What are you folding with?

Just for fun(?) I installed VirtualBox on my 2600K Windows 7 machine, installed Ubuntu, hacked Ubuntu to make it look like it had 16 CPU's and fired up F@H configured for BigAdv WU's.

Even though it took 4 days to process the 1 WU it worked out to just over 60K PPD. So my 2600K that was getting me anywhere between 15K - 30K PPD is now puling down about 60K PPD.

And I use this machine for other things too.

How would one implement that hack?

If i'm not mistaken - you could set the 16cores in Virtualbox (right now i could set 8cores in an i5-2540M), and the set the fah cores again inside ubuntu. Bleeder - can you shed some light please :)
 
Here is a guide.

http://www.overclock.net/t/1048365/ubuntu-setting-up-big-bigadv-p6903-for-dedicated-2600k

F@H Linux version 6.34


Bleeder, Is there any guide for version linux 7.1.52 f@h?


A greeting.

This is the guide I followed. But with the following changes:

-I used Ubuntu 11.10 instead of the 10.10 as was directed in the guide. But still followed the partitioning recommendations of the guide. Not sure if this matters but I've read that 10.10 is better for folding but it's also not supported by Ubuntu any more.

-I installed the F@H v7 client instead of the v6. And skipped ALL sections that configure "Langouste decoupler" because v7 does this for you already (downloading the next WU's while the current is still processing). Also, skip ALL sections that detail any configuration of the F@H v6 client including the creation of the "fah2smp.sh" file.

-The steps that you create and configure the corefix.sh file is where the "core Hack" comes in. The only change here is I specified 16 cores. This is different than setting Virtual Box to 16 core if you only have 8. Doing that will KILL performance as Virtual Box will actually try to use 16 cores. The core hack just makes the F@H client believe you have 16 cores.

-I configured the v7 client's SMP slot with "client-type=bigadv" and "max-packet-size=big" although I'm not really sure that max-packet-size is needed.

Don't forget to reboot after doing all this so the corefix.sh can startup. Let it run for a few WU's as it took my machine 3 or 4 WU's before it finally grabbed a 6903.

Hope this helps... :)
 
If i'm not mistaken - you could set the 16cores in Virtualbox (right now i could set 8cores in an i5-2540M), and the set the fah cores again inside ubuntu. Bleeder - can you shed some light please :)

At first I did try setting VirtualBox to 16 cores as that would be way easier but doing that brought my machine to it's knees. I'm guessing that it's because of the additional overhead of VirtualBox trying to schedule 16 virtual threads running at 100% on just 8 threads of actual hardware.
 
Finally have two laptops up and running. And soon I'll have a new dedicated tri-SLI 680, i7 3930K rig up sometime next week. Happy folding!
 
Finally have two laptops up and running. And soon I'll have a new dedicated tri-SLI 680, i7 3930K rig up sometime next week. Happy folding!

Sweet bejezus! You trying to get into the 100K PPD club? Plenty of room... all are welcome! :lol:
 
Finally have two laptops up and running. And soon I'll have a new dedicated tri-SLI 680, i7 3930K rig up sometime next week. Happy folding!

I'm waiting to be able to fold on my 680.. Doesn't seem like everything for that is ironed out yet.
 
Well damn! Hopefully Stanford and nVidia can come up with a way to take advantage of the Kepler architecture pretty soon. I'd figured they had a solution for it already. Oh well, Mass Effect 3 and X3 wont play themselves anyway.

I'm hoping to finally be a serious folder! Finally! Have to make up for all that lost time...
 
Finally have two laptops up and running. And soon I'll have a new dedicated tri-SLI 680, i7 3930K rig up sometime next week. Happy folding!

Hey Kai... wondering how that 3930K build is coming along? Got it running yet? Like it? Have you tried to Fold with it? :)
 
Last edited:
At first I did try setting VirtualBox to 16 cores as that would be way easier but doing that brought my machine to it's knees. I'm guessing that it's because of the additional overhead of VirtualBox trying to schedule 16 virtual threads running at 100% on just 8 threads of actual hardware.

Thanks for pointing that out - i also thought it was the easiest way to fold on 16cores :D

btw, thanks for the linux tips and the 16core hack :fold:
 
Hey Kai... wondering how that 3930K build is coming along? Got it running yet? Like it? Have you tried to Fold with it? :)

Glad you asked! I just got it running a few hours ago, although the 680's are getting around 17k a piece right now. But hey, anything is better than nothing! I have the 3930 OC'd to to 3.8 an it's getting aroudn 30k at the moment. Not too shabby without an optimized kepler core.
 
Glad you asked! I just got it running a few hours ago, although the 680's are getting around 17k a piece right now. But hey, anything is better than nothing! I have the 3930 OC'd to to 3.8 an it's getting aroudn 30k at the moment. Not too shabby without an optimized kepler core.

Very nice! :up: So about 64K PPD right out of the gate and it's not even tweaked yet . Can't complain about that at all.

Have fun with that new rig man and it's great to see some really big additional points coming in for the team. :D
 
Back
Top