EggXpert

Tekran:

I have a dilemma I need help with.  I am trying to figure out what will be the best processor/mobo combo to go with when I upgrade.  My focus will be on getting back to folding.  I have 3 oiptions I see that I can do but I am not sure which would be best in price to performance as well as what would be best from the viewpoint of reduced power comsumption.  I'll list each option below and then ask more specific questions:

Option 1:

Using my current build in my sig, I can upgrade to one of the BD processors and only need to replace the processor.  I have heard that power consumption of the BD can be high when under load though so that could result in high electrical usage.  With this option, I would not need to get another water cooling option as my H70 would be compatible due to the socket being the same.

Option 2:

Ivy Bridge when it comes out.  Every thing but the motherboard and processor could be reused from my existing build as my RAM is already 1.5v like SB requires.  Since Ivy Bridge is mainly a die shrink, it likely will use the same RAM specs.  Ivy Bridge is designed for reducing power compsumtion and heat but I don't know how it will affect performance for F@H and overclocking.

Option 3:

Sandy Bridge-E when it comes out.  I would need to replace the CPU, motherboard, RAM, and possibly my H70 with this option.  My other components can be carried ovedr from my current build.  Performance I believe may be the best with this option but I am not sure what the potential electrical usage may end up as.

 

For all three options, I will be using a Crucial M4 256 GB SSD with them.  I just received it yesterday and will be installing it this weekend.  I will change my current RAID 0 to a RAID 1 array after I get my apps reinstalled.  The RAID 0 will become my data storage drive.  Any thoughts or ideas on what the best option is will be greatly appreciated.

 

A bit of clarification.  Part of the reason I am looking at parts upgrades is because I have been having BSODs more and more frequently lately but I have not been able to narrow it down to a specific cause.  Most of the errors have been 0x7A errors which are fairly generic.  The only one I ahve been able to narrow down further was the 0xF4 error I got last week which is when I decided to go ahead and get the M4 SSD since it was on sale.  Hopefully, it is just instability in my current RAID array that is causing the BSODs.  By using the SSD as a boot drive, I'll be able to eliminate the array as a potential issue if the problems persist.  If they do, I will be looking at going with either option 2 or 3 depending on performance to cost evaluations.  If the BSODs are resolved after I switch to the SSD, I will have to look at whether BD is feasible for me to get since I already have the 1090T.

Tekran:

Most of the errors have been 0x7A errors which are fairly generic.  The only one I ahve been able to narrow down further was the 0xF4 error I got last week which is when I decided to go ahead and get the M4 SSD since it was on sale.

0x7A is BSOD due to corrupted virtual memory / swap data...

0x74 is BSOD due to corruption in the operating system's memory area... Memtest should pick that up if it's caused by memory instability.. Otherwise it could be coming from your virtual memory / swap data if you have one..

 Edit: If I were you, I'd run a full scandisk (including sector scan)..   One of the drives might be making bad sectors...

The second one was 0xF4, not 0x74.  When I looked up the 0xF4 error, it indicated hard drive as a common issue.  That is part of why I am switching to the SSD, I doubt the hard drives are bad but the RAID 0 array may be unstable.  Going to run hard drive tests on them after I get the SSD set up so that I can check the SMART status on them as I can't check that in the array.  Planning to format the drives in the array before I change over to RAID 1 although I am thinking about not using an array at all.  I back my data up regularly enough that I would not lose much if I lost a drive so mirroring the drives won't benefit me much.  I'll run scandisk as well to check the drives after I can take the drives out of the array.

Tekran:

Going to run hard drive tests on them after I get the SSD set up so that I can check the SMART status on them as I can't check that in the array.  

Your RAID drivers should be able to check the status...  There should be a Windows component of your RAID drivers that gives you a RAID access panel.. For AMD chipsets, it's basically a webpage on a loopback port..   You can check the SMART status of the array and set up new arrays from the page.  I used to have an AMD RAID on my primary gaming PC for a while before I bought my nvidia 900 series chipset.  My other PC runs a 3ware card.

I am trying to remember what program I was using to try to check the status, I'll see if I can find the place to check the status using Windows when I go home at lunch time.

I just remoted into my home computer from work and according to AMD's RAID expert program, the SMART status for both drives in the array is Healthy.

