Just a moment if you please...

[Cerebral]

I am an idiot.

I, for whatever reason, cannot really figure out which client is for what or why they do what they do.

Knowing this I would hope someone can use an understandable means of communication to explain to me what the hell I am doing and how I can do it better.
All things said the more machines I can successfully get to fold the closer we come to a cure or more correct?
I would like to thank any and all in advance who reply.
I have 2 machines I am focusing on now and more later.
The first two are AMDs. 1)Athlon X2 5600+ (Windsor Cores) (2.8 Ghz), 4 GB PNY Optima DDR2 800, 1 PCIe 2.0 PNY Vertigo nVidia Geforce 9600 GT 1GB DDR3. 2)Phenom II X4 965 (Deneb cores) 4 Ghz, 8 GB Kingston HyperX Genesis DDR3 1600, 1 PCIe 2.0 Gigabyte HD 6870 OC (915 Mhz Core, 1050 Mhz Mem Stock) 1 GB DDR5.
I have others: amd, PS3 and 2 more intels to add but I want to get these running. Hopefully I won't need help once I understand which clients I need to keep up with.
Right now I am using the standard uni core client from the DL page and the GPU3 for the nVidia rig.
I figure an SMP client and gpu3 with a working control client but I am not sure. Any suggestions?
On the AMD rig I am running F@H Control Client and it does not do what I ask it to. It does not throttle the processor at all. It maxes all four cores out and gives prority to user threads. I believe it's Beta.
For reference the nVidia rig completes far more WUs than the AMD and the AMD is far more powerful so what am I doing wrong?
If what I suggested exists and works in that it would surely work on the AMD rig so I guess I either have an answer already or more suggestions?
Any help please?

[Jesse_V]

First off, I doubt your an idiot. After all, you joined F@h which IMO gives you one more edge over everyone else. Welcome to the forum Cerebral!

Yes, more machines will defintely help with improving F@h's speed in its simulations, along the way to cures for diseases. F@h uses techniques that scale very well to additional processors, thus achieving a near-linear speedup for every additional computer that joins.

The Pande Lab is currently working on the next generation of software, called the v7 client. It is designed to not only unify all previous v6 clients into one (thereby making the client choices much less complicated) but also to give those clients an intuitive GUI. It's under heavy development, and has a few bugs (such as PPD/ETA not working right) but I expect those to resolved fairly soon. We'll see. Is this the "F@h Control Client" you are talking about? The throttling may not work perfectly because as far as I know some of the cores do not support that feature. It should make everything much easier for you on your PCs. The PS3 has its own client.

F@h WUs are not equal. Some take longer than others, some are designed for specific hardware, and there's a variety of proteins being tackled that have varying number of atoms and whatnot. Us donors typically use points/day as a measure of our scientific productivity. I've heard of a lot of AMD GPUs using one CPU core as well for their calculations. The CPU folding may influence those AMD GPUs then. Nvidia GPUs are less prone to this problem, by way of their drivers and whatnot. F@h is specifically designed to back down for user applications, so that's normal behavior.

[Jesse_V]

Also, to break down the clients for you: GPU3 is the third generation of the GPU software. F@h was the first to offer GPUs for distributed computing. GPU3 has several advantages over GPU2. It supports more scientific capabilities and is more stable. I don't know precisely why, but GPU3 was not ready for AMD GPUs, so that's why those are left with the second generation, GPU2. V7 effectively solves this problem by offering even more scientific capabilities to both AMD/ATI and Nvidia GPUs. Since GPUs use specialized hardware to finish specific types of calculations extremely quickly, and they are capable of running molecular simulations, F@h achieves a significant speedup from those as well.

SMP is for folding on multi-core processors. F@h was one of the first to offer this client. It allows the simulations to achieve a 4x or more speedup, allowing a calculation that would take years into a matter of months or less.

