Folding Forum

Hello F@H users! I've been folding for about a year and a half now using my laptop ever since I saw a show on discovery channle that featured F@H. Till now I've only run the CPU client, I didn't even know that a GPU version existed. I'm not a expert on computers, dont get me wrong, I've built two of them from scratch so I'm definetaly not a novice, I just wouldn't label myself as a "expert". So I finally got the time to get a good desktop gaming rig operational. 2 grand and a few weeks later and I now have a computer with the following stats:

I7 Qudcore 2.8
16 Gigs DDR2 @ 1600
4 TB's of HDD
Bluray Player
2x EVGA 260 GTX's SLI'ed together
and a EVGA P55 Mainboard.

It's a decent rig, should hold me over for years. I plan on overclocking the Vid cards, RAM and Processor, cooling isn't an issue. My CPU fan is so big it could be used as a blunt weapon.

Getting to the point, I want to fold better. I read online that a GPU can fold much more efficently than a CPU. Now, please don't laugh at me here, a GPU is like a CPU, just on a video card right? How good would my SLI'ed 260 GTXs be at folding? And also, can I fold with my CPU and GPUs at the same time? My laptop is a 2 year old asus with a Nvidia 9650, and a quad core 2.something. Would it be worth it to change that one to GPU as well?

Also, I heard that running a GPU F@H client on a system that is SLI'ed doesn't work as well as with just one, is this true for my rig? And how stable is the software? I have an upcomming deployment to Afghanistan soon, and actually wanted to just leave my systems on at home running F@H till I'm back (shouldn't be more than 4 months or so). One last question, I don't know what group to join, are there any military minded ones out there?

Thanks guys!

Welcome to the F@H Forum Burnham113,

When considering overclocking, please remember that F@H is stressful so once you have overclocked your system, you may need to tweak it so that it can fold all WUs without any issues.

CPU efficiency VS GPU efficiency depends on factors like; are you folding -bigadv or not, etc. You can think of the CPU as a minivan that is kinda slow but can carry loads of data while the GPU is a dragster, very fast but can carry very little load. F@H doesn't use SLI since it is too slow. You can fold on your CPU and GPUs at the same time although, you have to make sure that the priority for the CPU is idle while the GPUs have low/slightly higher (that is configurable via the F@H Client). Laptops produce a lot of heat and IMO, if you are folding on the CPU, don't fold on the GPU since it would produce a lot of heat and wouldn't perform that great when compared to the desktop GPUs.

With the latest Nvidia Drivers correctly installed, you will have to install two separate instances of the GPU Client and you can leave SLI enabled (some prefer to disable it when folding). The GPU2 Client is finalized while the GPU3 Beta Client isn't. I have folded 2400+ WUs on the GPU3 Beta Client on GTX 260 Gigabyte SOC and haven't encountered any issues at all (I did encounter some issues when I was tweaking my overclock but nothing after that). If you plan to use the SMP2 Beta Client v6.30 on your system, you might have to occasionally monitor the system so maybe you can ask a friend, family member or you can use Ultra VNC, TightVNC to monitor it yourself if you can. Please note that this Forum is Team neutral so nobody can recruit you. The best advise is that you may want to check your favorite site or Google something of your liking and see if they have a F@H Team. If you can't find any Team of your choice, you can always create one.

Thanks for the quick reply Panther!

So, can I fold on each card's GPU at the same time? So I can actually have 3 running total? (1 CPU and 2 GPU) Or is it possible to fold with each core at the same time, so... Six instances running at once might be possible? Also, when using SMP2 Beta Client v6.30, what would I haveto monitor? Temperture, or is there a manual process to it?

Glad to hear the GTX260 works well with this, and thank you for taking the time to explain CPU and GPU concepts to me, and bearing with me as I learn!

Burnham113 wrote:So, can I fold on each card's GPU at the same time? So I can actually have 3 running total? (1 CPU and 2 GPU) Or is it possible to fold with each core at the same time, so... Six instances running at once might be possible? Also, when using SMP2 Beta Client v6.30, what would I haveto monitor? Temperture, or is there a manual process to it?

Yes, the ideal folding setup for that rig would have 3 clients running at once. The SMP client will use all of your CPU's cores to work on one Work Unit at a time, so you only need to run one copy of it (multiple SMP clients running is bad). You could run 4 copies of the Classic client, but that will earn far fewer points than 1xSMP. I'd recommend getting a passkey from http://fah-web.stanford.edu/cgi-bin/getpasskey.py and using it at least on the SMP client, as the passkey will let you earn Quick Return Bonuses on SMP WUs. You can use that same passkey on the GPU clients, though it won't get you any bonus points there; it'll just ensure that, in the unlikely event that someone tries to use your username for dubious things, everything you returned with a passkey can be verified as non-dubious.

