New F@H Team Members Post Here First

[qubit]

Your latest recruit

I've finally done it people and joined the TPU Folding team. Username is 'Qubit'.

Buck, I finally found the team I used to be in and my work units while reading the install FAQ. I just guessed I'd used qubit in the past, so I stuck qubit in the username search, found a few hits and looked through them. Are my old work units going to transfer to TPU? I've just created a passkey, too.

I have to go out in a bit, so I've got it going on the bog standard Stanford client for now, which is only running off the CPU. Obviously, as I've got a high-ish performance PC with a GTX 580, I want to use that.

I've noticed that Stanford have a GPU client, or there's the third party one by jedi95 pointed to by Buck's FAQ post. As I'm not familiar with either of them, so is it a case of jedi95's GPU client being much more advanced than Stanford's GPU client?

 

[thebluebumblebee]

I've finally done it people and joined the TPU Folding team. Username is 'Qubit'.

Buck, I finally found the team I used to be in and my work units while reading the install FAQ. I just guessed I'd used qubit in the past, so I stuck qubit in the username search, found a few hits and looked through them. Are my old work units going to transfer to TPU? I've just created a passkey, too.

The WU's are always yours, but they stay with the team that you were Folding with at the time.

I have to go out in a bit, so I've got it going on the bog standard Stanford client for now, which is only running off the CPU. Obviously, as I've got a high-ish performance PC with a GTX 580, I want to use that.

"bog standard Stanford client"? You should be using the SMP client for your CPU.* More on that below.

I've noticed that Stanford have a GPU client, or there's the third party one by jedi95 pointed to by Buck's FAQ post. As I'm not familiar with either of them, so is it a case of jedi95's GPU client being much more advanced than Stanford's GPU client?

All Folding clients are Stanford's. There is a front end called FAH GPU Tracker V2, and it does make multi-client Folding easier. It will analyze your hardware and offer to install the appropriate client(s).

If you have any further questions, just ask.

 

[qubit]

The WU's are always yours, but they stay with the team that you were Folding with at the time.



"bog standard Stanford client"? You should be using the SMP client for your CPU.* More on that below.


All Folding clients are Stanford's. There is a front end called FAH GPU Tracker V2, and it does make multi-client Folding easier. It will analyze your hardware and offer to install the appropriate client(s).

If you have any further questions, just ask.

Another awesome, full reply. Thanks dude, I'll run that tracker V2 post-haste.

I've so far contributed one whole unit to the team.

 

[qubit]

Hmmm... it doesn't look like I'll be able to run F@H optimally.

I ran the tracker, set everything up and got just the GPU to do the work. However, there's the most annoying modulated high-pitched whistling sound coming from the speakers and also directly from the graphics card. It's clearly audible and not possible to ignore, so it's no go. While I didn't expect the audio breakthrough, I expected the graphics card to make some noise, from previous experience of driving cards at very high framerates. I guess I could throttle it back a bit, but it's still gonna make a noise, just less, so it's pointless.

On to the CPU: I can only run the single core client right now, because the cooling isn't up to it. My Zalman cooler is analog and has a resistor to run it at a constant slowish speed for quietness, which it does very well. This is fine for everyday use and games (just about) but the CPU was climbing the high 70's before I quit the client. In fact, I had to kill the client with the Task Manager too, because the tracker couldn't stop it. :shadedshu

So there you go. I'm gonna continue to run it, just not with the performance us enthusiasts would like. Any suggestions for improving this situation welcome.

 

[Nergal]

Hello Techpowerup!

just wanted to let you know I joined under the name "Netherlord"

got a new laptop, i5 2410 8GB ram; and my old rig, E8400 + HD6850(thats new), running

I`m also considering of building a cluster@home

found myself a nice master thesis and I`ll be looking into it this week

At my full disposal I could create a cluster with the following:

i5, 8GB (ddr3), 540GT
E8400, 4GB (ddr2), HD6850
P4@3GHZ, 4GB
P4 celeron@2.4GHZ, 1GB DDR
P3@1GHZ, 384MB SD
AMD Athlon@1.2GHZ or a +2000, 1GB DDR

worth the effort?

 

[thebluebumblebee]

Would you tell us what client(s) you're using? The HD6850 needs client V7 to get decent PPD.

 

[Nergal]

Installed the V7
what settings would I set best? (32bit XP...I know...)

 

[thebluebumblebee]

Installed the V7
what settings would I set best? (32bit XP...I know...)

To get the new fahcore_16 for your AMD card, please check this post http://www.techpowerup.com/forums/showpost.php?p=2248985&postcount=1508

 

[LightningJR]

Hi everyone! I'm now folding for TPU. I have two PCs running it. One is an Athlon II X2 240 @ 3.6Ghz with a 9500GT folding about 16hrs a day on both and the second is an Athlon II X3 @ X4 3.6Ghz with a HD4830 running somewhere between 7-10hrs a day. I have a question. With the second rig I run both the GPU and CPU and I get about 4,800PPD (1,800 on the GPU) but if I turn off GPU folding I get the exact same result.. The CPU keeps up and gives me the same 4,800PPU. I'm wondering if there's ANY reason to fold both if the PPD doesn't increase? I'm also going to test and see if I get the same result on the first rig. (I find it sad that the 9500GT gets 2,000PPD and the 4830 gets 1,800PPD :/ How is it that the ATI card with it's double performance gets less than the NVidia is beyond me.. )

 

[thebluebumblebee]

Welcome!
Due to the design of the ATI pre-5xxx series, there is really no reason to Fold on them when considering PPD/watt and, that the ATI GPU puts a load on the CPU and slows it down. The Nvidia cards are just better suited for Folding, although the AMD 6xxx series is starting to get competitive. No need to test the 9500GT.

 

[LightningJR]

No need to test the 9500GT.

Too late

But yeah, the CPU on the 9500GT rig may have went down by 100PPD.. That's about it.. So ~4,500PPD on one and ~4,800PPD on the other.. now I want to reach 10,000PPD... :/ And so it starts..

Another thing. I have a partial PC just crying to be folding. If anyone can point me to a cheap socket 754 PCIE mobo I would be grateful.
Thanks

JR

 

[thebluebumblebee]

I don't mean to be Mr. Po-Po, but if you're thinking about Folding on an Athlon 64 CPU, it's also not worth it.(personal experience) Want to add big points, (in comparison to what you have) find a GTX460. They're good for 10,000 PPD before OC.(based on my MSI GTX 460 which has a stock clock of 750MHz)
What Client(s) are you using?

 

[LightningJR]

Well I would use the other 9500GT that I have lying around to fold not the Athlon 64.. So it would be worth it if I could get a mobo for ~$20.

Well im using FAH GPU Tracker V2.

 

[thebluebumblebee]

As for the 754 MOBO, do you have a "free geek" around your area?

If you want to try to get more points from the 4830, you could try the instructions in step 5 of this thread and I added a clarification comment at post #24 in that thread.

 

[LightningJR]

Unfortunately in such a small town not really. Im one of the very few geeks here..

Thanks for the info. Will those variables affect other things as well as F@H?

 

[thebluebumblebee]

I believe that the EV's are only for the Folding client.

 

[dannylill1981]

hi guys gonna start folding tonight for tpu but using the system in my sig whats the best client to use to take full advantage of it.?

 

[thebluebumblebee]

Hi and welcome. I'd recommend just following this thread. Run GPU2 on your GPU and SMP on your CPU. I don't know how well F@H is going to run on that GPU since it only has 32 Shaders. The GPU OC priority for F@H is Shaders, Core and then memory. The SMP client is really for those who want to dedicate their PC to Folding. If you are not at that point, you can run up to 4 instances of the uniprocessor client which have a much longer deadline, but will yield much less points than the SMP client.

 

[MadDogTen]

Hello, I'v just starting trying this today, seems to being running well right now ~855 ns/day (Assuming this is normal for my build) using the GPU (Seems to be settling at ~72C/160F). System Specs (CPU is overclocked, and this shows how much, otherwise the one on this site is Correct): http://valid.canardpc.com/show_oc.php?id=2203406
50% Done Now with a 50,000,000 Work Load.

I would rather only use my GPU for this, as this is my Main Computer, and I still want to be able to do things at the same time (Which is seems to be working fine right now).

I will try to do at least 1 Work Load Per a Day.

 

[thebluebumblebee]

Hi, and welcome MadDogTen. What client are you using? My advice would be to only Fold on the GPU.

 

[MadDogTen]

Hi, and welcome MadDogTen. What client are you using? My advice would be to only Fold on the GPU.

I'm using Folding@Home Client Version 6.41r2 GPU3. I am planning on switching to the No Console Version after the current workload is completed (I'm at 93%).

and yes, Also the fast that I have a Sucky CPU.


Edit: Also, is it normal for the ns/day to slowly increase? It started at about 755 ns/day, then when I checked next to about 830 ns/dau, then 855 ns/day, and now 860 ns/day, and I'v been doing the same thing the whole time.

 

[thebluebumblebee]

Console or system tray version does not matter. Turning off the viewer will speed it up, but that doesn't matter if you are only doing 1 WU/day. The longer you work on a WU, the better the client is able to figure how fast you are doing it. I don't know the system tray client, (which is what I believe that you have) but I believe that you can set it to "last 3 frames" to judge the speed. I think it's set to "effective" which takes into account the upload/ download times.
What is the "no console client" that you're talking about? FAHtracker? Client V7?

 

[MadDogTen]

Console or system tray version does not matter. Turning off the viewer will speed it up, but that doesn't matter if you are only doing 1 WU/day. The longer you work on a WU, the better the client is able to figure how fast you are doing it. I don't know the system tray client, (which is what I believe that you have) but I believe that you can set it to "last 3 frames" to judge the speed. I think it's set to "effective" which takes into account the upload/ download times.
What is the "no console client" that you're talking about? FAHtracker? Client V7?

My Bad, I meant the "No-Nonsense Console" One, Second one on this list.

The 1 WU/day is just a Min. Amount I will do in a day. Any more then that just depends on what I'm doing.

 

[thebluebumblebee]

No reason to switch, unless you switch to Client V7 (the client of the future) which has some nice features like being able to pause a WU anytime you want. But, the sys tray client for GPU3 is really a nice easy client.

 

[MadDogTen]

No reason to switch, unless you switch to Client V7 (the client of the future) which has some nice features like being able to pause a WU anytime you want. But, the sys tray client for GPU3 is really a nice easy client.

Ok, Thanks! I will switch to V7, Thanks for the Link!