Well, this has been an interesting few days of learning.  I was able to get my OS and apps installed on my SSD Thursday night.  I reconnected my HDDs Friday morning and immediately after I did so, I started getting a message that the BOOTMGR was missing.  I ended up having to reinstall the OS again.  I then tried to use audit mode to transfer the Users folder to a drive other than the SSD.  I received a fatal error every time I tried.  I then had to reinstall a third time and I was able to use the directions I found online to move the Users folder to one of my HDDs so I ket the space free on the SSD.  I got everything installed successfully but as I went to close the case back up, I discovered what I think was the source of my problem with my BSODs.  I have the Coolermaster HAF 932 Blue case and there is an 8 pin extender cable for the CPU that comes with the case so that the EPS/ATX 12v 8-pin cable can be routed on the back of the motherboard tray.  Something had caused the plastic to turn brownish on the extender cable and on the PSU 8 pin cable, the connector had what appeared to be melted plastic (not badly melted but like something caused a temporary heat surge in the adapter).  I had to go buy a new PSU and adapter cable today so I am finally back up and running.  I am holding off F@H for a few days to I can confirm my system is stable now.

Well I finally got everything straight and am folding again, had to reduce the OC to 3.8ghz and the ram at 1080 using 100% of the CPU and it still only uses 25% of the Ram but I'm getting bigadv wus worth 12,000+ ppd which is 10 times better than what I was getting with the GPU folding so hopefully this will help the team climb back up.

So that I can start to figure out what will perform the best for me, I have 2 options I can see right now for folding.  First, I can do GPU and CPU folding but I am not sure is I can assign a specific core to each GPU and then  use the others for -bigadv.  The second option is to just go for -bigadv if I  understand it correctly and not even worry about using the GPUs for folding.  Which option will bgive me the best performance and allow me to complete WUs quickly?

ETA:  Right now, I am set for smp for the method and I have added both gpus.  The exact info for the smp is smp:6 since it is set to use all cores currently.  I have not added any switches to set specific options yet.  Just test to make sure I can complete work units successfully right now.

What is the current core type for F@H?  For my GPUs, I am getting 0x11 which I thought was the old core type prior to 0x16.  For SMP, I am getting 0xa4.  Not sure what the current core info should be.

Figured out the core issue for GPUs while hunting on the F@H forums.  Needed to set the option for client-type advanced on each of my GPU slots.  Also realized I needed to free up a cpu core for each gpu client running.

If you are running just smp clients for the CPU folding A4 core is right, as for GPU folding I stop long time ago due to ineffiicient in energy.  -bigadv unit will use A5 core, if you are doing cpu -bigadv + 1 gpu client you probably yield just about the same ppd as a -bigadv on its own and you are using much more power with cpu + gpu vs a cpu client.

The trick is that all the new A4/A5 cores clients calculate points based not on their base points, but rather the bonus point.  For example, the WU7000-7028 I meant earlier are only 600 credits each but when I upload it before the deadline I will get bonus point based on my time.  Comes down to folding time, the less time it takes you to complete a client the more bonus points you get.  Just to use an example, Hog get 2299 points from WU7015 and he used about 13hrs to complete, however I run the same client and I use about 3hrs and 20mins to complete now my upload points is around 4700 points it doubles his points and use much less time to upload. 

Some of the huge -bigadv units like the P6903/P6904 people dropping off well over 350k-800k ppd per WU depending on the completion time, so if you are dropping off 350k over 2 days time you really don't need to waste more electricity on gpu folding that yield you 10-20k ppd.

There's a whole bunch of different cores depending on what your hardware can do..

 I think there are 2 cores for nvidia (8800 core & Fermi core), 1 core for AMD graphics, at least 1 core for SMP and another core for the classic client.

 

Ok, it looks like I should switch to straight -bigadv then.  Can you walk me through what I will need to do to set the client up for -bigadv?  Running both GPUs doubles my wattage used so if I can do more folding with less power consumption, that works for me.  I guess I hadn't really considered how the bonus points would factor in.

To get estimate on the bonus point use the bonus point calculator.  Just select the WU number and enter the approximate time to complete one frame and you should able to calculate the points.

As for the guide, there are two version one yields better result but is running Linux Unbuntu so I guess not many people going to do that.  I would link you to the windows guide of -bigadv.  Just a note, I heard the -bigadv units are heavy on the cpu so doing other tasks while doing the -bigadv can be a hassle (expect lags).

To get bonus point you would need a passkey from folding@home page and you need to complete 10 WU with that passkey in order to start getting the -bigadv units.

Click here.

 

Yo team mates you seen the new version of F@H they have coming out?

http://www.stanford.edu/group/pandegroup/cgi-bin/edit/English/WinGuide

https://fah-web.stanford.edu/projects/FAHClient/#Documentation

Great find Hog, I probably start folding later this year or early next year since I got something to do at the moment.

Well I just downloaded it to MommaMary's rigg and she is folding again too so just maybe we can do some catching up again with the new F@H 7.1.

Anyone else experiancing hard core errors, lock-ups on DUAl GTX4xx setups on the GPU3 model?

Can you possibly link me to the GPU2 updated HP version I can't find it anywere (made for fermi)