Are the p1499_tet_1499's as good as they seem to be?

[DanG]

13:37:33] Project: 1499 (Run 46, Clone 0, Gen 6)
[13:37:33]
[13:37:33] Assembly optimizations on if available.
[13:37:33] Entering M.D.
[13:37:40] Protein: p1499_tet_1499
[13:37:40]
[13:37:40] Writing local files
[13:37:43] Extra SSE boost OK.
[13:37:43] Writing local files
[13:37:43] Completed 0 out of 125000 steps (0)
[13:42:03] Writing local files
[13:42:03] Completed 1250 out of 125000 steps (1)
[13:46:24] Writing local files
[13:46:24] Completed 2500 out of 125000 steps (2)
[13:50:44] Writing local files
[13:50:44] Completed 3750 out of 125000 steps (3)
[13:55:05] Writing local files
[13:55:05] Completed 5000 out of 125000 steps (4)
[13:59:25] Writing local files
[13:59:25] Completed 6250 out of 125000 steps (5)
[14:03:45] Writing local files
[14:03:46] Completed 7500 out of 125000 steps (6)
[14:08:06] Writing local files
[14:08:06] Completed 8750 out of 125000 steps (7)
[14:12:26] Writing local files
[14:12:26] Completed 10000 out of 125000 steps (8)



By my math, it's about 4 minutes and 21 seconds per frame, 100 frames and 364 points per WU, which is about 1200 points a day, and almost 8500 per week.

Or is my math screwed up somewhere? For some reason EMIII doesn't show a point value for this protien, but the stanford site shows it as 364.

Can it be this good?

[muddocktor]

Yep, they are delicious. My Conroe can eat these at a rate of about 2000 ppd. :D They are the same basic series type as the earlier p147x/p1481/p1495/p1497 wu's. These wu's love a big L2 cache.

[Leonardo]

They are as good as they appear!

Now, if we can get the QMDs back! :hair: :bigggrin: :)

[DanG]

I'm 62% through the second one today, now if only the other core would pick these up, I'd be set.

[qparadox]

I'm getting 12 minutes / frame on one on my 3000+ (not oc'd), definitely not complaining about 400 PPD from that machine :).

[Oriane]

Are AMDers being discriminated against? :) I have yet to see a p14anything.

[Leonardo]

Check your client configuration. Make sure advanced methods (admethods) is set to 'Yes.'

[Oriane]

From client.cfg:


[settings]
...
asknet=no
bigpackets=yes
machineid=2
local=70

[http]
active=no
host=localhost
port=8080
usereg=no

[clienttype]
type=3

_______________
From FAH WIKI for client types:

S: type=3
O: Request work units without deadlines (no/yes) [no]?
O: Set -advmethods flag always, requesting new advanced
scientific cores and/or work units if available (no/yes) [yes]
N: This is option was added in client v5.04.

I pretty sure I have it right- yes?

I may simply have not drawn one yet and/or not have had EM running to log it when I did.

[Thermalfish]

Still havent seen any of these. Melts, melts everywhere! I cheer when a Villin shows up.

[csimon]

Still havent seen any of these. Melts, melts everywhere! I cheer when a Villin shows up.
Have you had any (beta WU) yet?

Note: Inadvertent reference to private beta WU removed.

-Prof :)

[Thermalfish]

No, I havent. Ive only scored maybe a dozen 600 pointers in 490 WUs

[csimon]

How much memory do you have?

[Thermalfish]

a mix of 512, 1024, and 2048.

[Leonardo]

Do you have Folding client configurations set to accept large downloads?

[Thermalfish]

Yeah they are. These are mostly powermacs. Maybe thats it.

[nonstop301]

I'm facing an error with the FAH console when it's about to start a task with p1499_tet_1499 which I inadvertently interrupted.

