Folding@Home

[Zargon]

1) I don't really know which numbers go to where, so I'll just write what I see when I see the words CPU on that "general" screen: Pentium(R) 4 CPU 2.20 GHz 2.19 GHz, 256 MB of RAM
Hope that helped.

2)Is using the internet very CPU intensive? My computer isn't usually on without someone doing something on it - mostly, using the internet. The longest time it's been on without us using it was about 6 hours, though even that much is VERY rare.

3)p3652_Seq22_Amber03_, GROMACS core
There are about 450 frames left (about 1/3 of the whole project) left, but apparently it'll be done in 2 days. Strange, since the other 2/3 took about half of December and 13 days of January.

4) In today's log file:
Assembly optimizations on if available.
[15:14:36] Entering M.D.
[15:15:02] (Starting from checkpoint)
[15:15:03] Protein: p3652_Seq22_Amber03_Extended
[15:15:03]
[15:15:03] Writing local files
[15:16:19] Completed 198513 out of 1500000 steps (13)
[15:16:21] Extra SSE boost OK.

However, in the log files of earlier ones (such as December 29, as a random example):
Assembly optimizations on if available.
[22:48:39] Entering M.D.
[22:48:45] Protein: p3699_Seq16_Amber03_Native
[22:48:45] Writing local files
[22:49:34] Gromacs error.
Did that have anything to do with it?

First, I apologize in advance that I have answered more extensively than needed as I'm aiming it for all readers (and hopeful inclusion of the FAQ that I've been unable to complete so far).

1) although not the fastest, your P4 2.2GHz system should be able to handle projects fairly quickly (I'd say on average about 100 hours). If the computer is only on about 30 hours a week, that means it could take several weeks.
BTW PIII/Athlon/Duron 600MHz are about as slow as I would suggest to run this project if those systems are usually on (e.g. file server) and those could take up to 4 times longer than P4. PII and K6-III+ or slower should not bother running this project as they will often not be able to make the preferred deadline.

There are many different areas covered by folding @ home. Although the end goal is disease research, part of the work is to do basic research so to be able to attack certain aspects of some diseases. As there are many different areas being researched at the same time, there are several different tools available to do that. From our point of view, we can see different tools (namely cores) being used.
Here is a list of all the current projects running, including what core they use (listed in the code column). http://fah-web.stanford.edu/psummary.html
One thing to note is that the various projects have huge differences in points and in deadlines (in general that isn't correlated). By the way, preferred deadline is the important deadline.
The various cores (code) use different parts of the CPU. For example:
The Gromac core uses SSE (or 3DNow+) extensively. This runs very quickly on a C2D, quite quickly on a P4 and fairly quickly on an AMD CPU (K10s should be better).
Amber use double precision numbers. This runs very quickly on an AMD CPU, quickly on a C2D and slowly on a P4.
So different projects work better on different types of computers.
Note the current general client doesn't recognize CPUs. The new beta client however does try to better match project to CPU type.
For details on the various cores http://fahwiki.net/index.php/Cores

256MB RAM is tight on Windows XP. So I'd suggest only accepting the smallest projects (up to 5MB). That should help keep the whole thing (code and project) as small as possible to minimize memory swapping.
To reduce memory even further there is also the console version (preferably with the service option).

2) Internet usage (bandwidth usage). Except when getting a new project (and possibly the related core with it) and sending back the results, there is no Internet usage at all.
By not accepting large projects, the bandwidth usage for receiving a project and sending back the results will be very small. (this is the suggested setting for those with a modem, slow Internet connection or limited bandwidth)
Using the Internet is normally not that CPU intensive, although if have quite a few tabs or windows open (especially with flash or other encoded media), watching videos, etc. it can use quite a bit of the CPU especially on an older CPU like the P4.

3) Different parts of the project can take different times. But the most likely explanation is that the method to calculate how long it will finish the project is not taking into account how busy the computer will be and how long the computer will be on.

4) That log look good - especially the SSE boost! Despite the project name including the word "Amber", this is Gromacs SREM (GROST) core based project so that does use the SSE.
This is a fairly large project so it most likely will take more than average amount of time complete.
However the log file with the error (for project 3699 that is also a GROST core/code based project) is worrisome. It is possible that your system switched to safe mode for a while (runs the project much slower) so that could have been why it was taking so long.
Now the reason why that happened is a concern. Any chance your system is overheating? (e.g. too many dust bunnies!?) Or that someone has overclocked the computer? One suggestion is to reduce the cpu usage to around 80% (about 3/4 the way to the right edge of the cpu usage selection bar if using the windows graphical client).
If run into errors again, let us know and if you want, we can walk you through some diagnostics.
By the way, what is the folding name you have used?