The 2P/4P Owners Club

[flibberdipper]

I feel so underpowered and badgeless now.

[Whaler_99]

What do you think the performance is in Gflops of one of your folders?

 

I have no clue. Everything is detailed, so feel free to figure it out.

[tobben]

just out of curiousity, are any of you using your multi processor machines for anything besides folding?

 

in my case im using it as a makeshift centralised storage/backup solution for everyone in the house, it's the only pc that's on 24 hours a day, and not being used for anything else so the data is pretty secure.

[Judahnator]

just out of curiousity, are any of you using your multi processor machines for anything besides folding?

 

in my case im using it as a makeshift centralised storage/backup solution for everyone in the house, it's the only pc that's on 24 hours a day, and not being used for anything else so the data is pretty secure.

 

mine is a webserver, but nothing too exciting

[ray_p1]

just folding for now once i get better networking for my house then i want to do a media server.

[tobben]

has everyone had a look at this *http://folding.stanford.edu/home/changes-to-the-bigadv-threshold/*

 

it's kinda important .. whaler made a thread about it *http://linustechtips.com/main/topic/89924-4p-changes-coming-the-requirements-for-bigadv-increasing/*

 

considering the fact that almost all of our 4p systems are socket f this kinda hurts..almost seems to me as if they just got more bigadv folders than expected so they are trying to trimm off some of the fat.. would have thought there were few enough as it is compared to normal smp folders, it's not like they will be gaining smp folders by doing this. 

[Whaler_99]

It's funny, but it turns out the assumption is they may have to many - so trimming the fat. Seems like lately Pande really seems to be making the push to get folks to move to GPU folding.

 

Tear over at [H]ard though is trying to figure out if they are just changing core count requirements or the deadlines as well. If they just modify the core count requirements, speculation is there may be a work around. If they change the deadline though, like they did when going from 8 to 16, then that will make it a lot harder, as that isn't something we can fake...a 24 Core socket F can just complete most of the projects. If they tighten deadlines, would make it impossible.

[Jumper118]

has anyone tried running cinebench r15 on a dual or quad cpu set up? i am very interested to see what score dual xeons or quad opterons will do 

[FireFox]

I really want to do a dual 6 core Intel Gulftown system but looking at the AMD's I think I'd be wasting my time lol

[Whaler_99]

has anyone tried running cinebench r15 on a dual or quad cpu set up? i am very interested to see what score dual xeons or quad opterons will do 

 

I believe this is a Windows based app, where as our Folders are Linux based with zippo GPU/Graphics drivers. As well, the AMD Opteron are really around number crunching and this looks to try and eval the graphics capabilities of CPU's and GPU;s?

[flibberdipper]

I believe this is a Windows based app, where as our Folders are Linux based with zippo GPU/Graphics drivers. As well, the AMD Opteron are really around number crunching and this looks to try and eval the graphics capabilities of CPU's and GPU;s?

Whaler! *bear hug* You're back!!

[Jumper118]

I believe this is a Windows based app, where as our Folders are Linux based with zippo GPU/Graphics drivers. As well, the AMD Opteron are really around number crunching and this looks to try and eval the graphics capabilities of CPU's and GPU;s?

yes, i know mutli cpu folding rigs mostly run on linux, but does if anyone was running something like windows server 2008-maybe dual boot? the benchmark is rendering of an image and does scale very well multiple threads and clock speed. i think opteron's and xeon would score well. negative b has try it with dual six core xeon 's, but even though they got the highest score by quite a long way, they are 1366 which is fairly old. something like haswell or ivy 8 core xeons and 12-16 core opteron's should be beast when there are 2 or 4 of them.    

[tobben]

yes, i know mutli cpu folding rigs mostly run on linux, but does if anyone was running something like windows server 2008-maybe dual boot? the benchmark is rendering of an image and does scale very well multiple threads and clock speed. i think opteron's and xeon would score well. negative b has try it with dual six core xeon 's, but even though they got the highest score by quite a long way, they are 1366 which is fairly old. something like haswell or ivy 8 core xeons and 12-16 core opteron's should be beast when there are 2 or 4 of them.    

well... it probably won't scale that far and might not even work, rendering images isn't really something i want to do, especially with my all powerfull Integrated Matrox G200eW Graphics, there is software aimed towards benchmarking multi cpu setups, be it a 4p machine or several machines chained together, but i can't really be bothered with it, not to mention the most important thing! you would have to stop folding to run it, sooo noooope , for the most part benchmarking just isn't a priority to the people who build these machines.

[Jumper118]

well... it probably won't scale that far and might not even work, rendering images isn't really something i want to do, especially with my all powerfull Integrated Matrox G200eW Graphics, there is software aimed towards benchmarking multi cpu setups, be it a 4p machine or several machines chained together, but i can't really be bothered with it, not to mention the most important thing! you would have to stop folding to run it, sooo noooope , for the most part benchmarking just isn't a priority to the people who build these machines.

I know what ya mean but it would take like 5 seconds literally. obviously you don't have to but it would be good if you did 

this is on the older version with a 48 core rig

[Whaler_99]

I know what ya mean but it would take like 5 seconds literally. obviously you don't have to but it would be good if you did 

this is on the older version with a 48 core rig

 

 

Ahh, not 5 seconds. Most of us would need to shut the systems down, grab a spare drive, install Windows Server (do you have a licenses and install media?) get all the proper drivers and updates installed, as these could also affect performance. I would suspect would take me at least a day to get Windows Server up and tuned on one of my systems for something like this, and unfortunately I simply don't have the time for this right now. If this is so popular, I am surprised there are not more tests out there with high core counts.

[rvborgh]

Yes, i did some benchmarking when i was doing BIOS tuning on my quad opteron setup at 3.0 GHz.  

CB 11.5

and CB R15

i hope folks don't mind the noise... that was back when i was still waiting on 2 additional Noctua HSF for the top two processors.

 

EDIT: processors are 4 x Magny Cours ES (48 cores - unlocked clock multipliers) running in a SuperMicro H8QGi-f.  These are totally stable at 3.0 GHz... and 1.175 volts.  i can run it higher but wattage at 3.1 GHz slightly exceeds 1000w at times when running the benchmarks.



 

has anyone tried running cinebench r15 on a dual or quad cpu set up? i am very interested to see what score dual xeons or quad opterons will do 

[79 Potatoes]

I've 4P! The mb is my table :C

 

Yay! It was a lot of time since the last time I heard a jet ^^

[flibberdipper]

Well I should be getting a dual-PIII board... If I fold once does that count?

[Judahnator]

Well I should be getting a dual-PIII board... If I fold once does that count?

 

I see no reason why it wouldnt.

 

I did have a hilarious mental image though. Like one of those anti-drug commercials, "Folding, not even once!"