Folding@Home: THGC Needs You -Team 40051

[beanoslim]

Now setting my card up on a newer mobo to see if any better, think I'll stick with 14.4 for now.

 

[EXT64]

Looks like Core 17 will now work in Linux for AMD GPUs too (it has for a while for NVidia). If anyone is interested, it requires the beta or advanced flag.

 

[beanoslim]

Picked up a Sabertooth 990FX to see if card reacts any differently with regard to usage dips and spikes.

Its exactly the same, but it does have 4xPCI-e x16 slots so room for expansion 😉

 

[JPNpower]

Are my eyes screwed up or does one guy on the Curecoin team have the same PPD as the entirety of [H]?

 

[EXT64]

Yep, of course [H] is nowhere close to its prime as many switched to boinc, but still impressive. Assuming Curecoin gets some value, expect other high PPD members to show up. An AMD GPU Linux core is now in testing and available, so those miners with 200-300 AMD GPUs can now get in the game.

 

[JPNpower]

Excuse my ignorance, but why the sudden trend shift to BOINC? money again? The only benefit I see for it is that I can (if I want) run it on my Android.

And 300 AMD GPUs!? No wonder the crazy PPD... That's what, 4 GPUs per MB = 75 MBs... dang. I want to see what kind of gear/electricity/heat/econimmics/space management this guy does, AND only in his spare time. Impressive.

 

[icraft]

One day I want to run over 1000 dual GPU cards in a giant rack mount server farm. I'd have to be a millionare, but then I could actually cool them all. 😛 I'd fold for my team, but I could also do so much. Get 1 work unit done each and then move to another team. I'd be the ppd fairy! 😀

 

[beanoslim]

^Someone's been on the moonshine :lol:

 

[EXT64]

:lol: At least it was on a Saturday night.

The shift to boinc at [H] was because they were upset with the management of fah (around the time bigadv was cancelled) for their standard method of ignore/avoid problems rather than address and explain (and the actions of official fah forums mods at the time).

Let me see if I can find one of the pics. I can't find the original thread, but here is one of the setups. And for cooling, it is fans, lots of fans :lol:

http://www.hardocp.com/image.html?image=MTM5MjgxODY2Mzg0aWxaekVGOFVfMV8xX2wuanBn

 

[Unolocogringo]

Got the 2p and 650ti moved to the basement and they are all runnimg 5 to 7 degrees cooler.
Came across a mod for removing the back of a 8x slot so a 16x video card will fit in it. Use a soldering iron to slowly melt it away. Was afraid to use a hacksaw blade or dremel on a 2p board but the slow melt dosen't endanger the board or surface mounted bits.

Might give one slot a try on my next breakdown for cleaning. Just wanted to get the 2 moved and up and running for now.

 

[beanoslim]

You could just use an 8x to 16x pci riser, get them on eBay, I picked one up that runs from any slot, 1x, 2x, 8x etc.

 

[JPNpower]

That man alone probably drives the PC enthusiast GPU/MB/PSU economy. Dang!

I've actually seen something similar in person though. (not as big scale). The cacophony of various fans minding their own business was... unique.

 

[EXT64]

Folding does take a hit with the very small PCI-E slots (1x, 2x, don't know about 4x) so you will want to get an 8x riser if you get one.

 

[Unolocogringo]

Folding does take a hit with the very small PCI-E slots (1x, 2x, don't know about 4x) so you will want to get an 8x riser if you get one.

From my experience 4x is a little slower than 16x.
From 3 to 8 seconds per frame depending on the work unit.

 

[beanoslim]

Made absolutely no change to my ppd but I was running it with a x1 to x16 powered riser kit.

 

[Unolocogringo]

I have a pair of GTX 460 clocked the same. One on a 16x slot the other on a 4x slot. Depending on the work unit the one in the 4xslot is a little slower.
Not sure if it is secondary memory timings on the cards or the 4x slot .
But it is definitely a little slower.