Here's the log I receive:

[06:10:42] + Core successfully engaged
[06:10:47]
[06:10:47] + Processing work unit
[06:10:47] Core required: FahCore_78.exe
[06:10:47] Core found.
[06:10:47] Working on Unit 01 [January 19 06:10:47]
[06:10:47] + Working ...
[06:10:47]
[06:10:47] *------------------------------*
[06:10:47] Folding@Home Gromacs Core
[06:10:47] Version 1.90 (March 8, 2006)
[06:10:47]
[06:10:47] Preparing to commence simulation
[06:10:47] - Looking at optimizations...
[06:10:47] - Created dyn
[06:10:47] - Files status OK
[06:10:49] - Expanded 857201 -> 12361133 (decompressed 1442.0 percent)
[06:10:49] - Starting from initial work packet
[06:10:49]
[06:10:49] Project: 1499 (Run 598, Clone 0, Gen 8)
[06:10:49]
[06:10:50] Assembly optimizations on if available.
[06:10:50] Entering M.D.
[06:10:57] Protein: p1499_tet_1499
[06:10:57]
[06:10:57] Writing local files
[06:10:57] Gromacs error.
[06:10:57]
[06:10:57] Folding@home Core Shutdown: UNKNOWN_ERROR
[06:11:01] CoreStatus = 79 (121)
[06:11:01] Client-core communications error: ERROR 0x79
[06:11:01] Deleting current work unit & continuing...
[06:11:21] - Preparing to get new work unit...
[06:11:21] + Attempting to get work packet
[06:11:21] - Connecting to assignment server
[06:11:22] - Successful: assigned to (171.64.122.134).
[06:11:22] + News From Folding@Home: Welcome to Folding@Home
[06:11:22] Loaded queue successfully.
[06:11:35] + Closed connections
[06:11:40]
[06:11:40] + Processing work unit
[06:11:40] Core required: FahCore_78.exe
[06:11:40] Core found.
[06:11:40] Working on Unit 02 [January 19 06:11:40]
[06:11:40] + Working ...
[06:11:40]
[06:11:40] *------------------------------*
[06:11:40] Folding@Home Gromacs Core
[06:11:40] Version 1.90 (March 8, 2006)
[06:11:40]
[06:11:40] Preparing to commence simulation
[06:11:40] - Looking at optimizations...
[06:11:40] - Created dyn
[06:11:40] - Files status OK
[06:11:42] - Expanded 857201 -> 12361133 (decompressed 1442.0 percent)
[06:11:42] - Starting from initial work packet
[06:11:42]
[06:11:42] Project: 1499 (Run 598, Clone 0, Gen 8)
[06:11:42]
[06:11:43] Assembly optimizations on if available.
[06:11:43] Entering M.D.
[06:11:49] Protein: p1499_tet_1499
[06:11:49]
[06:11:49] Writing local files
[06:11:49] Gromacs error.
[06:11:49]
[06:11:49] Folding@home Core Shutdown: UNKNOWN_ERROR
[06:11:52] CoreStatus = 79 (121)
[06:11:52] Client-core communications error: ERROR 0x79
[06:11:52] Deleting current work unit & continuing...

Folding@Home Client Shutdown.

I don't know why the Gromacs error occurs and whether I have to change something in the console configuration to prevent this cycle of errors that prevents the task from starting properly.

Thank you in advance for all your comments and any advice you wish to provide.

[Leonardo]

Dan, that error may be due to memory faults. Maybe it's time to consider Memtest? Is your computer overclocked?

[nonstop301]

Hi Leonardo and thank you for your reply.

I'm not sure why the error occurred and whether it was related to the abrupt interruption of the 1499_tet_1499 task or bad memory like you suggest.

I made an adjustment in the FAH console configuration just now and removed the -advmethods flag. The Gromacs core fortunately started properly and I got a new task so I'm folding again.

I think the incorrect interruption of the 1499_tet_1499 must have had a negative effect on the client.cfg file and prevented the correct use of -advmethods flag. Once the task I received now completes I will add -advmethods again and hopefully there will be no client-core errors.

______
//EDIT:

I will perform the Memtest after the current work unit completes, to ensure there is nothing wrong with the memory stick of the computer.

The processor is overclocked and it has been performing very well for a low grade Sempron socket 754 :)

[Oriane]

nonstop-

Yes, run Memtest (http://memtest.org/) like Leo suggested. Your CPU might be overheating (when was the last time you cleaned out your PC?) as well. Keep checking your logs. Hardware problems can be intermittent and your WUs may continue to fail and abort even though the FAH servers will continue to give you new WUs. Try backing off your overclock if you still have problems.