The PS3 essentially takes the middle ground between flexibility in the types of calculations it can do and the speed at which they can do them. Its more flexible than GPUs, but less so than the more general CPUs. F@h was also the first to perform distributed computing on the PS3. All of these clients are designed to take advantage of specific types of hardware, and they supplement the standard uniprocessor clients that a lot of other projects only have. This setup gives F@h some pretty impressive performance. This page http://en.wikipedia.org/wiki/F@H#Software might be of some help.

[bruce]

One additional fact.

It has nothing to do with being an idiot. The information you've been looking at is really confusing. You can see a variety of clients ... all are Version 6 clients ... and deciding which one(s) to use is WAY too complicated. The V7 development that Jesse is talking about does not appear on the official download page yet. I expect that the multitude of choices that you see (all being V6) will be replaced with a single choice (being V7) and nobody else will be faced with the choices you're looking at.

Information about the beta V7 client can be found here. Click on the "Download Folding@home" link and you'll only have to pick your Operating System.

Whether you choose to participate in the beta testing of V7 or to choose one or more of the V6 clients, we're here to help.

[Cerebral]

TY and glad to be of service.

Now what you mentioned is right, it has PPD and ETA issues, they keep screwing up. I thought the GPU was tearing through material. Something came in with an expiration day in march and finished in 3 days. Didn't realize I was betaing the new software. Wonder how I got that? Ok well, I do and it reports EVERYTHING wrong. Is always saying in active when actually folding and always showing a processor colred as stopped or off (RED) when it is actually folding. It has broken several features in itself and changed so I suspect it has been updating itself. It also has no way to figure out the version number. The about is a web address.
If that's the case I guess i'll leave it alone. If I am wrong then what do I do?

Nevermind, I see I would choose the SMP and GPU3 DLs and run them on this like the nVidia rig.

How do I force the GPU to fold?

[Jesse_V]

FYI, the bug tracker is here: https://fah-web.stanford.edu/projects/F ... t/report/3 so you can see all the reported bugs in the software. This includes it reporting "active" incorrectly when you hover your mouse over the icon down by the clock. I'm not sure about the other things you are describing. Can you be more specific? Just in case you are reporting a new glitch that is not yet known, we'd like to know the specifics so the developers can track things down and address them. Thanks.

[Cerebral]

@Jessev
WOW, I have SO MANY of those bugs and a few I didn't see listed...

Yeah ok, before my day is out i'll enter them. It didn't look difficult.

Thank you BTW. I was actually somewhat lost until now.

And you like Portal.

@Bruce

TY For jumping on that. I appreciate the help.

[bruce]

If we're talking about V7, you can't force a GPU to fold.

Please post the first couple pages of the log. That can be done from the "log" tab in the middle of FAHControl.

Personally, I prefer copying the data from the actual file, itself ... your choice. When you installed FAH, in the "all programs" list, you'll find FAHClient and inside of it you'll find a link to the data directory which contains log.txt.

If we're talking about V6, forget all of that and I'll start over.

[Jesse_V]

By the way, there's no way for either of us to directly open a new ticket to report a bug. Post the specifics of you issue here, and someone else will be able to open a ticket for you if it is indeed a new problem. And I do believe you are talking about v7. But we're here to answer any questions you may have regarding anything about F@h.

The cake is a lie.

[bruce]

@Jessev
WOW, I have SO MANY of those bugs and a few I didn't see listed...
@Bruce

TY For jumping on that. I appreciate the help.

NP. You're in good hands with Jesse, but anybody with additional information can jump in at any time. Sometimes when there's a difficult problem being discussed, there may be suggestions from a dozen different people. There are a lot of knowledgeable people who help out -- and we're all volunteers, except for an occasional comment from somebody from the Pande Group.

As a way to interpret what bugs are being fixed (and what will probably happen next) look at the roadmap.

