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 : 2015 PRPNet January Challenge (Message 82938)
Posted 2808 days ago by zzcmeyer
I feel like saying....what the? :)

I completed more work units but zero near wieferichs.

Bummer :(

-- Craig
2) Message boards : Project Staging Area : 2015 PRPNet January Challenge (Message 82871)
Posted 2810 days ago by zzcmeyer
Hi
given up on NV cards and are move one ATI card into most boxes

Have hit the sweet stop with one 290x running @ 4.5m p/sec doing 4 threads in under 10 minutes. that's 24 per hour.
those other ATI cards generally are running over 2.1m to 3.1m p/sec and are doing 8 threads per 20 mins same 24 per hour.

NV cards may work hard but none of them can get within that sort of output
24 per hr.

Ross*


I don't want to get into an ATI v NV war here...

With my NV-Titan, I can get 1 per 35secs, which equates to about 100/hr.

Unfortunately my cpu died. So the titan is off wwwwcl as I can't drive it to 100% with the cpu I currently have.

-- Craig
3) Message boards : Project Staging Area : 2015 PRPNet January Challenge (Message 82826)
Posted 2811 days ago by zzcmeyer
As for selecting the optimal settings in wwww.ini. I just found settings that sent my GPU to 95-97% and thought "good enough".

I found that I needed all 6 "real" cores of my i7-3930k to drive my titan 100% with wwwwcl. I assume the program is cpu/gpu hybrid. It seem you need enough cpu resources to drive the gpu to high usage.

Also I ran 2x instances to cater for the drop when the code stops to get more work.

I also found that this setting needs to be set >= number of cores.:

threads=x
4) Message boards : Project Staging Area : 2015 PRPNet January Challenge (Message 82714)
Posted 2813 days ago by zzcmeyer
My guess is the parent executable (prpclient) calls wwwwcl and uses 1000. On linux with the ps command you can see this option being set:

[ec2-user@ip-172-31-13-232 prpclient-4]$ ps -aef | grep www[w]cl ec2-user 32248 31644 98 12:47 ? 00:02:43 ./wwwwcl -TWieferich -s1000 -p370229200000000000 -P370229300000000000 ec2-user 32259 31512 97 12:47 ? 00:02:35 ./wwwwcl -TWieferich -s1000 -p370229900000000000 -P370230000000000000
5) Message boards : Project Staging Area : GFN524288 GPU b limit (Message 81735)
Posted 2839 days ago by zzcmeyer
Package name:
prpclient-5.3.1-linux_64.zip

$ file genefer genefer: ELF 64-bit LSB executable, x86-64, version 1 (GNU/Linux), statically linked, for GNU/Linux 2.6.24, BuildID[sha1]=f354c4949715c4f31d944e4fd236f9170fc01650, not stripped


Looks like 'genefer' is the 64bit build.

All lines beginning with "gen":

$ grep ^/*gen prpclient.ini geneferexe=./geneferocl //geneferexe=./genefercuda geneferexe=./genefer
6) Message boards : Project Staging Area : GFN524288 GPU b limit (Message 81733)
Posted 2839 days ago by zzcmeyer
Aaah, that makes sense. Whoopsie.

I had this in prpclient.ini:

geneferexe=./geneferocl //geneferexe=./genefercuda //geneferexe=./genefer


It's now:

geneferexe=./geneferocl //geneferexe=./genefercuda geneferexe=./genefer


Lets see how it goes. In the linux 5.3.1 distribution it didn't have genefer64.
7) Message boards : Project Staging Area : GFN524288 GPU b limit (Message 81728)
Posted 2840 days ago by zzcmeyer
I'm using the g2.2xlarge nodes on Amazon's EC2.

I'm winding down. I did have a number of them, but since they started to all error out to cpu now, I've dropped them all off.

But I have 1x VM remaining, with 3x instances running. The VM is effectively 4x cores (x2 for hyperthreading). Total of 8 cores as seen by the OS.

Top excerpt:

top - 22:36:38 up 22 days, 10:32, 1 user, load average: 3.00, 3.00, 3.04 Tasks: 106 total, 4 running, 102 sleeping, 0 stopped, 0 zombie Cpu(s): 0.0%us, 5.8%sy, 87.8%ni, 6.4%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st Mem: 15404752k total, 803500k used, 14601252k free, 143420k buffers Swap: 0k total, 0k used, 0k free, 194596k cached PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 17765 ec2-user 39 19 124m 95m 1164 R 100.0 0.6 9466:02 pfgw64 15689 ec2-user 39 19 124m 95m 1168 R 99.5 0.6 10632:48 pfgw64 16098 ec2-user 39 19 124m 95m 1176 R 99.5 0.6 10408:43 pfgw64 1 root 20 0 19488 1600 1288 S 0.0 0.0 0:02.04 init



Even only 3x cores are used, I left it over night and it's still going (at a rate roughly 1/wk).

Thoughts?


8) Message boards : Project Staging Area : GFN524288 GPU b limit (Message 81705)
Posted 2841 days ago by zzcmeyer
If I could ask, that the time allowed to complete be increased on this?

If my gpu can complete it, the task gets done in a few hours.

But if it drops to CPU, it takes a little over a week and then the work becomes redundant.

-- Craig
9) Message boards : Project Staging Area : http://uwin.mine.nu/ / latest binaries (Message 80873)
Posted 2876 days ago by zzcmeyer
Thanks I've installed them.

But I'm getting this error:

$ ./prpclient
Could not determine version of PFGW being used. Missing version data

Running pfgw on its own I get:

$ ./pfgw64 -V
PFGW Version 3.7.7.64BIT.20130722.x86_Dev [GWNUM 27.11]

Error opening file helper.txt

Any ideas?
10) Message boards : Project Staging Area : http://uwin.mine.nu/ / latest binaries (Message 80861)
Posted 2877 days ago by zzcmeyer
Hi,

I'm having trouble connecting to uwin.mine.nu/PRPNet/ for the last 8hrs+ or so. I'm after the latest prpnet binaries to install on a new box.

Is this site coming back soon? Or is there a active mirror?

-- Craig


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2022 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 2.07, 1.45, 1.03
Generated 6 Oct 2022 | 3:46:41 UTC