Which client should I download?

[kaywalker]

I read the forums a bit but it's a bit confusing to me. I have a few questions.
Which client should I run? GPU? CPU?
Can I run both?

I'm on windows XP. Cpu is a pentium 4 2.8 Ghz. Graphic card is ati radeon 9800 pro 128 MB. 1 gig of ram.
I know my pc is not very good, but I'm willing to do my part until I can afford a better pc.

Thanks in advance.

[Jeannie]

You can download either the console or graphics client - the GPU client won't work on an ATI 9800, and you have a single CPU, so the SMP console won't work for you.

For Windows, you can download the console or the graphics clients in the older 'version 5', or the newer Beta 'version 6'.

The difference between the console and the graphics clients the way they look - with the graphics client, you'll always see FAH running in your system toolbar, and if you click on it, you'll see a graphic of a protein folding, plus some other information about the progress of the workunit you are processing (percent complete and estimated date/time of completion). With the console client, you won't get that 'picture', and you'll have to look in the log file that FAH creates to see your progress, or else use a 3rd party tool like FAH-MON to see it (you'll find info about 3rd party software tools here in this forum, also).

With the console client, you can choose to have it run either as a 'service', where it will start automatically whenever you boot your computer, or as a regular program that you will need to start when you log in. You'll have a choice for which option when you configure the client.

As to the difference between the older clients and the 'version 6 Beta' clients, I'd recommend the version 6 - even though it's still in Beta testing, it's pretty stable.

I started off a little more than a year ago by downloading and running a graphics client because it was so easy to install. I downloaded it and started it up right away, and later I figured out I needed to 'configure' it by right-clicking it in my system tray, so I could add my username and team number. It was running immediately, but would have given the points accumulated to 'anonymous' as the user name.

With the console client, as I said, you need to configure it, but it will start up the configuration dialog the first time you run it, where you can enter a username and team name and also answer "Y"es or "N"o to the question that asks if you want to run it as a service. You can later go in and change your configuration by running the console client with the -configonly flag.

Happy folding, and feel free to ask additional questions!

[kaywalker]

thanks for the response.

I had another question. The cpu and the gpu programs, are they essentially folding the same stuff with different hardware? or they are completely different folding material?

[Jeannie]

The console client and the graphics client both use just your CPU.

The GPU client runs using the processing power on certain ATI video cards, and I assume it also uses some of the power of your CPU.

Certain 'workunits' can be run by the GPU client, and others can be run by the console or graphics clients. I don't believe there is any overlap between those groups of workunits.

Similarly, the 'SMP' client must have a multi-core CPU, and only certain workunits will run with it. Again, there's no overlap with the other groups of workunits. In fact, I have two computers where I run the SMP client and two where I run the windows console client; once in a while, when SMP finishes one work unit, it will have to wait until someone at Stanford releases new work units meant for the SMP client. I think once I had to 'wait' more than 1 day until I could get a new workunit for an SMP client.

[kaywalker]

I see. By the way as I mentioned my cpu is a 2.8 ghz single core. Is it even worth contributing? it takes 2-3 days sometimes more to finish a work unit.

[gwildperson]

I see. By the way as I mentioned my cpu is a 2.8 ghz single core. Is it even worth contributing? it takes 2-3 days sometimes more to finish a work unit.

Yes, a 2.8 GHz machine is still worth running. The benchmark machine for the uniprocessor client is a 2.8 GHz P4 and it'll get 110 Points per day, no matter how many days it takes to finish a WU.

[jrweiss]

As well as being more than adequate to run F@H, you may be able to run large WUs and get a few bonus points. If you have 1 GB RAM or more you can try enabling large WUs in the configuration. If they use too much RAM for your purposes, simply turn off the switch again.

Some WUs can take a week or more. That's normal for the large ones, and is reflected in their deadlines.

[kaywalker]

As well as being more than adequate to run F@H, you may be able to run large WUs and get a few bonus points. If you have 1 GB RAM or more you can try enabling large WUs in the configuration. If they use too much RAM for your purposes, simply turn off the switch again.

Some WUs can take a week or more. That's normal for the large ones, and is reflected in their deadlines.

where can you see their deadlines?
Also I do have 1 gig of ram. But I don't see an option in configuration that let's me pick larger WUs. Could you please tell me where that is?

[Ren02]

2 days per WU is still decent. Actually, if you look at the client stats then the Windows clients get an average of 173 TFLOPS / 181955 clients = 0.95 GFLOPS per client. Now a 2.8 GHz P4 should be capable of 1.2 GFLOPS performance, so your machine is above average.

The deadlines for each poject can be seen here or if you prefer longer way, then go to http://folding.stanford.edu/ pick Stats, then: Project summary (directly from the servers, updated hourly).
There are 2 deadlines, preferred and final. If you don't make the preferred deadline, then the WU gets reassigned to someone else but the uniprocessor client deadlines are usually 20+ days so there shouldn't be any problem.
If you miss the preferred deadline but make the final deadline you'll still get points, but since the WU is reassigned you're not exactly helping the project..

[bruce]

Configuration settings are slghtly different depending on whether you'r running the V5 console client or the V6 console client. This is what you would see with V6. The xxxxx's are the value you prevously set (or the original default if you have not set anything yet.

User name [xxxxxxx]?
Team Number [xxx]?
Passkey [xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx]?
Launch automatically at machine startup, installing this as a service (yes/no) [no]?
Ask before fetching/sending work (no/yes) [no]?
Use proxy (yes/no) [no]?
Acceptable size of work assignment and work result packets (bigger units
may have large memory demands) -- 'small' is <5MB, 'normal' is <10MB, and
'big' is >10MB (small/normal/big) [normal]?
Change advanced options (yes/no) [no]?

User name [xxxxx]?
Team Number [xxxx]?
Passkey [xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx]?
Launch automatically at machine startup, installing this as a service (yes/no) [no]?
Ask before fetching/sending work (no/yes) [no]?
Use proxy (yes/no) [no]?
Acceptable size of work assignment and work result packets (bigger units
may have large memory demands) -- 'small' is <5MB, 'normal' is <10MB, and
'big' is >10MB (small/normal/big) [normal]?
Change advanced options (yes/no) [no]?

V5 has no passkey It has an option for UseIE Settings (which should be set to [no]) and the size of WUs has two choices, not three.

The deadlines are found on the project summary page which can be accessed from the stanford homepage or from a link at the top of this page.