If everything checks out but Wus are still failing, post with your FAH log. You might have a bad core.

Thermalfish-

Two days after my last post in this thread I FINALLY got a 1499 (with my posted client.cfg settings) after many moons. I might add they run pretty good on an AMD too. However, I doubt having a Mac is the issue- I think it might be what your assignment server has when.

[Leonardo]

I think the incorrect interruption of the 1499_tet_1499 must have had a negative effect An incorrect or too abrupt shutdown will will often corrupt the work unit itself, not the client or the config.sys file. But yes, the interruption could indeed be the culprit here.

[csimon]

Have you had any tethered vesicles yet? They are over 800 points and fold pretty fast too.

[profdlp]

What's all this rot about "feathered testicles"? :wtf:

[csimon]

What's all this rot about "feathered testicles"? :wtf:

omg ...he went there. :eek3:

Hehe thanks. I thought I had edited that myself and only now am I finding out that I only quoted myself. :aol:

Thank you Spike and Prof. :csimon:

[nonstop301]

I did a one of those tethered vesicles recently and they are pretty fast like you say and deliver good points too :)

It's back to p2125_lambda_5way_melt_4_10011 at the moment but I'm not complaining :)

[SPIKE09]

I nearly had a keyboard meets red wine incident, when Leonardo 1st posted about the "feathered testicles" at the FCF, he has the eagle avvy there too, wondered why an eagle was complaining about something like that.
back to the scheduled programming, the P149x series are great until you get 2 on the same machine frame times rose from 7min 30s with a five way patty on the other core to 20 mins, with 2 1499's. this is on an E6400 with 2 Gb of DDR2 667.

[csimon]

I nearly had a keyboard meets red wine incident, when Leonardo 1st posted about the "feathered testicles" at the FCF, he has the eagle avvy there too, wondered why an eagle was complaining about something like that.
back to the scheduled programming, the P149x series are great until you get 2 on the same machine frame times rose from 7min 30s with a five way patty on the other core to 20 mins, with 2 1499's. this is on an E6400 with 2 Gb of DDR2 667.
Check your Extra SSE boost OK.

[SPIKE09]

It is they are both hammering the CPU bigtime, they just don't like two good wu's running at the same time. probably both competing for the same system resource's.

[Donut]

Hmmm, Spike you have me re-thinking my next rig now. Do these wu's (I haven't seen any in a while) use a lot of memory, or is it possible they're bandwidth (fsb) hogs?

The reason I'm wondering is I was going to hold out for the 4xxx series Intels with the lower fsb.

[csimon]

It is they are both hammering the CPU bigtime, they just don't like two good wu's running at the same time. probably both competing for the same system resource's.
Strange. When I get double 149X's my production triples to almost 1200ppd.

[Leonardo]

The rate of production for 149X on my Pentium D9XX CPUs is unchanged if they are doubled up as compared to the machine folding one 149X and something else. With 1495s, 1497s, and 1499s each of the machines (signature) will produce about 1330 PPD. No such luck for the last few days. Most of the units have been 212Xs. Good for science, not good for ego. :rolleyes2

[csimon]

The rate of production for 149X on my Pentium D9XX CPUs is unchanged if they are doubled up as compared to the machine folding one 149X and something else. With 1495s, 1497s, and 1499s each of the machines (signature) will produce about 1330 PPD. No such luck for the last few days. Most of the units have been 212Xs. Good for science, not good for ego. :rolleyes2

Didn't you and mudd each have a core2duo a while back?

[Leonardo]

Not I. I'd love to upgrade these machines, but that would mean new CPUs and motherboards. I got all these D930s and the 915 for very little and already had the motherboards and DRAM. I've got my daughter's wedding to pay for this year in May so no major upgrades for a while. I may just skip Core Duo and dive right into quad core, depending on how long it takes before their prices become reasonable. I guess that will all depend on how much pressure AMD puts on Intel. AMD is the only reason Core Duo is affordable now.

