mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware > GPU Computing > GpuOwl

Reply
 
Thread Tools
Old 2020-09-04, 14:34   #2410
mathwiz
 
Mar 2019

1508 Posts
Default

I've just started playing with gpuOwl (latest release from the git repo).

How do I submit proofs generated by gpuOwl? Is there a script somewhere? For results.txt I know I can just paste that into the manual submission form.
mathwiz is offline   Reply With Quote
Old 2020-09-04, 14:45   #2411
moebius
 
moebius's Avatar
 
Jul 2009
Germany

3·7·19 Posts
Default

Quote:
Originally Posted by mathwiz View Post
I've just started playing with gpuOwl (latest release from the git repo).How do I submit proofs generated by gpuOwl? Is there a script somewhere? For results.txt I know I can just paste that into the manual submission form.
Go to this page

Kriesel proof upload reference material

Last fiddled with by moebius on 2020-09-04 at 14:46
moebius is offline   Reply With Quote
Old 2020-09-04, 14:46   #2412
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

464310 Posts
Default

Quote:
Originally Posted by mathwiz View Post
I've just started playing with gpuOwl (latest release from the git repo).

How do I submit proofs generated by gpuOwl? Is there a script somewhere? For results.txt I know I can just paste that into the manual submission form.
See post 26 of the gpuowl reference thread https://www.mersenneforum.org/showthread.php?t=23391
kriesel is offline   Reply With Quote
Old 2020-09-04, 18:58   #2413
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

11110001001002 Posts
Default

5500:
Code:
2020-06-05 17:13:16 gfx1012-0 OpenCL compilation in 3.10 s
2020-06-05 17:13:17 gfx1012-0 77936867 OK        0 loaded: blockSize 400, 0000000000000003
2020-06-05 17:13:21 gfx1012-0 77936867 OK      800   0.00%; 2982 us/it; ETA 2d 16:34; 1579c241dc63eca6 (check 1.27s)
2020-06-05 17:23:18 gfx1012-0 77936867 OK   200000   0.26%; 2991 us/it; ETA 2d 16:35; f0b04b45b0855bd2 (check 1.28s)
2020-06-05 17:33:15 gfx1012-0 77936867 OK   400000   0.51%; 2979 us/it; ETA 2d 16:10; c03f94396a5aa29e (check 1.27s)
2020-06-05 17:43:17 gfx1012-0 77936867 OK   600000   0.77%; 3004 us/it; ETA 2d 16:32; b9decd65ca71b629 (check 1.28s)
1080 Ti:
Code:
2020-09-04 13:24:28 GeForce GTX 1080 Ti-0 OpenCL compilation in 2.02 s
2020-09-04 13:24:29 GeForce GTX 1080 Ti-0 77936867 OK        0 loaded: blockSize 400, 0000000000000003
2020-09-04 13:24:32 GeForce GTX 1080 Ti-0 77936867 OK      800   0.00%; 2481 us/it; ETA 2d 05:43; 1579c241dc63eca6 (check 1.04s)
2020-09-04 13:32:54 GeForce GTX 1080 Ti-0 77936867 OK   200000   0.26%; 2514 us/it; ETA 2d 06:18; f0b04b45b0855bd2 (check 1.04s)
2020-09-04 13:41:12 GeForce GTX 1080 Ti-0 77936867 OK   400000   0.51%; 2483 us/it; ETA 2d 05:29; c03f94396a5aa29e (check 1.05s)
2020-09-04 13:49:27 GeForce GTX 1080 Ti-0 77936867 OK   600000   0.77%; 2473 us/it; ETA 2d 05:07; b9decd65ca71b629 (check 1.06s)
Attached Thumbnails
Click image for larger version

Name:	nvidia.png
Views:	48
Size:	42.2 KB
ID:	23255  
Xyzzy is offline   Reply With Quote
Old 2020-09-04, 21:53   #2414
mathwiz
 
Mar 2019

23·13 Posts
Default

Quote:
Originally Posted by kriesel View Post
OK. So basically:

1. manually upload results.txt
2. ./upload.py <username> <proof file>

?
mathwiz is offline   Reply With Quote
Old 2020-09-04, 23:21   #2415
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

B9916 Posts
Default

Quote:
Originally Posted by ATH View Post
I'm trying to force 3M FFT now and see if it helps.
Since I started to force 3M FFT all 15+ results have been good, so gpuowl is most likely using a too aggressive FFT selection around 54M.

I tested a 54.1M exponent in Prime95 and it chose 2880K compared to 2.75M = 2816K for gpuowl.

