sfilippone
|
3a25d7b04a
|
Fixes for LLVM compilation
|
2024-03-22 17:10:30 +01:00 |
|
sfilippone
|
373d841bce
|
Don't need renaming of psi_gth and psi_sct
|
2024-03-22 15:44:51 +01:00 |
|
sfilippone
|
472f16f0df
|
Fix compilation with --enable-serial
|
2024-03-19 14:00:03 +01:00 |
|
sfilippone
|
e0a4d362fa
|
Define flag TRACK_CUDA_MALLOC
|
2024-03-14 12:08:57 +01:00 |
|
sfilippone
|
48455190ec
|
Add GPU version of XYZW
|
2024-03-05 13:57:03 +01:00 |
|
sfilippone
|
a11f328e62
|
Added CUDA version of XYZW
|
2024-03-05 12:42:21 +01:00 |
|
sfilippone
|
86be8ebcd0
|
New method W%XYZW()
|
2024-03-04 16:29:47 +01:00 |
|
sfilippone
|
b5d5f97661
|
Improve cuda%zero()
|
2024-02-20 14:04:23 +01:00 |
|
sfilippone
|
0e269ed641
|
typo in Cabgdxyz
|
2024-02-20 13:57:32 +01:00 |
|
sfilippone
|
2391f64df6
|
X_cuda_vect%abgdxyz
|
2024-02-20 12:30:07 +01:00 |
|
sfilippone
|
93c71c4316
|
Fix %ZERO() on cuda
|
2024-02-20 10:25:31 +01:00 |
|
sfilippone
|
0568a83734
|
Fix ifdef and old code
|
2024-02-19 10:53:23 +01:00 |
|
sfilippone
|
83ededd02b
|
Implementatino of abgd_xyz
|
2024-02-13 12:54:37 +01:00 |
|
sfilippone
|
45f00e6e19
|
Fixed comments
|
2024-02-12 15:10:58 +01:00 |
|
sfilippone
|
6433dc797e
|
Fix CUDA implementation of %set_scal and %zero
|
2024-02-02 15:27:42 +01:00 |
|
sfilippone
|
097d63147a
|
Fix cuda dir makefile
|
2024-01-24 17:58:39 +01:00 |
|
sfilippone
|
3aa3c795e9
|
Refactor assembly and cnv
|
2023-12-23 13:15:01 +01:00 |
|
sfilippone
|
4d051c777d
|
Fix makefile and test program
|
2023-12-22 12:01:59 +01:00 |
|
sfilippone
|
49e99a3e82
|
Fix conversion and product to enable overlap with GPU
|
2023-12-22 12:01:41 +01:00 |
|
sfilippone
|
74cf138a6c
|
Merge branch 'repackage' into non-diag
|
2023-12-22 10:31:48 +01:00 |
|
sfilippone
|
be7571f568
|
Fix missing directive
|
2023-12-20 13:45:43 +01:00 |
|
sfilippone
|
e9d1238b43
|
Add detailed measurements.
|
2023-12-20 13:31:14 +01:00 |
|
sfilippone
|
1bc2a884e2
|
Adjust conditional compilation on CUDA version
|
2023-12-15 10:52:48 +01:00 |
|
sfilippone
|
6b65199afb
|
Check CUDA version for -dopt=on only from 11.7
|
2023-12-11 13:21:23 +01:00 |
|
sfilippone
|
0230fbb7af
|
Identufied problems with CSRG. Will fix in a branch
|
2023-12-07 12:09:00 +01:00 |
|
sfilippone
|
41491f7b9c
|
Fix HAVE_CUDA in test programs
|
2023-12-01 15:14:50 +01:00 |
|
sfilippone
|
b2b7b074df
|
Fix usage of HAVE_CUDA/HAVE_GPU (mostly disappeared)
|
2023-12-01 15:11:17 +01:00 |
|
sfilippone
|
e373ed7e0b
|
Modify configry to only use HAVE_CUDA, since SPGU is recompiled.
|
2023-12-01 15:10:39 +01:00 |
|
sfilippone
|
a6016f00fa
|
Bump PSBLAS version to 3.9
|
2023-12-01 15:10:13 +01:00 |
|
sfilippone
|
ab8631439f
|
Update configure script
|
2023-11-29 16:35:42 +01:00 |
|
sfilippone
|
6c9ca58282
|
Silly bug in coo insert
|
2023-11-29 16:14:30 +01:00 |
|
sfilippone
|
d3b2b7816d
|
Fix coo insert OpenMP. Fix Make.inc.in
|
2023-11-29 15:35:55 +01:00 |
|
sfilippone
|
655c86caed
|
Updated docs.
|
2023-11-29 10:20:38 +01:00 |
|
sfilippone
|
9b713c177b
|
Fix cuda interfaces for renaming
|
2023-11-28 19:46:20 +01:00 |
|
sfilippone
|
6fa0bf7fe7
|
Complete cuda renaming
|
2023-11-28 19:14:28 +01:00 |
|
sfilippone
|
ae7fad95d4
|
Merge branch 'development' into non-diag
|
2023-11-07 18:02:55 +01:00 |
|
sfilippone
|
a6ec655a97
|
Prepare merge
|
2023-11-07 17:57:32 +01:00 |
|
sfilippone
|
a2788bdf0b
|
New version with ND product
|
2023-11-07 13:39:44 +01:00 |
|
sfilippone
|
d718ef1e6d
|
Always allocate szs in psb_gather
|
2023-11-07 10:56:23 +01:00 |
|
sfilippone
|
baf18cebd7
|
Further fix for gather.
|
2023-11-03 14:39:01 +01:00 |
|
sfilippone
|
5caee551e5
|
Fixed IN_PLACE option for collectives.
|
2023-11-03 14:28:04 +01:00 |
|
sfilippone
|
d82b090289
|
Fix makefile for psi_acx & friends
|
2023-10-25 12:47:31 +02:00 |
|
sfilippone
|
6aa7987d52
|
Rename GPU into cuda, and merge SPGPU code.
|
2023-10-19 14:22:32 +02:00 |
|
sfilippone
|
2732336915
|
Fix gpu/makefile
|
2023-10-16 16:15:00 +02:00 |
|
sfilippone
|
81e9121c91
|
Add GPULDLIBS into Make.inc (and fix configry)
|
2023-10-02 13:08:56 +02:00 |
|
sfilippone
|
d0d4e45877
|
Fix for I gpu types from template: use psb_sizeof_ip
|
2023-09-29 13:59:40 +02:00 |
|
sfilippone
|
3a9116bc42
|
Merge psblas-ext into psblas step 2: GPU formats and configry.
|
2023-09-29 13:54:58 +02:00 |
|
sfilippone
|
1d5faa388d
|
Merge psblas-ext into psblas, step 1: ext storage formats.
|
2023-09-27 16:23:17 +02:00 |
|
sfilippone
|
d1bf46b0b1
|
Merge branch 'dev-openmp' into development
|
2023-09-07 09:50:59 +02:00 |
|
sfilippone
|
e31dd52c41
|
Fixed CRITICAL in hash_mod
|
2023-09-07 09:48:18 +02:00 |
|
sfilippone
|
def0635c53
|
More OMP directives in cd_inloc
|
2023-09-07 09:25:37 +02:00 |
|
sfilippone
|
efa29bc3ad
|
Fix configure for GCC when using --with-fcopt
|
2023-09-05 17:00:49 +02:00 |
|
sfilippone
|
3e29e603d2
|
Fix necessary flags for GCC
|
2023-09-05 16:59:03 +02:00 |
|
sfilippone
|
41be1357c3
|
Set defaults for SPSPMM depending on OpenMP compilation.
|
2023-08-30 13:45:51 +02:00 |
|
sfilippone
|
94c9060579
|
Merge branch 'dev-openmp' into development
|
2023-08-24 09:16:43 +02:00 |
|
sfilippone
|
f4f37924e2
|
Update contributor list
|
2023-08-24 09:15:25 +02:00 |
|
sfilippone
|
815ecd5cc1
|
Merge branch 'dev-openmp' into development
|
2023-08-22 10:27:28 +02:00 |
|
sfilippone
|
f001ebbad3
|
Final fix for COO on OMP
|
2023-08-22 10:22:46 +02:00 |
|
sfilippone
|
26bf4c5d69
|
Fixed COO csput for OMP/not OMP
|
2023-08-21 17:26:37 +02:00 |
|
sfilippone
|
3aa748b0e3
|
Finish dual OMP/notOMP g2lv1_ins
|
2023-08-21 17:22:16 +02:00 |
|
sfilippone
|
08c1ab0cd1
|
Fix tril/triu in COO for non-OMP paths.