[csimon]

Not I. I'd love to upgrade these machines, but that would mean new CPUs and motherboards. I got all these D930s and the 915 for very little and already had the motherboards and DRAM. I've got my daughter's wedding to pay for this year in May so no major upgrades for a while. I may just skip Core Duo and dive right into quad core, depending on how long it takes before their prices become reasonable. I guess that will all depend on how much pressure AMD puts on Intel. AMD is the only reason Core Duo is affordable now.
Don't remember who else it was then. You're doing well with what you have though I must say. Priorities!:thumbsup:

[Leonardo]

The D9 series was Intel's second generation dual core CPUs. They are 65nm and comparatively cool, that is, if you compared with the D8XX series (Smithfield) - which are essentially dual Prescott. The D9XX series are Presler. Clock for clock though, the Conroe (Intel 3rd generation dual core) will outpace the D9s by 30% or perhaps even more. Two things the Preslers and Conroes have in common -- 65nm construction and great overclocking.

EDITED: goofed up core code names - fixed now

[Oriane]

... These wu's love a big L2 cache....

That's what I think -

Toledo core @ 2.5GHz


...
[13:09:28] Project: 1499 (Run 154, Clone 3, Gen 14)
[13:09:28]
[13:09:28] Assembly optimizations on if available.
[13:09:28] Entering M.D.
[13:09:35] Protein: p1499_tet_1499
[13:09:35]
[13:09:35] Writing local files
[13:09:42] Extra SSE boost OK.
[13:09:43] Writing local files
[13:09:43] Completed 0 out of 125000 steps (0)
[13:24:35] Writing local files
[13:24:36] Completed 1250 out of 125000 steps (1)
[13:39:28] Writing local files
[13:39:28] Completed 2500 out of 125000 steps (2)
[13:54:21] Writing local files
[13:54:21] Completed 3750 out of 125000 steps (3)
[14:09:14] Writing local files
[14:09:14] Completed 5000 out of 125000 steps (4)
[14:24:06] Writing local files
[14:24:07] Completed 6250 out of 125000 steps (5)
[14:38:59] Writing local files
[14:38:59] Completed 7500 out of 125000 steps (6)
[14:53:51] Writing local files
[14:53:52] Completed 8750 out of 125000 steps (7)
[15:08:44] Writing local files
[15:08:44] Completed 10000 out of 125000 steps (8)
[15:23:36] Writing local files
[15:23:36] Completed 11250 out of 125000 steps (9)
[15:38:29] Writing local files
[15:38:29] Completed 12500 out of 125000 steps (10)
[15:53:21] Writing local files
[15:53:21] Completed 13750 out of 125000 steps (11)
[16:08:13] Writing local files
[16:08:13] Completed 15000 out of 125000 steps (12)
...


The Toledo has 1MB/core compared to Conroe's 2MB/core- this, I think, speaks for itself.

[Oriane]

… compared to Conroe's 2MB/core- ...

Sorry- this is not correct- I’m thinking sooo AMD/Pentium D. The Conroe (the E6600 that DanG uses (http://www.short-media.com/forum/showthread.php?t=52539)) has 4MB of shared (http://www.extremetech.com/article2/0,1697,1988795,00.asp) cache- a feature that could be as significant as being twice the size of my Toledo’s total core-split cache.

I don’t know much about cache operation, but I have to appreciate that his Conroe outperforms my system on this WU by 340%- and this looks to me like the most significant reason why.

[Thermalfish]

Still havent had any of these.

Firing up a dual Xeon 2.8 with 2 Gigs RAM today. See what happens.

(Sorry Mirage)

Nup: just set it up and what do I get? A pair of 2125 melts. :hiding:

[SPIKE09]

I think the 2nd level cache is the big thing here too as the e6400 only has 2Mb not the 4Mb of it's big brothers.

[mirage]

Still havent had any of these.

Firing up a dual Xeon 2.8 with 2 Gigs RAM today. See what happens.

(Sorry Mirage)

Nup: just set it up and what do I get? A pair of 2125 melts. :hiding:

A dual-core Opteron is waiting here to start folding. If I can bring it online, you might not catch me so soon. How about eating the dust a little longer :rarr:

[Thermalfish]

Hehe ok! Ive just gotten back from a multi-day festival. I ate several kilos of dust already!

I am only 10 days behind you. Better get a move on!!

[mirage]

Every move has the right moment as Confucius (probably) said wisely :D

[Donut]

Don't tell anybody, I got one!

EM3 says 54min/frame on a PD805.:bigggrin: