PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
1) Message boards : Project Staging Area : Call for wwwwcl beta testers (OpenCL) (Message 54896)
Posted 3079 days ago by sesef
Cool, but we need to fix the stupid ntdll.dll error first. If you have more knowledge about debugging and c++ this can much help here ;)


Any tips how to reproduce ntdll problem? I tried different b and t values on my 32bit host witch Geforce 310M and didn't get any crash.
2) Message boards : Project Staging Area : Call for wwwwcl beta testers (OpenCL) (Message 54859)
Posted 3079 days ago by sesef
Is this a new GPU or an old GPU? I suspect that it doesn't support OpenCL 1.1, even though you have the 1.2 driver installed.


It's possible to make this code work on OpenCL 1.0.

You have to change:

__kernel void wieferich_kernel(__global const ulong *prime, __global short *rem, __global short *quot)


into

__kernel void wieferich_kernel(__global restrict const ulong *prime, __global int *out)


in end of code all rem and quot pack into one int like this.

rem[gid] = 1; quot[gid] = sp_quot;


out[gid] = 1; out[gid] |= ((int)sp_quot) << 16;


Also you can change all short buffers into int/uint buffers but witch one int buffer and packing data into one int you save some bandwidth.
3) Message boards : Project Staging Area : Call for wwwwcl beta testers (OpenCL) (Message 54394)
Posted 3090 days ago by sesef
Still many compiling errors on windows. Need to wait a bit longer :/


Remember to set /openmp for Multi-Threaded build.

I've posted wwwwcl 2.1.0 at home.roadrunner.com/~mrodenkirch/wwwwcl_2.1.0.zip. The changes are:


Did you consider using a http://gitorious.org/, SVN or something else. I think it will be much better to share source than upload zip files.
4) Message boards : Project Staging Area : Call for wwwwcl beta testers (OpenCL) (Message 53662)
Posted 3101 days ago by sesef
I've made some code changes to increase gpu load. Changed single thread sieve to multi-thread and what I've got.

wwwwcl.exe -p1334568e11 -P1334569e11 -TWieferich -t6 -b2000
wwwwcl v2.0
Sieve started: 133456800000000000 <= p < 133456900000000000
p=133456849813392709, 41.95M p/sec, 5.65 CPU cores, 49.8% done. ETA 02 May 01:
21
Sieve complete: 133456800000000001 <= p < 133456900000000000 2535955284 primes
tested
Elapsed time: 65.63 sec. (4.77 init + 60.76 sieve) at 38640587 p/sec.
Processor time: 364.20 sec. (23.12 init + 341.08 sieve).
Seconds spent in CPU and GPU: 270.71 (cpu), 49.33 (gpu)
Percent of time spent in CPU vs. GPU: 0.85 (cpu), 0.15 (gpu)
CPU/GPU utilization: 0.17 (cores), 0.03 (devices)


If someone want to test, binary is here (32bit win): http://dl.dropbox.com/u/1452459/wwwwcl.exe
5) Message boards : Project Staging Area : Call for wwwwcl beta testers (OpenCL) (Message 53654)
Posted 3101 days ago by sesef
Win7 x65 Radeon 7970@1080 mhz load 19%


wwwwcl.exe -p1334568e11 -P1334569e11 -TWieferich -t5 -b2000
wwwwcl v2.0
Sieve started: 133456800000000000 <= p < 133456900000000000
p=133456826811639179, 11.21M p/sec, 1.04 CPU cores, 26.8% done. ETA 01 May 22:
p=133456853946666931, 11.23M p/sec, 1.04 CPU cores, 53.9% done. ETA 01 May 22:
p=133456880919407209, 11.24M p/sec, 1.04 CPU cores, 80.9% done. ETA 01 May 22:
36
Sieve complete: 133456800000000001 <= p < 133456900000000000 2535955284 primes
tested
Elapsed time: 228.27 sec. (2.64 init + 225.53 sieve) at 11109501 p/sec.
Processor time: 237.76 sec. (3.68 init + 234.08 sieve).
Seconds spent in CPU and GPU: 917.40 (cpu), 48.97 (gpu)
Percent of time spent in CPU vs. GPU: 0.95 (cpu), 0.05 (gpu)
CPU/GPU utilization: 0.20 (cores), 0.01 (devices)
6) Message boards : Number crunching : AMD HD 7970 ppd? (Message 47440)
Posted 3202 days ago by sesef
I've figured out what is wrong. 7970 have problems with multiple kernel execution (?)

