10:23 -05:00
43a45ec97b
SSC_START
2022-12-20 05:10:23 -05:00
b00a4142e5
A=A fix
2022-12-20 05:10:23 -05:00
3791bc527b
Logging pulled in from dirichlet branch
2022-12-20 05:10:23 -05:00
d8c29f5fcf
Updated FFT test for PETSc
2022-12-18 12:05:00 -05:00
281f8101fe
Matt FFT test
2022-12-17 20:35:33 -05:00
472ed2dd5c
Merge branch 'feature/dirichlet' of https://github.com/paboyle/Grid into feature/dirichlet
2022-12-17 20:17:09 -05:00
4f85672674
Simpler test for PETSc
2022-12-17 20:16:11 -05:00
dc747c54be
Merge branch 'develop' into feature/dirichlet
...
Conflicts:
Grid/qcd/action/fermion/WilsonCompressor.h
Grid/stencil/Stencil.h
2022-12-13 08:24:58 -05:00
140684d706
Head to head vs HMC
2022-12-13 08:15:38 -05:00
5bb7ba92fa
Test for DDHMC force term
2022-12-13 08:15:11 -05:00
b54d0f3c73
Smaller deltaH down to 7000s on t=0.5 trajectory
2022-12-13 08:14:27 -05:00
ff6777a98d
Variable depth experiments
2022-12-13 08:13:51 -05:00
07acfe89f2
Merge pull request #417 from rrhodgson/feature/fermtoprop
...
Feature/fermtoprop
2022-12-06 12:45:03 -05:00
rhodgson
40234f531f
FermToProp accelerator_for -> thread_for
2022-12-06 17:34:51 +00:00
rhodgson
d49694f38f
PropToFerm fix
2022-12-06 15:48:54 +00:00
dc6a38f177
Minor cleanup
2022-11-30 17:13:12 -05:00
82c1ecf60f
Block lanczos added
2022-11-30 16:08:40 -05:00
67f569354e
Partial dirichlet changes
2022-11-30 15:51:13 -05:00
97a098636d
FermToProp
2022-11-30 15:36:35 -05:00
e13930c8b2
Faster fermtoprop case
2022-11-30 15:11:29 -05:00
5fa573dfd3
partial send fix
2022-11-25 00:51:04 -05:00
f6402cb6c4
AUDIT removal
2022-11-25 00:50:33 -05:00
bae6c263dc
Audit
2022-11-25 00:47:01 -05:00
d71672dca9
Bug fix
2022-11-25 00:46:35 -05:00
121c9e2ceb
Tracing
2022-11-25 00:45:21 -05:00
63a30ae34f
Tracing
2022-11-25 00:45:05 -05:00
7d8231ba32
Tracing
2022-11-25 00:44:57 -05:00
b690b1cbe9
Audit
2022-11-25 00:43:57 -05:00
c0fb20fc03
Audit check for wrongly locked data
2022-11-25 00:43:12 -05:00
bc9579dac6
Old code path removed
2022-11-25 00:40:45 -05:00
a5c77f8b95
Tracing moved in order
2022-11-25 00:40:27 -05:00
3dbfce5223
Tests clean build on HIP
2022-11-16 20:15:51 -05:00
e51eaedc56
Making tests compile
2022-11-15 22:58:30 -05:00
e2a938e7f7
GPU happy for compile...?
2022-11-15 17:48:18 -05:00
Christopher Kelly
rhodgson
Peter Boyle
Peter Boyle
fjosw
Makis Kappas
Christoph Lehner
Chulwoo Jung