Page 2 of 11

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Sat May 03, 2008 8:00 pm
by apd183
uncle_fungus wrote:4519-4521 at the moment
One of my computers that I know does NOT have the -advmethods option got a 4520 today. It is the 5.03 Windows GUI client and is also not set for big WUs. Should I be getting these WUs on this computer?

[edit: fixed version number]

Re: New Bonus Amber Core WU's

Posted: Sun May 04, 2008 12:11 am
by bruce
GTron wrote:1) If using the Linux v6.02beta1 client, does this imply setting WU size to "small" or is defaulting to "normal" correct/preferred?
2) Or is the Linux v6.02beta1 client, without the -smp flag, not a "classic client" able to process these WUs?
From what I've heard, small and normal may both work but I'm not sure anybody really knows yet.

The LInux client without the -smp flag is, indeed, a uniprocessor client so it should also be able to get these WUs. Again, it's all new so I'm not sure anybody knows yet.

Once vvoelz increases the priority, there will be plenty of them, and I'm sure we'll all know almost immediately which settings matter and which do not.

Re: New Bonus Amber Core WU's

Posted: Sun May 04, 2008 12:21 am
by GTron
Tobit wrote:
GTron wrote:1) If using the Linux v6.02beta1 client, does this imply setting WU size to "small" or is defaulting to "normal" correct/preferred?
2) Or is the Linux v6.02beta1 client, without the -smp flag, not a "classic client" able to process these WUs?
Neither will work as these bonus units aren't able to run on the SMP client. These projects require the Amber core which isn't available on the SMP and other non-classic clients. See this thread for the official announcement.

Meanwhile, I woke up this morning to find two of my classic clients chewing on each their own 4519. :D
Actually, my Linux v6.02beta1 clients, without the -smp flag, are happily folding some 4519 and 4520s now too! :D
For WU size, I took the "normal" default.

Tobit must have been misled by the download page which currently describes this client as "SMP client console version". However, it is an INTEGRATED client, not just an SMP client, as the v6 Client FAQ clearly states -- "integration of SMP support into the standard client". I suppose the download page misleadingly says SMP since it is the only current Linux client to support SMP -- misleading because it is ALSO a standard/CPU/uniprocessor/classic (choose your favorite term) client when run without the -smp flag.

Fold On
Greg

Edit: Bruce, you beat me to it while I was typing! Thanks. (And no, I haven't tried "small" yet.)

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Sun May 04, 2008 1:38 am
by sneakers55
I've got a WU that is running on AMBER.

What I'd love to get is more for DOUBLE GROMACS B. My T7800 gets HOT but I'm getting about 780 PPD per core!

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Sun May 04, 2008 6:49 am
by theMASS
sneakers55 wrote:I've got a WU that is running on AMBER.

What I'd love to get is more for DOUBLE GROMACS B. My T7800 gets HOT but I'm getting about 780 PPD per core!
I think you'd be better off running SMP. My T9300 (2.5 GHz) does 880-925 PPD/Core. Vista 32-bit with VMware Server running the Linux client

4520 (Run 42, Clone 96, Gen 1)

Posted: Sun May 04, 2008 6:00 pm
by apd183
apd183 wrote:
uncle_fungus wrote:4519-4521 at the moment
One of my computers that I know does NOT have the -advmethods option got a 4520 today. It is the 5.03 Windows GUI client and is also not set for big WUs. Should I be getting these WUs on this computer?
Here's some more info.

First lines of log:

Code: Select all

# Windows Graphical Edition ###################################################
###############################################################################

                       Folding@Home Client Version 5.03

                          http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\Program Files\Folding@Home


[19:42:27] - Ask before connecting: No
[19:42:27] - User name: apd183 (Team 682)
[19:42:27] - User ID not found locally
[19:42:27] + Requesting User ID from server
[19:42:31] - Machine ID: 1
[19:42:31] 
[19:42:31] Work directory not found. Creating...
[19:42:31] Could not open work queue, generating new queue...
[19:42:34] Initialization complete
[19:42:34] + Benchmarking ...
WU 4520 (Run 42, Clone 96, Gen 1) received:

Code: Select all

[18:05:42] - Preparing to get new work unit...
[18:05:42] + Attempting to get work packet
[18:05:42] - Connecting to assignment server
[18:05:43] - Successful: assigned to (169.230.26.30).
[18:05:43] + News From Folding@Home: Welcome to Folding@Home
[18:05:43] Loaded queue successfully.
[18:05:45] 
[18:05:45] + Processing work unit
[18:05:45] Core required: FahCore_82.exe
[18:05:45] Core found.
[18:05:45] Working on Unit 05 [May 3 18:05:45]
[18:05:45] + Working ...
[18:05:45] 
[18:05:45] *------------------------------*
[18:05:45] Folding@Home PMD Core
[18:05:45] Version 1.03 (September 7, 2005)
[18:05:45] 
[18:05:45] Preparing to commence simulation
[18:05:45] - Looking at optimizations...
[18:05:45] - Created dyn
[18:05:45] - Files status OK
[18:05:45] - Expanded 19825 -> 126414 (decompressed 637.6 percent)
[18:05:45] 
[18:05:45] Project: 4520 (Run 42, Clone 96, Gen 1)
[18:05:45] 
[18:05:45] Assembly optimizations on if available.
[18:05:45] Entering M.D.
[18:05:52] Protein: p4520_2CVR-16-10ns_minout
[18:05:52]
[18:05:52] Completed 0 out of 5000000 steps  (0)
[18:20:46] Writing checkpoint files
[18:28:29] Writing local files

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Sun May 04, 2008 6:50 pm
by Lgringo
i recently got an AMBER along with family GROMACS, etc. but on first glance it would appear that the bonus is in the negative direction ... :-|
FAHspy snapshot at:
http://navcompars.net/gringo/x8spy.gif

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Sun May 04, 2008 11:56 pm
by alancabler
Howdy Lgringo,
There's a good discussion about p45xx points in this thread.
Points tend to average correctly after several completed WUs.

