Peter Boyle
58d32a4d0e
Assertion should never hit, but did due to a bug
2015-05-10 15:24:37 +01:00
Peter Boyle
6bb17502f9
Moving operator stuff into separate file so that we can switch on/off replacement with
...
expression templates
2015-05-10 15:23:49 +01:00
Peter Boyle
8299bc39ea
Fixing breakage in the Comms non compile
2015-05-10 15:23:09 +01:00
Peter Boyle
7f04b85368
Bringing expression templates for faster vector loops
2015-05-10 15:22:31 +01:00
Peter Boyle
a115f3b086
ET ready benchmark with bytes counted assuming loop interchange
2015-05-10 15:18:04 +01:00
Peter Boyle
27c2d13968
Updated todo list
2015-05-10 15:13:50 +01:00
Peter Boyle
5415180676
Wilson perf improvements with Gauge prefetching
2015-05-06 06:37:21 +01:00
Peter Boyle
7b0dd6c5d6
Cleaned up for Linux
2015-05-05 22:09:22 +01:00
Peter Boyle
cb4b82b09f
streaming store cases
2015-05-05 18:14:09 +01:00
Peter Boyle
cd990ba13d
Streaming store option
2015-05-05 18:13:06 +01:00
Peter Boyle
249165d1b2
Added streaming stores
2015-05-05 18:09:28 +01:00
Peter Boyle
b720222d98
Updated bandwidth test
2015-05-05 18:08:53 +01:00
Peter Boyle
0e8415de1b
Added a makefile
2015-05-05 17:56:42 +01:00
Peter Boyle
2b46ad38e2
Back to vector for now; cost of init loop is clear in the a*x + y
...
loop in memory benchmark and must move to better container class.
2015-05-03 09:48:13 +01:00
Peter Boyle
9d93d1e6d4
Comms and memory benchmarks added
2015-05-03 09:44:47 +01:00
Peter Boyle
253362f978
Added a comms benchmark
2015-05-02 23:51:43 +01:00
Peter Boyle
ea52562527
Added a comms benchmark
2015-05-02 23:42:30 +01:00
Peter Boyle
6a39089a43
Starting a benchmarking sub dir
2015-05-02 17:52:36 +01:00
Peter Boyle
bdf18941a2
Improving the byte swap support for portability
2015-05-01 10:57:33 +01:00
Peter Boyle
d904e2b9ac
Merge branch 'master' of https://github.com/paboyle/Grid
2015-04-30 16:40:13 +01:00
Peter Boyle
c0ead94791
Integrated Lebesgue code and been playing with alternate implementations of the wilson dop without
...
any particular success in increasing the performance.
2015-04-30 16:39:06 +01:00
Peter Boyle
7ac997bd58
Merge pull request #1 from mspraggs/patch-1
...
Added <map> include to GridNerscIO.h
2015-04-30 09:46:48 +01:00
mspraggs
24fc71b2e9
Added <map> include to GridNerscIO.h
...
Adding this allows clang to compile Grid to completion.
2015-04-29 23:44:03 +01:00
Peter Boyle
d8ffa09e3b
Benchmark wilson dhop now; 14.6GF on one core, not as fast as SU(3)xSU(3) [23GF] but still not too shabby.
...
Disassembling output shows ugly sequences in the permute sector. Could comparatively benchmark with and without
the if-else structure to see how much I'm losing.
Drops to 9GF as it falls out of cache. Moving to Lebesgue ordering should help there. Substantive progress.
2015-04-29 06:50:18 +01:00
Peter Boyle
dcc23faa4a
Fixed the stencil sector and Wilson now agrees between stencil based implementation
...
and the cshift based implementation. Managed to reduce the volume of code in this
sector a little, but consolidation would be good, perhaps taking common
logic out into simple helper functions
2015-04-29 06:23:56 +01:00
Peter Boyle
b0485894b3
Shaken out stencil to the point where I think wilson dslash is correct.
...
Need to audit code carefully, consolidate between stencil and cshift,
and then benchmark and optimise.
2015-04-28 08:11:59 +01:00
Peter Boyle
0b7d389258
Reworking CSHIFT and Stencil. Implementing Wilson and discovered rework is required
2015-04-27 13:45:07 +01:00
Peter Boyle
35cfef2129
Big updates with progress towards wilson matrix
2015-04-26 15:51:09 +01:00
Peter Boyle
c678f2d255
Starting the implementation of wilson; incomplete and committing non-functional code which
...
is not yet included from elsewhere or linked to the build system.
2015-04-25 14:33:02 +01:00
Peter Boyle
d5fd34b6e8
Update to TODO list
2015-04-25 13:04:26 +01:00
Peter Boyle
2d8cf9e456
Added two spinor functionality required to support the Wilson hopping term.
2015-04-25 12:54:06 +01:00
Peter Boyle
dc970c6442
Dirac done ; remove from TODO
2015-04-24 22:56:37 +01:00
Peter Boyle
fc32450360
Improved the gamma quite a bit.
...
Serial rng's which are set on node zero and broadcaste
2015-04-24 20:21:40 +01:00
Peter Boyle
2a67214f9d
static names and enum list
2015-04-24 19:12:14 +01:00
Peter Boyle
71d5927a66
Vectors now too and right multiple of matrix with gamma
2015-04-24 19:08:29 +01:00
Peter Boyle
f2ac20e7ab
Removed summation
2015-04-24 18:42:44 +01:00
Peter Boyle
750dd5f5fd
Cleared the code out from Grid_summation to lattice/Grid_lattice_transfer.h
2015-04-24 18:41:34 +01:00
Peter Boyle
74432432b6
Moved code from summation into transfer and reduction
2015-04-24 18:40:44 +01:00
Peter Boyle
b8eef54fa7
First implementation of Dirac matrices as a Gamma class.
2015-04-24 18:20:03 +01:00
Peter Boyle
e2e3ea5742
Reorganised the TODO. Really getting somewhere
2015-04-23 20:42:30 +01:00
Peter Boyle
4b4dcc4c13
Rename Grid_QCD
2015-04-23 20:42:09 +01:00
Peter Boyle
afe6c4f64f
move
2015-04-23 20:41:22 +01:00
Peter Boyle
62e8d2d127
Slice summation working. May move this into lattice/Grid_lattice_reduction however
2015-04-23 15:13:00 +01:00
Peter Boyle
b7416d79e3
Begginings of slice summation and subblocking
2015-04-23 11:04:59 +01:00
Peter Boyle
2f8431ab03
Consolidate index to coor in a single routine
2015-04-23 11:04:19 +01:00
Peter Boyle
a9e574dd27
Snippets from Guido to optimise Reduce
2015-04-23 08:31:40 +01:00
Peter Boyle
73c0db82d5
Better description of Intel's many ISA targets
2015-04-23 08:02:51 +01:00
Peter Boyle
eb58297a43
Fixing endian on linux I hope
2015-04-23 07:51:15 +01:00
Peter Boyle
1851327d19
Got the NERSC IO working and fixed a bug in cshift.
2015-04-22 22:46:48 +01:00
Peter Boyle
a5b0c492d7
Rework of RNG to use C++11 random. Should work correctly maintaining parallel RNG across
...
a machine. If a "fixedSeed" is used, randoms should be reproducible across different machine
decomposition since the generators are physically indexed and assigned in lexico ordering.
2015-04-19 14:55:58 +01:00
