WUs are too big

Hello mates. I've been folding for quite a while now. The system I will be referring to is my Socket A system, which has been my main folding rig (and my main rig in general) up until I was given an Opteron system. I recently gave this Socket A system to my buddy, switching from a router which had ports forwarded and such, (comcast cable) to a network/router that did not. That said, I never had to, and never did, open any ports to make folding worth. Using the GUI client, on this new network, I got the Attempting to Get Work Pack constant loop. I uninstalled and downloaded the Console Client, as that is what I'm now using on my current Socket 939 system, and am familiar with it.

So I installed and set up the Folding Console client. I picked Normal out of small/normal/big work units. This is where the problem begins...before, using the graphic client, this system would only get 500-2500-5000 step work units. Now, I have a 500K step client. it did 1% in 6 hours (granted it may not have have 100% cpu to use during that time, but it may not even make the deadline). Should I be using small WUs? Or does the console get larger WUs in general? I would get 250 step WUs on this system and they'd be the one's that took the longest, as opposed to the higher (2500-5000) step WUs. Now it has a 500,000 step WU, much like my (much faster) dual core Opteron receives.

So my question is, what option should i set to get the WUs lower (to something that can be completed in the deadline) or any other suggestions that you guys have.

Single core WU's generally have 30 days or longer to complete. 6 hours x 100% = 600 / 24 = 25 days..... so yeah you should make it. Could you post up the WU project number and we can figure out if it is a heavy WU and what the deadlines are? Honestly I've never selected anything but BIG for the work unit size even when I was running F@H on my Athlon XP.

Now, I have a 500K step client. it did 1% in 6 hours (granted it may not have have 100% cpu to use during that time,

Click to expand...

If the system was under a heavy load while the client was running for that 1% then 6 hours may not be accurate, the following frames may run faster if the client is allowed to use more of the CPU.

Thank you for the reply! I understand the constraint that normal or big sized work units take up a large amount of time and processing capabilities. I just find the jump from 2500-5000 step to 500,000 step WUs to be large. I guarantee that this client can finish said large WUs in 30 days or less.

Kowalski3500 said:

I just find the jump from 2500-5000 step to 500,000 step WUs to be large.

Click to expand...

Stanford might just be getting spoiled by the GPU and Big Advanced clients and are pushing the older client. I'm sure there is a good percentage of systems running the single core client that are pretty fast but just are not running 24/7 so they can't make the SMP deadlines.

There is no relationship between the number of steps and the amount of time it takes to fold a WU. Almost any Any Socket A AMD Athlon should finish within the deadline of all the uniprocessor clients, unless it is run less than 8/5 or, in the case of an XP-M is running at reduced speed (as in cmos battery died and the chip reverted to minimum bus and multiplier speeds. What cpu and what project? My Athlon XP-Ms are running even the largest projects, like p1773 (2138 points) in about 1:30:00/frame and making 330 ppd. The deadline on these WUs is uncharacteristically short at 14 days preferred. We need to figure out why it's taking 6 hours/frame.

@Chas, just curious- what board are your XP-Ms running in? Perhaps a dually? Wire mod?

I just bought an XP-M for "old times sake" and of course it will be required to fold, at least part of the time.

Abit NF-7S. When new, all of them would do 2.4 GHz (I had 9 at one point). I think the motherboards are getting tired of of them and the two I have left running are down to 2.2 and 2.3 GHz.

I loved my 2 Abit NF-7 boards; sold 'em when I went Intel for folding. :-(

I just removed all the specs from my old comp from my signature, but heres the lowdown

XP-M Barton @2300 (200x11.5) @1.575v (very very stable)
in A78NX-E Deluxe

Folding stats:
Gromacs 3.3
FahCore_a0.exe
Project 1771 (Run 6, Clone 33, Gen 6)

Lookin at my buddys console client here
every 1.5-2 hours it checkpoints 1% of progress
330-503
503-1022 (he was most likely playing wow)
1022-1154
And similar times now. Since the beginning of this post he's got to 40% And I'm pretty sure he hasn't been folding 24/7

I just bought an A78NX-E for the XP-M, will have 1/2 a gig of RAM and a quadro fx 500/600 graphics card. It'll fold at least one WU as an 'initiation' to join the household menagerie.

For clarification, in this older system which I have been referring to, there is:

2GB of OCZ Gold or Platinum RAM= the 200 FSB stated previously.
The numbers (330-503, 503-1022, etc.) were military time of 1% worth of folding. So every increment = 1% of folding according to the Console.

That said, 40% since "the beginning of this post" actually means 40% since the beginning of this thread.

In addition, I've had to change the Clock Rate of this processor down to 2200 instead of 2300 (200X11 vs. 200X11.5) because of instability. There goes my previous statement of "Very Very Stable". (Temperature issues: 50C under load at my house and config vs. 56+ at my buddy's house and config)

That said, now, system #2 (the Athlon XP-M system) is folding @ 2200 (200x11)@1.575 and so far, should be stable (fingers crossed). And I will update with results/further thoughts.

Running into so many problems with that system when it comes to folding. Couldn't get a new WU nor transmit the finished one. He's running Qwest DSL (at freakin' 20 kilobytes per second download speed, surely not what he's paying for) but anyway I'm pretty sure it has to do with firewall junk and I'm not up to messing with it anymore at the moment.

So it's been retired as a folding machine :/