"checkCUDAErr(clWaitForEvents(1, &comp_done_event), "Waiting for computation to finish. (clWaitForEvents)");" in bottom of loop in check_ns function fix the problem, but it's increase CPU time, and it's about 14% slower.

Here is compiled version of tpsieve which works on 7970: http://dl.dropbox.com/u/1452459/tpsieve.zip

I hope there is another way to fix it, or its just a bug in AMD drivers.
7) Message boards : Sieving : ppsieve ATI/OpenCL testing (Message 46718)
Posted 3212 days ago by sesef
With AMD 7970 I've got some problems

20:39:02 (1032): Can't open init data file - running in standalone mode
Sieve started: 42070000000000 <= p < 42070010000000
Thread 0 starting
Detected 512 multiprocessors (2560 SPUs) on device 0.
Computation Error: no candidates found for p=42070000039579 between 755788 and 1001548.
20:39:03 (1032): called boinc_finish
22:15:22 (2496): Can't set up shared mem: -1. Will run in standalone mode.
Sieve started: 42070000000000 <= p < 42070010000000
Thread 0 starting
Detected 512 multiprocessors (2560 SPUs) on device 0.
Computation Error: no candidates found for p=42070000039579 between 755788 and 1001548.
22:15:23 (2496): called boinc_finish


Drivers: AMD Catalyst 12.1 Preview + OpenCL 1.2 package

2012-01-11 22:08:54 | | ATI GPU 0: ATI unknown (CAL version 1.4.1658, 3072MB, 3033MB available, 11520 GFLOPS peak)
2012-01-11 22:08:54 | | OpenCL: ATI GPU 0: Tahiti (driver version CAL 1.4.1658 (VM), device version OpenCL 1.1 AMD-APP (851.6), 6144MB)
2012-01-11 22:08:54 | | ATI GPU is OpenCL-capable


8) Message boards : AP26 - AP27 Search : AP26 on ATI Radeon (Message 21151)
Posted 3905 days ago by sesef
Newest version http://www.sesef.pl/AP/AP26PGx86-461.zip

With BOINC support and app_info.xml

Results for Radeon 2k, 3k, 4k, 57xx, 56xx, 55xx, 54xx, series
11 366384 6998839830228583
10 366384 785468127438811
25 366384 6171054912832631
10 366384 1459178643530617
10 366384 9735298495263823
10 366384 1727805601738891
10 366384 2138118918508471
11 366384 3007107694524497
10 366384 1255892682660361
11 366384 7036595108074969
10 366384 8188609810134857
10 366384 2411963918614357
10 366384 13700995611657901
14 366384 15879407069784169
11 366384 6250872237076277
10 366384 11977568522771779
10 366384 1540799946122147
12 366384 14782924219657043
13 366384 2167218735183577


Results for Radeon 58xx, 59xx
11 366384 6998839830228583
10 366384 785468127438811
25 366384 6171054912832631
10 366384 9735298495263823
10 366384 1727805601738891
10 366384 2138118918508471
11 366384 3007107694524497
10 366384 1255892682660361
11 366384 7036595108074969
10 366384 8188609810134857
10 366384 2411963918614357
10 366384 13700995611657901
14 366384 15879407069784169
10 366384 1540799946122147
13 366384 2167218735183577
9) Message boards : AP26 - AP27 Search : AP26 on ATI Radeon (Message 20545)
Posted 3928 days ago by sesef
I forgot to post correct version SOL-AP26.TXT

11 366384 6998839830228583
10 366384 785468127438811
25 366384 6171054912832631
10 366384 1459178643530617
10 366384 9735298495263823
10 366384 1727805601738891
10 366384 2138118918508471
11 366384 3007107694524497
10 366384 1255892682660361
11 366384 7036595108074969
10 366384 8188609810134857
10 366384 2411963918614357
10 366384 13700995611657901
14 366384 15879407069784169
11 366384 6250872237076277
10 366384 11977568522771779
10 366384 1540799946122147
12 366384 14782924219657043
13 366384 2167218735183577
10) Message boards : AP26 - AP27 Search : AP26 on ATI Radeon (Message 20542)
Posted 3928 days ago by sesef
New version with checkpoints and some speed improvements.

Link to current version: http://www.sesef.pl/AP/AP045.zip

Also, How much CPU % is required to feed the GPU?


Its CPU+GPU app. I moved sieving to GPU, CPU is still checking primality, so CPU usage is dependent on the core type and speed.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2020 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 0.01, 0.01, 0.00
Generated 28 Oct 2020 | 6:56:35 UTC