The public beta is currently V7.1.43. There are currently three milestones listed:
Public Release (34 tickets)
Open Beta Phase 2 (4 tickets)
V7.1.44 (1 ticket)

Public Release is somewhere out there in the future, but before it can happen we need to get to Open Beta Phase 2 which currently has four open tickets. In The short term, Development will probably be working mostly on those four items plus the one ticket that's already listed under Milestone: V7.1.44. The fact that ticket #794 is already listed under V7.1.44 means that there is a very high probability that it will be fixed in that version, even though V7.1.44 doesn't exist yet. Also, any of the four other tickets MIGHT be included when that version is frozen. If, when working on Tickets #395, #408, #563, or #526 there is some other ticket listed under Public Release that happens to be easy to fix at the same time, something from Public Release might also be included, but that's not too likely.

If development decides to freeze V7.1.44 (whether it closes 1 ticket or 5 or even 6 or 7), it will be tested by an internal PandeGroup test team. Depending on the results of that testing, a new milestone called V7.1.45 may be created or V7.1.44 may be released for Open Beta testing. As you can see, the current V7.1.43 (5th Open-Beta) was preceded by V7.1.38 (4th Open-Beta) so there were several intermediate versions that were frozen and sent for testing. The number of times this might happen cannot be predicted.

New tickets can also be opened if new problems are encountered, but that is becoming less and less frequent.

[Napoleon]

Are these rigs running 24/7? What other uses do they have besides folding? Which OS are they running? IMHO it would be a good approach to focus on one particular rig at a time. I'd propose to go through the motions of configuring v7 beta client optimally both for FAH and the OP's daily use cases, starting with rig 1 (Athlon X2 CPU + NVidia 9600GT GPU), for example. Once it's done, I'd expect it to get progressively easier for the OP to make educated choices with the rest of his rigs.

[Cerebral]

Ok I realized there was no way for me myself to enter tickets so I have no choice but to report bugs here.
But before that,
@Napoleon
Yes, it would make it easier for me to figure out clients once that works. Rig 1 and 2 (And the last AMD rig I just put up) all run 24/7. They all fold 24/7 within reason (System updates, power outages as i'm in rural WA and of course th occassional pause to get some performance back from the v7 Beta as it SUCKS THE LIFE out of my cpu and gpu).

Currently rig 1 is running v6.23 of the Uni-Core F@H and the GPU3 version. I know there is a better way. They both have icons in the task tray and use GUIs. I use that rig as a media server on a VERY underutilized network (No one cares much for the media, just the internet, so the machine lays idle most of the time). It has no other business.

Rig 2 is my game rig. I am 2 Double precision cards and 8 GB away from an actual useful gaming rig, but I am no longer the gamer I was in my youth. I play Skyrim and BF3 maybe 2-3 times a week for maybe an hour or 2 max. I watch movies at night and i surf the web on it as well. It has an IGP 4250 in it that I leave off and I was wondering if it could be turned on and used to fold? This rig is running the F@H v7 Beta Control client and a GPU version that supports my card (It comes up as a Barts XT so it must be working). Although it plows through WUs I never see the numbers for it so I wonder what the hell it actually is doing. It does more work than the nVidia rig and gets FAR less WUs. I know this is not a contest, and if the Beta is more important to the project right now than folding itself, which might be the case as confusing as all these non auto sensing clients are concerned, I will continue to run it and report bugs as permissable.

I went ahead and added the last AMD machine to the mix last night out of boredom. Interesting results.

AMD Athlon II X2 (2.8 Ghz), 4 GB DDR3 1333, IGP Radeon HD 4250 w/1GB from system mem as default. It is running GPU2 which has a systray icon and GUI and I seemed to DL a wierd version of the SMP client. SMP2. It is run from the command line. Good thing I had a Vic-20 with a 110 Baud Modem and acoustic coupler or I might never get what a batch file is or why it's like unix or basic commands and how they work. Fortunatley I find CMD far less work as I never move my arm as much typing as with a mouse anyway. It seems to be running fine for now. Suggestions there would be greatly appreciated. That machine is INUNDATED as it is the family machine and is use for everything from 2 adults and 2 children.