Last fiddled with by ATH on 2020-09-04 at 23:26
ATH is offline   Reply With Quote
Old 2020-09-05, 00:04   #2416
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

22·3·643 Posts
Default

980 Ti:
Code:
2020-09-04 17:42:56 GeForce GTX 980 Ti-0 OpenCL compilation in 1.83 s
2020-09-04 17:42:58 GeForce GTX 980 Ti-0 77936867 OK        0 loaded: blockSize 400, 0000000000000003
2020-09-04 17:43:04 GeForce GTX 980 Ti-0 77936867 OK      800   0.00%; 4221 us/it; ETA 3d 19:23; 1579c241dc63eca6 (check 1.73s)
2020-09-04 17:57:13 GeForce GTX 980 Ti-0 77936867 OK   200000   0.26%; 4258 us/it; ETA 3d 19:56; f0b04b45b0855bd2 (check 1.75s)
2020-09-04 18:11:28 GeForce GTX 980 Ti-0 77936867 OK   400000   0.51%; 4263 us/it; ETA 3d 19:49; c03f94396a5aa29e (check 1.75s)
2020-09-04 18:25:42 GeForce GTX 980 Ti-0 77936867 OK   600000   0.77%; 4262 us/it; ETA 3d 19:34; b9decd65ca71b629 (check 1.75s)
Xyzzy is offline   Reply With Quote
Old 2020-09-05, 02:18   #2417
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

11011111001012 Posts
Default

Quote:
Originally Posted by ATH View Post
Since I started to force 3M FFT all 15+ results have been good, so gpuowl is most likely using a too aggressive FFT selection around 54M.

I tested a 54.1M exponent in Prime95 and it chose 2880K compared to 2.75M = 2816K for gpuowl.

Here is a gpuowl run on a Radeon VII. Note the pErr values show there is a 1 in 200 chance of a 0.5 roundoff error during the entire PRP test. This roughly what Mihai and I were targetting (actually it is a bit higher as I never analyzed SMALL_HEIGHT=512).

Does your nVidia run show significantly different pErr values??

Code:
george@haswell:~/testbed/gpuowl$ ./gpuowl -prp 54094109 -use STATS
2020-09-05 02:09:22 gpuowl v6.11-351-ge930f9e-dirty
2020-09-05 02:09:22 config: -device 0 -log 30000
2020-09-05 02:09:22 config: -prp 54094109 -use STATS
2020-09-05 02:09:22 device 0, unique id ''
2020-09-05 02:09:22 gfx906+sram-ecc-0 54094109 FFT: 2.75M 256:11:512 (18.76 bpw)
2020-09-05 02:09:22 gfx906+sram-ecc-0 Expected maximum carry32: 50F80000
2020-09-05 02:09:23 gfx906+sram-ecc-0 OpenCL args "-DEXP=54094109u -DWIDTH=256u -DSMALL_HEIGHT=512u -DMIDDLE=11u -DPM1=0 -DAMDGPU=1 -DMM_CHAIN=2u -DMM2_CHAIN=2u -DMAX_ACCURACY=1 -DWEIGHT_STEP_MINUS_1=0x1.73cb21a140106p-3 -DIWEIGHT_STEP_MINUS_1=-0x1.3aab2a02e6dbp-3 -DSTATS=1  -cl-unsafe-math-optimizations -cl-std=CL2.0 -cl-finite-math-only "
2020-09-05 02:09:32 gfx906+sram-ecc-0 OpenCL compilation in 8.82 s
2020-09-05 02:09:33 gfx906+sram-ecc-0 54094109 OK        0 loaded: blockSize 400, 0000000000000003
2020-09-05 02:09:33 gfx906+sram-ecc-0 validating proof residues for power 8
2020-09-05 02:09:33 gfx906+sram-ecc-0 Proof using power 8
2020-09-05 02:09:34 gfx906+sram-ecc-0 54094109 OK      800   0.00%; 1012 us/it; ETA 0d 15:12; c9130b584a5c2453 (check 0.51s)
2020-09-05 02:10:04 gfx906+sram-ecc-0 Roundoff: N=29673, mean 0.246350, SD 0.014346, CV 0.058235, max 0.364949, z 17.7 (pErr 0.429960%)
2020-09-05 02:10:04 gfx906+sram-ecc-0 54094109 OK    30000   0.06%; 1001 us/it; ETA 0d 15:02; 8035402484acbe73 (check 0.53s)
2020-09-05 02:10:34 gfx906+sram-ecc-0 Roundoff: N=30475, mean 0.246477, SD 0.014465, CV 0.058689, max 0.376085, z 17.5 (pErr 0.523875%)
Prime95 is offline   Reply With Quote
Old 2020-09-05, 03:26   #2418
moebius
 
moebius's Avatar
 
Jul 2009
Germany

18F16 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Here is a gpuowl run on a Radeon VII.
Here is a gpuowl run on a Vega 64 availabale in Germany (maybe in EU) for 269,- EURO , but only with PCI Express x16 3.0 max Solution 2560 x 1600 only DVI

