Folding Forum

Removing one GPU will not help, as the other is already at x16 and they both performed the same on 9415 work units, so there is no pcie bottle neck.

2m PPD is only for 11432 project and you would get that probably on the other GPU too with a 11432.

But the Nvidia Tesla V100 does 1700k PPD also on 9415 work units on Linux.

So that is the question if the Titan V can do that on Linux too?

It is already known that p94xx doesn't perform as well on Windows on most high end cards as the do on Linux ...

And you didn't mention which version you use, but be sure to use the latest drivers ...

coffeechaps wrote:... does anyone know if this would help?

It might or it might not help. There's no simple answer.

Linux drivers are different than Windows drivers. NVidia drivers are different than AMD/ATI drivers. One FAH project may be different than another FAH project. The only safe answer is: It might help; it might not; it can't hurt.

There has been a few reports where somebody studied how much you can reduce the speed of the CPU running FAHCore_21 begins to degrade performance but even this test has not been run on multiple combinations of the speed of the PCIe bus for a specific GPU on a specific OS. There have also been a few reports studying reducing the speed of the PCIe bus to establish at what point a loss in performance begins to kick in (and like the CPU speed, that's going to depend on GPU & Project).

It's reasonable to say that a slot running at x8 is probably fast enough that you won't see any degradation for most projects on most GPUs, but as soon as someone makes that claim, somebody will find an exception. It's also reasonable to say that CPU speed doesn't matter ... but exceptions can be found if it is reduced enough.

so it sounds like i need to get this working on linux to proceed. I had no luck with the new ubuntu 16.04, but from what i have read herer and many other places, Mint should be the way to go. I will put mint on and see what I can do

in my testing PCIe x16 down to x8 only showed a 2-3% drop in performance

@Nathan_P: You mean on Windows only?

foldy wrote:@Nathan_P: You mean on Windows only?

No, this was all on linux

For installing the Nvidia driver on ubuntu, just do For FAH, download the client package and install it. Instructions here: http://folding.stanford.edu/support/faq ... -advanced/. It should start running automatically, I've had no issues with it on ubuntu 16.04. Didn't even need to reboot after installing the Nvidia driver.

Just get the client package and install it, I have not been able to get the fahcontrol one to work. You can either just watch the log file (in ~fahclient/log.txt) to see how much time it takes per frame for comparison, or configure the client for remote access if you have another windows machine you can use to monitor it.

For remote access, edit /etc/fahclient/config.xml and restart the FAHClient service.

You may need the "python-support" package too.

im out of town until this weekend, but i will try these changes when i return. first and for most i will get Mint working. thanks everyone

Hi there.
Haven't been around for many years - too many other things in life ....

I got a Titan V and a 1080 Ti. Put them in one of my older folding systems and just restarted the rig to see what happens. No further optimizations, just watching.

Config:
Win 10
FahClient 7.4.4.
Geforce Driver 388.71

Observation:
Both cards started with 9431 simultaneously.
1080 Ti : 970.000 ppd
Titan V: 940.000 ppd

1080 Ti (data from GPU-Z):
Clock: 1870 MHz
Mem: 1260 Mhz
GPU Temp: 81C
Fan 60%
GPU load 82%
Power consumption 85%

Titan V:
Clock: 1342 MHz
Mem Speed: 212 Mhz
GPU Temp: 57C
Fan 50%
GPU load 79%
Power consumption 38%

CPU i7-3770K:
CPU load 35%, no folding
Both cards run at PCI 3.0 x8


In a nutshell:
1) The Titan V delivers about the same performance as the 1080 TI with half the power consumption
2) something is not properly working with the Titan V, looks like it is just "idling" around. Need to look for further details and potential root causes.

rgds,
Andy

PS:
Started also FAHBench:

Implicit solvent / single precision:
1080 Ti: 385.81 ns/day
Titan V: 532.24 ns/day (+38%)

Implicit solvent / double precision:
1080 Ti: 23.99 ns / day
Titan V: 242.13 ns / day (~ 10x)

If It was me, I would uninstall and reinstall the folding software, it does configuration at install it does not do later.

Folks are selling these cards for $90: https://www.ebay.com/itm/Nvidia-Tesla-M ... 1438.l2649
I'm not familiar with them. Are they worth 90$ for folding???
I wonder if price gouging on graphics cards will end now that we're past xmas.

Anyone have some ballpark PPD numbers for Titan V, Titan X and Titan Xp running under Linux??? TIA
I found some here: https://docs.google.com/spreadsheets/d/ ... edit#gid=0
Nvidia Tesla V100 1700 10000 300 0,2 5,7 5,8 PPD on Linux. Windows guess 1200k PDD only
Nvidia Titan V 1500 3000 250 0,5 6,0 6,5 Currently only 1200k PPD cause of small work units
Titan Xp new 1200 1200 250 1,0 4,8 5,8 PPD on Linux. Windows only 900k PPD
Titan X Pascal 1100 1200 250 0,9 4,4 5,3 PPD on Linux. Windows only 900k PPD

The Tesla series (including the M2090) falls into FAH's oldest/lowest group of supported GPUs. It's very likely to work but it'll probably be pretty slow. You won't know unless you try it out.
It was introduced in 2011 and there have been a lot of GPU improvements since then.

Personally, I'd spend a little more and get a more recent-generation GPU.
Do you need a low-profile card? Try the GTX 1050 or 1050TI for slightly more $$ which use 1/3 the power and producing more.