|
2023-08-21 16:32:55 +02:00 |
|
sfilippone
|
ca82520b88
|
Reworked CSR TRIL/TRIU for OpenMP
|
2023-08-01 10:16:22 +02:00 |
|
sfilippone
|
5e691d5bff
|
Some improvements for openmp vector updates
|
2023-07-24 13:21:20 +02:00 |
|
sfilippone
|
bb9f213551
|
Define and implement OMP version of TRIL/TRIU
|
2023-07-18 18:19:03 +02:00 |
|
sfilippone
|
2f403e0df7
|
Rework cp_{from|to}_fmt for better OpenMP performance
|
2023-06-28 11:16:53 +02:00 |
|
sfilippone
|
d378266f33
|
Fix synatx error
|
2023-06-21 13:37:23 +02:00 |
|
sfilippone
|
fd0aea4afa
|
Merge branch 'dev-openmp' of github.com:sfilippone/psblas3 into dev-openmp
|
2023-06-21 12:22:08 +02:00 |
|
sfilippone
|
a66778f270
|
Improve coo and merge development
|
2023-06-21 12:21:27 +02:00 |
|
sfilippone
|
e72edf70a1
|
Merge branch 'development' into dev-openmp
|
2023-06-21 12:16:58 +02:00 |
|
sfilippone
|
677789469c
|
Reworked ensure_size
|
2023-06-21 12:14:47 +02:00 |
|
sfilippone
|
347352fe1e
|
Make spins work in OpenMP from either par or serial
|
2023-06-02 11:07:23 +02:00 |
|
sfilippone
|
db0e4db507
|
Minimize debug sttements in hash_ins
|
2023-06-02 11:06:53 +02:00 |
|
sfilippone
|
1941affe7a
|
Exposed error in AMG test when not parallelizing generation loop
|
2023-05-30 12:26:58 +02:00 |
|
sfilippone
|
494e29dd2e
|
Cosmetic adjustments to COO and BSRCH
|
2023-05-23 15:56:48 +02:00 |
|
sfilippone
|
739dc78a75
|
Merge branch 'development' into omp-threadsafe
|
2023-04-27 13:07:42 +02:00 |
|
sfilippone
|
7e5dc20e03
|
Define new options for BSRCH, clean interface
|
2023-04-27 13:05:46 +02:00 |
|
sfilippone
|
40cc78854a
|
Improve implementation of fix_coo using exscan
|
2023-04-21 13:41:16 +02:00 |
|
sfilippone
|
91d3e66547
|
Merge branch 'omp-threadsafe' of github.com:sfilippone/psblas3 into omp-threadsafe
|
2023-04-21 12:02:38 +02:00 |
|
sfilippone
|
74a8217520
|
Fixed silly bug in EXSCAN and usage in CSR_IMPL
|
2023-04-21 12:02:18 +02:00 |
|
sfilippone
|
f3efea0a89
|
Take out IBASE from exscan, makes no sense.
|
2023-04-20 12:31:32 +02:00 |
|
sfilippone
|
08ff37332a
|
Enable OpenMP in test/pargen
|
2023-04-19 14:06:40 +02:00 |
|
sfilippone
|
05b684ddbb
|
Updated use of exscan in CSC
|
2023-04-19 13:23:48 +02:00 |
|
sfilippone
|
9c248a31e2
|
Refactored EXSCAN and its OpenMP usage.
|
2023-04-19 12:12:58 +02:00 |
|
sfilippone
|
02dd204351
|
Implement psi_exscan and use in _from_coo
|
2023-04-18 17:00:22 +02:00 |
|
sfilippone
|
dbd55321f8
|
Fixed CSR mv and cp _from_coo with OpenMP.
|
2023-04-17 21:00:20 +02:00 |
|
sfilippone
|
6ba7d93159
|
Fix CRITICAL in LIST%G2L_INS
|
2023-04-16 12:03:16 +02:00 |
|
sfilippone
|
5a5712b4f0
|
Rely on CRITICAL inside G2L_INS implementation
|
2023-04-16 12:02:54 +02:00 |
|
sfilippone
|
f068d73ef1
|
First working version
|
2023-04-16 11:30:09 +02:00 |
|
sfilippone
|
8459ea28f5
|
Modified matrix build procedures with OpenMP
|
2023-04-14 18:25:43 +02:00 |
|
sfilippone
|
eb11e5e053
|
Put CRITICAL(name) in G2L_INS
|
2023-04-14 17:48:15 +02:00 |
|
sfilippone
|
0f1603a2e9
|
The current version of test/omp seems to be working. To be completed
|
2023-04-13 16:40:29 +02:00 |
|
sfilippone
|
98945f36b5
|
Fix nrm2 with overlap
|
2023-03-31 13:37:20 +02:00 |
|
sfilippone
|
e6a70410e0
|
Make sure to use install -p
|
2023-03-30 15:50:21 +02:00 |
|
sfilippone
|
c05b32c202
|
Reset status for csr_impl.
|
2023-03-29 14:41:35 +02:00 |
|
Salvatore Filippone
|
21164cbcbd
|
Merge pull request #7 from sfilippone/development
Updates for emergency performance fix.
|
2018-04-23 08:03:32 +01:00 |
|
Salvatore Filippone
|
ed1b966d33
|
Merge pull request #6 from sfilippone/development
Version 3.5.1
|
2018-04-10 12:36:46 +01:00 |
|