 

[EXT64]

Yeah, I have no evidence. Might even depend on the Core and Project. I did see some people post numbers showing 1x slowing the card down, but obviously that wasn't the case for beano. Might also depend on your CPU and PCI-E configuration (latency).

 

[Unolocogringo]

Mine exhibit the same on Core 15 or 17 units. Although I have not swapped the cards to test. On core 15 units I get a steady 99% usage in Afterburner and GPU-Z.
Core 17 shows a lot of drops to 96.99.98.99 98.99.99.99.99.99.99.96....... with a little variation in the pattern occasionally. So I can not say a definite on them on core 17. But I do get slightly higher points for similar work units when ran at the same time.
Update

Just finished a pair of 13000
16x=38852 points
4x=38801 points

 

[beanoslim]

Maybe the points difference was so minimal I didn't register that it was regularly less.

Lucky my new board has 4 pci-e x16 slots

 

[beanoslim]

Anyone know how to stop the web page opening on the new client and going straight to advanced control?

 

[anti-painkilla]

I have it open in the system tray and only open it when I need to, then select the Advance Control.

 

[Unolocogringo]

My solution.

Right click task bar icon
Advanced Controll

Expert tab.
Edit
Remove the line Web control= True.

Close and save.

Now start all programs
Folding @ home folder.
Find the F@H control Icon and drag to desktop.
Create shortcuts here.
Now you have an Icon on your desktop for F@H control Advanced.

 

[beanoslim]

Thanks beloved patriot rick

Anyone else getting bad wu's?

Could my OC be the cause?

07:44:42:WU01:FS01:0x17:Completed 2200000 out of 2500000 steps (88%)
07:44:42:WU01:FS01:0x17:Bad State detected... attempting to resume from last good checkpoint
07:44:42:WU01:FS01:0x17:Max number of retries reached. Aborting.
07:44:42:WU01:FS01:0x17:ERROR:exception: Max Retries Reached
07:44:42:WU01:FS01:0x17:Saving result file logfile_01.txt
07:44:42:WU01:FS01:0x17:Saving result file log.txt
07:44:42:WU01:FS01:0x17:Folding@home Core Shutdown: BAD_WORK_UNIT
07:44:42:WARNING:WU01:FS01:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
07:44:42:WU01:FS01:Sending unit results: id:01 state:SEND error:FAULTY project:9102 run:21 clone:2 gen:56 core:0x17 unit:0x0000003c0a3b1e81537c104be337f7d1

06:46:26:WU01:FS01:0x17:Completed 1750000 out of 2500000 steps (70%)
06:46:26:WU01:FS01:0x17:Bad State detected... attempting to resume from last good checkpoint
06:46:26:WU01:FS01:0x17:Max number of retries reached. Aborting.
06:46:26:WU01:FS01:0x17:ERROR:exception: Max Retries Reached
06:46:26:WU01:FS01:0x17:Saving result file logfile_01.txt
06:46:26:WU01:FS01:0x17:Saving result file log.txt
06:46:26:WU01:FS01:0x17:Folding@home Core Shutdown: BAD_WORK_UNIT
06:46:26:WARNING:WU01:FS01:FahCore returned: BAD_WORK_UNIT (114 = 0x72)
06:46:26:WU01:FS01:Sending unit results: id:01 state:SEND error:FAULTY project:9102 run:40 clone:15 gen:57 core:0x17 unit:0x000000400a3b1e81537c1aa885bc8ba0

 

[Unolocogringo]

Could be. I know for a while Certain work units would fail to ramp up clock speeds on one of my 460s.
Upped the voltage to 1.075 from 1.062 and seems to have fixed it.
Have not had a 9102 work unit yet.

 

[EXT64]

It is possible (caused by the OC). However it could just have been a random couple of bad WUs as well. I have run 9102s (so they aren't all bad 😉 ). If you seem to be getting bad WU somewhat consistently you could try backing off the OC a little to see if it helps.