P10200

[Pette Broad]

Not sure what the problem is but I got this from 171.67.108.39. using -advmethods flag. I can't find it on the summary page and it's showing up as a p11289 worth 69 points using FahSpy Server 171.67.108.39 looks to be running O.K.

Code: Select all

[18:17:12] + Attempting to send results [February 12 18:17:12 UTC]
[18:20:06] + Results successfully sent
[18:20:06] Thank you for your contribution to Folding@Home.
[18:20:06] + Number of Units Completed: 1912

[18:20:07] - Preparing to get new work unit...
[18:20:07] + Attempting to get work packet
[18:20:07] - Connecting to assignment server
[18:20:08] - Successful: assigned to (171.67.108.39).
[18:20:08] + News From Folding@Home: Welcome to Folding@Home
[18:20:08] Loaded queue successfully.
[18:20:26] + Closed connections
[18:20:26] 
[18:20:26] + Processing work unit
[18:20:26] Core required: FahCore_a4.exe
[18:20:26] Core found.
[18:20:27] Working on queue slot 07 [February 12 18:20:27 UTC]
[18:20:27] + Working ...
[18:20:27] 
[18:20:27] *------------------------------*
[18:20:27] Folding@Home Gromacs GB Core
[18:20:27] Version 2.25 (Sept. 16, 2010)
[18:20:27] 
[18:20:27] Preparing to commence simulation
[18:20:27] - Assembly optimizations manually forced on.
[18:20:27] - Not checking prior termination.
[18:20:27] - Expanded 1515984 -> 2956216 (decompressed 195.0 percent)
[18:20:27] Called DecompressByteArray: compressed_data_size=1515984 data_size=2956216, decompressed_data_size=2956216 diff=0
[18:20:27] - Digital signature verified
[18:20:27] 
[18:20:27] Project: 10200 (Run 0, Clone 9595, Gen 2)
[18:20:27] 
[18:20:27] Assembly optimizations on if available.
[18:20:27] Entering M.D.
[18:20:35] Completed 0 out of 250000 steps  (0%)
[18:44:45] Completed 2500 out of 250000 steps  (1%)
[19:08:53] Completed 5000 out of 250000 steps  (2%)
[19:33:01] Completed 7500 out of 250000 steps  (3%)
[19:57:09] Completed 10000 out of 250000 steps  (4%)
[20:21:15] Completed 12500 out of 250000 steps  (5%)
[20:45:23] Completed 15000 out of 250000 steps  (6%)
[21:09:50] Completed 17500 out of 250000 steps  (7%)
[21:34:07] Completed 20000 out of 250000 steps  (8%)
[21:58:20] Completed 22500 out of 250000 steps  (9%)
[22:22:33] Completed 25000 out of 250000 steps  (10%)
[22:46:49] Completed 27500 out of 250000 steps  (11%)
[23:11:30] Completed 30000 out of 250000 steps  (12%)
[23:35:47] Completed 32500 out of 250000 steps  (13%)

Thanks,

Pete

[John_Weatherman]

It's on the Summary page now see http://fah-web.stanford.edu/cgi-bin/fah ... ed?p=10200

[GreyWhiskers]

I just picked up one of the 10200 WUs last night on my old HP P4 HT 3.2 GHz system. Just looked at Windows XP TaskMan, and see it is using 109 MBytes memory - more than I usually see for the uniprocessor WUs. This seems no problem since I have 2 GBytes on the system, and TASKMAN is showing only a 1381 MByte commit charge for everything running.

I expect about 50-60 ppd on this slot, secondary to an ATI GPU on the system - and that's about what I'm seeing. Big surprise is the 2 hours 25 minutes TPF and 9.87 days ETA. I looked at the project page and it has a 136 day deadline on the single core A4 client for 493 points.

PPD at the moment is a little low, but the system just finished a 14 hour Norton full scan, and my system is one of the "7%ers" using FAH v7 client. I'll be interested in seeing what the next few frames look like.

Good news for me is that this will just run in the background. I don't seem to have any particular problem web surfing, working Quicken, or other things I will do with the system.