I'm not sure what you're asking about with respect to monitoring. You should certainly monitor your temperatures to make sure they're not too high, but it sounds like you're asking about something else.

Core temp is a great app for monitoring the temperature of your CPU's cores. Otherwise Speedfan is another great option

The good thing with the i7's is they have a lot of circuitry dedicated to power and temperature management, so they are very good at slowing themselves if they ramp up to high temperatures unexpectedly, although running the CPU above 60-70 degs and core temps above 80 *may* reduce the life expectancy of the CPU, especially if you want to keep it for a lengthy period of time, although that said, they appear very resilient to heat.

Oh, it's just that panther said that "If you plan to use the SMP2 Beta Client v6.30 on your system, you might have to occasionally monitor the system so maybe you can ask a friend, family member or you can use Ultra VNC, TightVNC to monitor it yourself if you can."

So is there some manual component not present in classic mode? Or is it just error prone?

The Classic Client is considered a set-and-forget Client but the SMP2 Client isn't. However, the SMP2 Client has come a long way and with the latest version, I haven't had any issues at all with 440+ WUs folded. Also note that the Classic Client is a finalized Client while the SMP2 Client is still classified as Beta Client. Please note that regardless of what Clients you use, there will be a possibility that you will be assigned a Bad WU which may or may not cause your Client to get stuck. If that happens, you will have to manually "reset" the Client so that it can continue to fold (Details). I have encountered <20 bad WUs in over 4140 WUs that I recorded so it is very rare especially with default settings.

Everybody is giving good advice but nobody has asked a critical question. How many hours per week do you intend to run Folding? None if these "ideal" recommendations mean anything unless you meed the deadlines, so THEY ONLY APPLY if you run 24x7 and you didn't say that. If you fold a few hours a day, the best option is cpu clients because the deadlines are much longer than for the smp client.

PantherX, Zagen30, k1wi - you all know better than to overlook that "detail"

Burnham113 wrote:...wanted to just leave my systems on at home running F@H till I'm back (shouldn't be more than 4 months or so)...

I inferred that that meant 24/7, nonetheless, a clear answer by Burnham113 would be much better.

My apologies! I'm still new to all this

I intend on running it practically 24/7 on this desktop (except when gaming, when folding will be paused) and on 3-4 additional older desktops that I'm buying just for this purpose to run 24/7:
http://cgi.ebay.com/ws/eBayISAPI.dll?Vi ... :MEBIDX:IT

I'm also running it pretty much 24/7 on my laptop, may switch that one to SMP2 and GPU2, right now its just running classic.

Thank you all so much for your help!

Okay, some more questions, you guys got me really interested

So, I have a passkey now, but what can someone do with it thats bad? Tamper with files and get you banned or something?

Now, the GPU3 client, does that earn more points than the GPU2 client, does it fold faster? What are it's advantages? And the SMP2 Beta, what advantages does it have over the normal SMP client?

Thanks!

Let's say that somebody distributes illegal copies of the F@H Client and uses your username. If that user is found guilt of that, than PG Members will ban that username, team number and will nullify all points. Since you are using a passkey that no-one else knows so PG Members can differentiate between the violator's points and your points (Details).

The GPU3 Beta Client supports the Nvidia Fermi GPUs. The points will vary with the WUs assigned to you. Do note that GPU2 Client will be replaced by GPU3 Client but that is somewhere in the future. Also, FahCore_11 Projects (GPU2) will eventually be replaced by FahCore_15 Projects (GPU3) as it is the future of GPU folding. However, there isn't any ETA so it depends on your choice to install either Client (your GTX 260 can fold both FahCore_11 and FahCore_15 WUs).

The SMP2 Beta Client uses multicores to fold a single WU and has the Bonus Point System (Details). It is much easier to install and is very stable. The SMP1 Client is discontinued so your only choice is either SMP2 Beta Client v6.30 or the Classic Client v6.23 which doesn't have the Bonus Points system yet (It will be in the future but there isn't any ETA).

Thanks PantherX!

Happy folding everyone!

So, the SMP2 client works in command prompt, not a standalone thingamabob like the GPU? Do they plan to make a Nvidia specific one like the GPU3?

SMP is CLI (comman line interface) console only. No GUI.

They will not make a GUI for the v6 SMP client. v6 client development has stopped in favor of v7.

They do plan to make a v7 client. We don't know what that looks like yet.