By the way, I haven't seen anything slowing your points down, lately. :D
You'll soon be the #1 donor on Team 7.
Good job!

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Mon May 05, 2008 5:44 am
by vvoelz
Heads up for three new 'bonused' AMBER Core WUs coming onling in the next day or so:
p4522, p4523, and p4524. See http://fah-web.stanford.edu/cgi-bin/fahproject?p=4522 for point details.

Vince
Pande Lab

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Mon May 05, 2008 3:03 pm
by Tobit
Excellent Vince, thanks for the info!

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Mon May 05, 2008 5:02 pm
by Tobit
26 hours to complete project 4520 (Run 32, Clone 2, Gen 5). 00:16:32 per frame

Linux 5.04 client, 996MB RAM available, 3.2GHz Celeron-D with 256K L2 - benchmarking reports 1162
Arguments: -advmethods -verbosity 9

$10 says the lack of cache is killing the performance. :biggrin:

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Mon May 05, 2008 7:40 pm
by Jeff_Grant
I see some Intels on fahinfo.org for 4521 and 4519. I have one 4521 on an older athlonxp 1500 that I threw up there as well. As everyone gets these units, can you add to the database?

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Mon May 05, 2008 9:00 pm
by GTron
I'm seeing quite a bit of variability in the expected completion times on the 4520s (started at the same time) as shown below. The spread is about 9:18 hours:min between #2 and #3. Probably safe to say "Your Mileage WILL Vary"! :ewink:
These are all running on the same dedicated Q6600@3GHz, Ubuntu 8.04, Linux v6.02beta1 client.

Greg

Code: Select all

Mon May  5 14:32:35 MDT 2008

1)
 Index 3: folding now 51.3 X min speed; 9.67% complete
  server: 169.230.26.30:8080; project: 4520, "p4520_2CVR-16-10ns_minout"
  Folding: run 22, clone 66, generation 2; benchmark 0; misc: 500, 300
  issue: Mon May  5 12:09:59 2008; begin: Mon May  5 12:09:58 2008
  expect: Tue May  6 10:59:06 2008; due: Mon Jun 23 07:21:57 2008 (49 days)
 
2)
 Index 4: folding now 40.6 X min speed; 7.67% complete
  server: 169.230.26.30:8080; project: 4520, "p4520_2CVR-16-10ns_minout"
  Folding: run 17, clone 66, generation 2; benchmark 0; misc: 500, 300
  issue: Mon May  5 12:09:59 2008; begin: Mon May  5 12:09:58 2008
  expect: Tue May  6 17:01:55 2008; due: Mon Jun 23 07:21:57 2008 (49 days)
 
3)
 Index 7: folding now 59.9 X min speed; 11.7% complete
  server: 169.230.26.30:8080; project: 4520, "p4520_2CVR-16-10ns_minout"
  Folding: run 24, clone 66, generation 2; benchmark 0; misc: 500, 300
  issue: Mon May  5 12:09:59 2008; begin: Mon May  5 12:09:58 2008
  expect: Tue May  6 07:43:23 2008; due: Mon Jun 23 07:21:57 2008 (49 days)
 
4)
 Index 3: folding now 44.5 X min speed; 8.67% complete
  server: 169.230.26.30:8080; project: 4520, "p4520_2CVR-16-10ns_minout"
  Folding: run 20, clone 66, generation 2; benchmark 0; misc: 500, 300
  issue: Mon May  5 12:09:59 2008; begin: Mon May  5 12:09:58 2008
  expect: Tue May  6 14:29:23 2008; due: Mon Jun 23 07:21:57 2008 (49 days)


Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Mon May 05, 2008 10:35 pm
by vvoelz
vvoelz wrote:Heads up for three new 'bonused' AMBER Core WUs coming onling in the next day or so:
p4522, p4523, and p4524. See http://fah-web.stanford.edu/cgi-bin/fahproject?p=4522 for point details.
Sorry folks, there were some problems with these. In the next few hours, I will be putting up three replacement WUs: p4525, p4526, and p4527. As before, see http://fah-web.stanford.edu/cgi-bin/fahproject?p=4525 for point details.

Vince
Pande Lab

Re: New Bonus Amber Core WU's [-advmethods option]

Posted: Mon May 05, 2008 11:37 pm
by uncle_fungus
Jeff_Grant wrote:I see some Intels on fahinfo.org for 4521 and 4519. I have one 4521 on an older athlonxp 1500 that I threw up there as well. As everyone gets these units, can you add to the database?
Once they projects appear on psummary, they'll automatically be added to fahinfo.org