BTW, I don't have this v7 slot set up for "client-type advanced", so this must be a regular WU now.

No problem - just a curiosity.

[bruce]

The TPF and the ETA are often take considerable time to stabilize. Find a few frame messages for that WU and calculate the time it takes to progress from (X-1)% to (X)% and from (X-2)% to (X-1)%. Are the numbers equal? If so, is that the reported TPF? Multiply by 100 to get the total time for the WU or multiply by the remaining number of frames to get the ETA.

BTW, I don't have this v7 slot set up for "client-type advanced", so this must be a regular WU now.

See viewtopic.php?f=24&t=14714&p=164344#p164344

[GreyWhiskers]

I just got another of these on my unicore Pent 4/HT 3.2 GHz, and have a couple of observations.

a. Projects 10200 - 10205 are assigned to the general population - and there is a pretty good project description page, Project 10200, but there still isn't any entry in either psummary page, after more than 5 months on the scene. http://fah-web.stanford.edu/psummaryC.html; http://fah-web.stanford.edu/psummary.html

b. I can reiterate that this is quite a resource hog - Windows TASKMAN showing over 110 MiBytes memory usage (no problem for me, but that is a LOT for a uniprocessor WU).

c. This is showing about 92 minutes TPF - projecting completion in about seven days for 493 points -- about 70 ppd if something doesn't go haywire in the meantime!!. With all the discussion on quick returns for science, I still don't understand at all a 136 day deadline, with no timeout. This could be years before a meaningful number of serial runs are accomplished.

d. This just goes to show what a great variety of different projects, different scientific approaches there are in FAH!!. I had gotten spoiled by two P7021s in a row - the last (see below) completed in a little less than three days, getting 1022 points, monumental -- stupendous -- for this processor.

Code: Select all

Previously completed WU:
11:25:40:Sending unit results: id:02 state:SEND project:7021 run:3 clone:281 gen:2 core:0xa4 

snip

11:26:15:Server responded WORK_ACK (400)
11:26:15:Final credit estimate, 1022.00 points

Code: Select all

Projects for single-core (A4) clients.

Project	# Atoms	WU duration	Points	Deadline (days)
10200-10203, 10205	58794	0.5 ns	493	136
Projects for multi-core (A3) clients.

Project	# Atoms	WU duration	Points	Deadline (days)	Timeout (days)	k-factor
10204	58794	0.5 ns	343	3	2	3

[bruce]

Deadlines for A4 projects are currently under consideration. This is the first FahCore that has been able to run either as a SMP project or as a uniprocessor project. In the past, SMP projects could be given shorter timeouts than Uniprocessor projects for two very specific reasons:
1) Machines that are true uniprocessor hardware tend to be older hardware (and therefore slower).
2) SMP projects can easily be assigned based on whether you have a dual / quad / octo- / big (etc) hardware so deadlines can be more or less customized to your hardware with information that the Assignment Server can detect, but in all cases, they're assumed to be more powerful machines which can manage shorter deadlines.
3) Some people don't run FAH 24x7. No matter what the speed of their machine, they generally can only meet long deadlines if FAH only runs when they're checking their email (or whatever).

Some projects can tolerate longer deadlines; some cannot. In the past, those part-time people could run (multiple?) uniprocessor clients and still make useful contributions to FAH even though they can't meet the deadlines normally associated with SMP projects. So the question becomes (A) should FAH tell those people that it doesn't need their contributions or (B) should they be allowed to continue to contribute as long as they're actually doing scientifically useful work?

If the uniprocessor client has a choice between projects with longer deadlines (read: mostly FahCore_78) and projects with shorter deadlines (read: mostly FahCore_a4 projects) they'd be OK but their client doesn't have such a choice. If A4 projects can benefit from being assigned to either SMP or Uniprocessor clients, then some accommodation must be made for those part-time donors in group C, whether it's done by providing projects with long deadlines, by providing some Opt-in/Opt-out option in the client, or by telling them FAH doesn't want them to continue to contribute.