The PS3 and the 2 32 Bit intels can wait as they will prove a most daunting task and I know that already.
I am curious if the XBOX 360 will ever get F@H. I have one of those too and it is never used. To wit it is a fat model that is still in perfect operating shape, if that tells anyone who knows about xbox hardware how little I use it.
Almost forgot, I have an Atom netbook that I can set to fold at night using the unicore. Is a 1.66 Ghz single core atom enough to fold?
And lastly all units (Cept the consoles and netbook - NB runs Win 7 Basic x86) run Windows 7 ultimate x64 (The intels run Win 7 Ultimate x86).

Suggestions?

I will make a list and upload the info from the control panel on my next run through here.
I'll try and get in at least once a day until I get this all under control if you all don't mind helping.
Thank you so much again.

~Cerebral

[bruce]

Ok I realized there was no way for me myself to enter tickets so I have no choice but to report bugs here.
But before that,
. . .
rig 1 is running v6.23 ...

Let me stop you right there. Tickets are being opened for the the V7 beta. No development work is being done on V6.

[Cerebral]

@ Bruce

I caught that already, but Rig 2 is running v7 so I need help there as well as advice on the rest.

I have had no power for 3 days and just got it back. My apologies for not being here.

Please help me.

[PantherX]

... of course th occassional pause to get some performance back from the v7 Beta as it SUCKS THE LIFE out of my cpu and gpu...

I am aware that V7 can make more efficient use of your system. However, if the CPU is at 100%, you will not notice any issues with your daily applications since F@H runs at idle settings. If you are talking about the GPU, then there are reports that with Nvidia GPUs, the desktop lags and this is because there isn't any priority setting for GPUs so far thus we advice you to pause/exit the GPU Client.

...rig 1 is running v6.23 of the Uni-Core F@H and the GPU3 version. I know there is a better way. They both have icons in the task tray and use GUIs. I use that rig as a media server on a VERY underutilized network (No one cares much for the media, just the internet, so the machine lays idle most of the time)...

Since the machine is free for most of the time, this sounds like a plausible candidate for you to install the V7 (UNI + GPU) and test its features out.

...It has an IGP 4250 in it that I leave off and I was wondering if it could be turned on and used to fold? This rig is running the F@H v7 Beta Control client and a GPU version that supports my card (It comes up as a Barts XT so it must be working). Although it plows through WUs I never see the numbers for it so I wonder what the hell it actually is doing. It does more work than the nVidia rig and gets FAR less WUs...

AFAIK, the IGP is too weak to fold but if you can experiment with it, please do post your findings here. If you are unsure about the V7 performance, then please post the entire log here so that users with similar setups can give you feedback about your performance.

...It is running GPU2 which has a systray icon and GUI and I seemed to DL a wierd version of the SMP client. SMP2. It is run from the command line. Good thing I had a Vic-20 with a 110 Baud Modem and acoustic coupler or I might never get what a batch file is or why it's like unix or basic commands and how they work. Fortunatley I find CMD far less work as I never move my arm as much typing as with a mouse anyway. It seems to be running fine for now. Suggestions there would be greatly appreciated. That machine is INUNDATED as it is the family machine and is use for everything from 2 adults and 2 children...

Not sure what suggestions you want if everything is working fine. If the system has multiple accounts, then you can set up the SMP as a Service so it runs in the background regardless of who is running the system as long as it is powered up. The GPU2 can't be setup as a service if you are using Windows Vista/7 but is possible with Windows XP.

...I am curious if the XBOX 360 will ever get F@H. I have one of those too and it is never used. To wit it is a fat model that is still in perfect operating shape, if that tells anyone who knows about xbox hardware how little I use it...

There won't be any support for XBox (viewtopic.php?f=3&t=8342).

...I have an Atom netbook that I can set to fold at night using the unicore. Is a 1.66 Ghz single core atom enough to fold?...

My guess is that if it runs 24/7, then you can setup a UNI Slot on it. If you are unsure, you can experiment with it and report your findings here.