C:\Users\gesch\gpuowl-v6.11-364-g36f4e2a>gpuowl-win -prp 54094109
2020-09-05 05:07:11 gpuowl v6.11-364-g36f4e2a
2020-09-05 05:07:11 config: -user geschwen
2020-09-05 05:07:11 config: -cpu AMD_RXVega64
2020-09-05 05:07:11 config: -prp 54094109
2020-09-05 05:07:11 device 0, unique id ''
2020-09-05 05:07:11 AMD_RXVega64 54094109 FFT: 2.75M 256:11:512 (18.76 bpw)
2020-09-05 05:07:11 AMD_RXVega64 Expected maximum carry32: 50F80000
2020-09-05 05:07:12 AMD_RXVega64 OpenCL args "-DEXP=54094109u -DWIDTH=256u -DSMALL_HEIGHT=512u -DMIDDLE=11u -DPM1=0 -DAMDGPU=1 -DMM_CHAIN=2u -DMM2_CHAIN=2u -DMAX_ACCURACY=1 -DWEIGHT_STEP_MINUS_1=0xb.9e590d0a0083p-6 -DIWEIGHT_STEP_MINUS_1=-0x9.d559501736d8p-6 -cl-unsafe-math-optimizations -cl-std=CL2.0 -cl-finite-math-only "
2020-09-05 05:07:12 AMD_RXVega64 ASM compilation failed, retrying compilation using NO_ASM
2020-09-05 05:07:18 AMD_RXVega64 OpenCL compilation in 6.88 s
2020-09-05 05:07:19 AMD_RXVega64 54094109 OK 0 loaded: blockSize 400, 0000000000000003
2020-09-05 05:07:19 AMD_RXVega64 validating proof residues for power 8
2020-09-05 05:07:19 AMD_RXVega64 Proof using power 8
2020-09-05 05:07:20 AMD_RXVega64 54094109 OK 800 0.00%; 1035 us/it; ETA 0d 15:33; c9130b584a5c2453 (check 0.47s)
2020-09-05 05:11:01 AMD_RXVega64 Stopping, please wait


https://www.ebay.de/itm/ASUS-ROG-Str...l4275.c10#rwid

and a present for Mr. Ernst Mayer, available again by chance.
https://www.mindfactory.de/product_i...-_1296273.html

Last fiddled with by moebius on 2020-09-05 at 04:13
moebius is offline   Reply With Quote
Old 2020-09-05, 14:42   #2419
ATH
Einyen
 
ATH's Avatar
 
Dec 2003
Denmark

2,969 Posts
Default

Quote:
Originally Posted by Prime95 View Post
Here is a gpuowl run on a Radeon VII. Note the pErr values show there is a 1 in 200 chance of a 0.5 roundoff error during the entire PRP test. This roughly what Mihai and I were targetting (actually it is a bit higher as I never analyzed SMALL_HEIGHT=512).

Does your nVidia run show significantly different pErr values??
No, it is weird. I will do an entire exponent with 2.75M FFT and -use STATS on.

I had about 11 DC mismatches (not all confirmed yet) with P100 and V100 cards, and now none since I started using 3M FFT. All failures are >54M, I have done many DCs in 51M-53M without any issues.

Code:
root@cuda2:/content/drive/My Drive/test# ./gpuowl -prp 54094109 -use STATS
2020-09-05 14:35:09 gpuowl v6.11-380-g79ea0cc
2020-09-05 14:35:09 config: -use CARRY32,OUT_WG=64,OUT_SIZEX=8,OUT_SPACING=4,IN_WG=64,IN_SIZEX=8,IN_SPACING=4
2020-09-05 14:35:10 config: -prp 54094109 -use STATS
2020-09-05 14:35:10 device 0, unique id ''
2020-09-05 14:35:10 Tesla V100-SXM2-16GB-0 54094109 FFT: 2.75M 256:11:512 (18.76 bpw)
2020-09-05 14:35:10 Tesla V100-SXM2-16GB-0 Expected maximum carry32: 50F80000
2020-09-05 14:35:11 Tesla V100-SXM2-16GB-0 OpenCL args "-DEXP=54094109u -DWIDTH=256u -DSMALL_HEIGHT=512u -DMIDDLE=11u -DPM1=0 -DMM_CHAIN=2u -DMM2_CHAIN=2u -DMAX_ACCURACY=1 -DWEIGHT_STEP_MINUS_1=0x1.73cb21a140106p-3 -DIWEIGHT_STEP_MINUS_1=-0x1.3aab2a02e6dbp-3 -DCARRY32=1 -DIN_SIZEX=8 -DIN_SPACING=4 -DIN_WG=64 -DOUT_SIZEX=8 -DOUT_SPACING=4 -DOUT_WG=64 -DSTATS=1  -cl-unsafe-math-optimizations -cl-std=CL2.0 -cl-finite-math-only "
2020-09-05 14:35:15 Tesla V100-SXM2-16GB-0

