Moo! Wrapper

Good at maths? Then why not figure on helping these projects - Gerasim, GPUGrid, Moo! Wrapper, NFS, NumberFields, ODLK, PrimeGrid, Rakesearch, SRBase
Post Reply
UBT - JohnR
Posts: 391
Joined: Sun Apr 30, 2006 1:00 am

Moo! Wrapper

Post by UBT - JohnR »

New Project at http://moowrap.net/index.php

Trying to take the place of DNETC but the message boards are showing lots of problems at the moment.

I am unable to find our team at the moment.
Temujin
Posts: 2259
Joined: Mon Mar 13, 2006 12:00 am

Post by Temujin »

the team's there now :thumbup:

It's currently a GPU and windows only project, with apps for both NVidia and ATI
They have said linux apps will follow shortly

It pays massively stupid amounts of credit.
I'm not sure how long that will remain so, or even if the awarded credits will remain, the project is alpha afterall
UBT - JohnR
Posts: 391
Joined: Sun Apr 30, 2006 1:00 am

Post by UBT - JohnR »

Moo! Wrapper is very demanding of the computer.
On my AMD 4 core machine I have 2 ATI 5850's
Moo sometimes only uses 1 GPU, but to keep both GPU's working I can only run 2 tasks on the CPU, and I have to let Moo use the other 2 cores.
It is worth running task manager and GPUZ to monitor the computer.  If you have more the 1 GPU you can run GPUZ several times to monitor them all at the same time.
I have not seen the same problem with only 1 GPU in a computer.
Zydor
Posts: 437
Joined: Mon Nov 01, 2010 12:00 am

Post by Zydor »

Moo! Wrapper is heavy on CPU useage, but nonetheless I like it way above DNETC because the latter has always had issues with multiple GPUs - not in all cases of course, would be a dead project by now if that was the case.  But 58XX multiple GPUs on DNETC can get real iffy at times, and the wait states / WU crashes in crunching are very annoying.

I also intensly dislike the continual sniping by a DNETC admin at Moo! - he was at it again within hours of their return online yesterday. It is unprofessional and a few more descriptive expressions I could use. That reason in itself is enough to keep me away from DNETC, quite apart from the fact that performance and reliability of Moo is distinctly better, with an end result of better credit RAC from it than DNETC despite both using the same Core Application.  I am also finding that my crunching results over at DistrNet are around 10% higher - Moo is producing more StatUnit crunching per beast than DNETC.  All in all its a good job by Teemu at Moo, thumbs up to him.

Anyone switching from DNETC to Moo, or thinking of it, advantage is stability, and better output - the latter will make up for loss of CPU WU credits with the higher use of CPU by Moo.  Results at Moo are sensitive to CPU availability for GPU WU use, and the temptation to run a full set of CPU WUs inparrallel from another project should be avoided, it will slam Moo hard and not worth doing.

As will be apparent, I  like Moo - but horses for course as they say :)

Regards
Zy
Zydor
Posts: 437
Joined: Mon Nov 01, 2010 12:00 am

Post by Zydor »

Teemu has adjusted down the CPU assist per GPU from the old 0.78 CPU per GPU to 0.05 CPU per GPU without any detrimental performance hit.

That makes a big difference in BOINC Scheduling, and gives space now for all CPU Cores to be scheduled a CPU WU from the  Client Queue of WUs (from any Project in the Client queue). This removes what was a percieved big disadvantage for some.  So if you were in that category, worth giving it a spin again to see if it suits.

The reliability of Moo is not yet approaching the standard of Collatz - but  its pretty damn close, its excellent. Teemu is doing a really good job.  The lack of posts on the Forum is testiment to zero hassles, as happens at Collatz where days can go by before a new post is seen.  All in all the output of Moo is now way above what DNETC used to be, mainly  due to reliability and excellent up time, but also the StatUnit output to DNET is consistently around 10% higher than DNETC was, and now also consistently exceeds PG ATI PPS Sieve app for BOINC Credits.  (NVidia should stay with PG, nothing can touch those credit levels they give out for NVidia)

No doubt there will be issues come along, all Projects get them of course, not claiming Moo will not, however its as solid as a BOINC Project gets, and Teemu carefully tests changes before throwing them out there  .... thats Bliss  :wink:

Regards
Zy
Joshrandom
Posts: 5602
Joined: Sat Jun 23, 2007 1:00 am

Post by Joshrandom »

Just thought I'd point out that Moo also allows those that crunch it to have the credits that are earned over at DNET awarded to an account over which the user has full control. This means that it is possible to earn credits for the team over on Moo, and then earn the team credits on DNET for the same completed work units.  :D

For those that are interested, this is the link to the stats for the team that NaRyan set up for us over at DNET. 8)
Joshrandom
Posts: 5602
Joined: Sat Jun 23, 2007 1:00 am

Post by Joshrandom »

Due to some no doubt unforeseen problem :wink: , Moo appears to be granting double credits just at the moment, so get them while you can. :santa:
UBT - mickyb69
Posts: 1655
Joined: Wed Sep 19, 2007 1:00 am

Post by UBT - mickyb69 »

Only just seen your post  :lol:
4100 credits for 20 mins on an old HD5600 ati gpu
Happy days   :wink:
UBT - Mikee
Marvin the Dalek
Posts: 4395
Joined: Wed Mar 15, 2006 12:00 am
Location: North Wales

Re: Moo! Wrapper

Post by UBT - Mikee »

Hi

So I was wandering about and stumbled on this page http://stats.distributed.net/projects.php?project_id=8
Seems there's enough work up to 134 years - though the key may be cracked any time before then.

Just as a bit of encouragement there's a decent prize for the discoverer of the key of $10,000 - $1000 of which goes to the discoverer and the remainder split - see http://www.distributed.net/RC5

Mike
Follow us on Twitter... http://twitter.com/UKBOINCTeam

Image Image

Image
UBT - Timbo
UBT Forum Admin
Posts: 9673
Joined: Mon Mar 13, 2006 12:00 am
Location: NW Midlands
Contact:

Re: Moo! Wrapper

Post by UBT - Timbo »

UBT - Mikee wrote:Hi

So I was wandering about and stumbled on this page http://stats.distributed.net/projects.php?project_id=8
Seems there's enough work up to 134 years - though the key may be cracked any time before then.

Just as a bit of encouragement there's a decent prize for the discoverer of the key of $10,000 - $1000 of which goes to the discoverer and the remainder split - see http://www.distributed.net/RC5

Mike
Hi Mike

Nice find...though I'm surprised that 7 members are still participating, according to the website.

I was running this, until a PC crash a couple of years ago, and I had to rebuild everything...but I never got around to re-installing dnetc...:-(

regards
Tim
Post Reply