2020-09-05 14:35:15 Tesla V100-SXM2-16GB-0 OpenCL compilation in 3.89 s
2020-09-05 14:35:15 Tesla V100-SXM2-16GB-0 54094109 OK        0 loaded: blockSize 400, 0000000000000003
2020-09-05 14:35:15 Tesla V100-SXM2-16GB-0 validating proof residues for power 8
2020-09-05 14:35:15 Tesla V100-SXM2-16GB-0 Proof using power 8
2020-09-05 14:35:16 Tesla V100-SXM2-16GB-0 54094109 OK      800   0.00%;  304 us/it; ETA 0d 04:34; c9130b584a5c2453 (check 0.23s)
2020-09-05 14:36:16 Tesla V100-SXM2-16GB-0 Roundoff: N=200098, mean 0.246357, SD 0.014325, CV 0.058147, max 0.365655, z 17.7 (pErr 0.415867%)
2020-09-05 14:36:16 Tesla V100-SXM2-16GB-0 54094109 OK   200000   0.37%;  304 us/it; ETA 0d 04:33; 40642cc2c4948d4e (check 0.22s)
2020-09-05 14:37:17 Tesla V100-SXM2-16GB-0 Roundoff: N=200900, mean 0.246390, SD 0.014403, CV 0.058455, max 0.411859, z 17.6 (pErr 0.471407%)
2020-09-05 14:37:18 Tesla V100-SXM2-16GB-0 54094109 OK   400000   0.74%;  305 us/it; ETA 0d 04:33; c24b99c4d1130b46 (check 0.22s)
2020-09-05 14:38:19 Tesla V100-SXM2-16GB-0 Roundoff: N=200900, mean 0.246320, SD 0.014335, CV 0.058195, max 0.382945, z 17.7 (pErr 0.420977%)
2020-09-05 14:38:19 Tesla V100-SXM2-16GB-0 54094109 OK   600000   1.11%;  305 us/it; ETA 0d 04:32; 68a6004748a738a2 (check 0.22s)
2020-09-05 14:39:20 Tesla V100-SXM2-16GB-0 Roundoff: N=200900, mean 0.246378, SD 0.014331, CV 0.058168, max 0.361628, z 17.7 (pErr 0.420930%)
2020-09-05 14:39:20 Tesla V100-SXM2-16GB-0 54094109 OK   800000   1.48%;  305 us/it; ETA 0d 04:31; 98ac6e527b6749ef (check 0.22s)
2020-09-05 14:39:37 Tesla V100-SXM2-16GB-0 Stopping, please wait..
2020-09-05 14:39:37 Tesla V100-SXM2-16GB-0 Roundoff: N=55738, mean 0.246389, SD 0.014440, CV 0.058606, max 0.371908, z 17.6 (pErr 0.499662%)
2020-09-05 14:39:37 Tesla V100-SXM2-16GB-0 54094109 OK   855200   1.58%;  307 us/it; ETA 0d 04:32; 3dfde62dfb22e13c (check 0.22s)
2020-09-05 14:39:37 Tesla V100-SXM2-16GB-0 Exiting because "stop requested"
2020-09-05 14:39:37 Tesla V100-SXM2-16GB-0 Bye

Last fiddled with by ATH on 2020-09-05 at 14:55
ATH is offline   Reply With Quote
Old 2020-09-05, 15:25   #2420
Viliam Furik
 
Jul 2018
Martin, Slovakia

229 Posts
Default

Quote:
Originally Posted by moebius View Post
...and a present for Mr. Ernst Mayer, available again by chance.
https://www.mindfactory.de/product_i...-_1296273.html
I checked the link. It seems that they only ship to some European countries, unfortunately not including Slovakia.
Viliam Furik is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
mfakto: an OpenCL program for Mersenne prefactoring Bdot GPU Computing 1657 2020-10-27 01:23
GPUOWL AMD Windows OpenCL issues xx005fs GpuOwl 0 2019-07-26 21:37
Testing an expression for primality 1260 Software 17 2015-08-28 01:35
Testing Mersenne cofactors for primality? CRGreathouse Computer Science & Computational Number Theory 18 2013-06-08 19:12
Primality-testing program with multiple types of moduli (PFGW-related) Unregistered Information & Answers 4 2006-10-04 22:38

All times are UTC. The time now is 07:38.

Sat Oct 31 07:38:30 UTC 2020 up 51 days, 4:49, 2 users, load averages: 2.28, 2.10, 1.93

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.