Folding new release: extra-large work units

[SciFrog]

http://foldingforum.org/viewtopic.php?t=10697

I hope to get a 3Ghz Clovertown 8-core this week to reach 17k PPD just for this rig :D

That should help the team. Anyone else?

[lyzardking]

http://foldingforum.org/viewtopic.php?t=10697

I hope to get a 3Ghz Clovertown 8-core this week to reach $17k PPD just for this rig :D

That should help the team. Anyone else?

Is yours (if you think you can take it)

;)

[SciFrog]

I should have it sometime next week, first unit out hopefully by Friday...

[lyzardking]

I should have it sometime next week, first unit out hopefully by Friday...

To hearing how these big units work out.

[twoodcc]

I should have it sometime next week, first unit out hopefully by Friday...

care to tell us the full specs of the machine your getting?

[SciFrog]

Rig has been delayed but is on its way.
I am getting a 8 core 3Ghz (Clovertown X5365?), 13GB ram, RAID card. I really wanted the 2009 models but the new price of the mid range threw me off entirely, too bad for Apple. The 2007 model is 20% slower for half the price and still under warranty. I just need to find a nice card to replace the Radeon X1900.

I now expect to run these at 35mins per frame, 2.5 days, over 20k PPD :eek:

There is a big bonus for returning quickly. A 3.2Ghz Nehalem might bring in 40k PPD, returning the unit in 1.5 days.

That and two other Core 2 duo should put me at #1 cruncher in the team... And top 3 units in a few months.

PS: this is the same PPD as 13 Mac Minis...

[twoodcc]

Rig has been delayed but is on its way.
I am getting a 8 core 3Ghz (Clovertown X5365?), 13GB ram, RAID card. I really wanted the 2009 models but the new price of the mid range threw me off entirely, too bad for Apple. The 2007 model is 20% slower for half the price and still under warranty. I just need to find a nice card to replace the Radeon X1900.

I now expect to run these at 35mins per frame, 2.5 days, over 20k PPD :eek:

There is a big bonus for returning quickly. A 3.2Ghz Nehalem might bring in 40k PPD, returning the unit in 1.5 days.

That and two other Core 2 duo should put me at #1 cruncher in the team... And top 3 units in a few months.

PS: this is the same PPD as 13 Mac Minis...

wow. now wait a min. are you getting a 2008 mac pro?

[SciFrog]

2007, revision 2,1 Clovertown.
The 2008 is revision 3,1 (a little confusing)...

[twoodcc]

2007, revision 2,1 Clovertown.
The 2008 is revision 3,1 (a little confusing)...

now what OS are you going to be folding in?

[Eidorian]

I can get 2100 - 2700 PPD off of my lowly 4830. :eek:

That's nothing compared to a monster 9600 GSO folding system either.

[SciFrog]

SL but it doesn't really matter...

GPU Folding is windows only... No Microsoft here...
Now if only someone figures out WINE GPU, the 8 cores and a GTX 285 could be a killer rig.

[twoodcc]

I can get 2100 - 2700 PPD off of my lowly 4830. :eek:

That's nothing compared to a monster 9600 GSO folding system either.

but do you fold with it?

SL but it doesn't really matter...

GPU Folding is windows only... No Microsoft here...
Now if only someone figures out WINE GPU, the 8 cores and a GTX 285 could be a killer rig.

oh ok. yeah i've just got the smp running on my mac pro. but i'm running more on my pc though (GPU and 2 SMPs right now, but still trying to figure the best method)

[SciFrog]

What is are your setups? Maybe we can help.

[Eidorian]

but do you fold with it?I do fold with my 4830. It's not the speed that a nVidia card can turn out but it's fast enough for my.

[twoodcc]

What is are your setups? Maybe we can help.

i currently have 4 machines that are folding. 2 are laptops, and i won't continue to fold with those for much longer (white macbook 2.16, 2.8 umbp).

i have a 2006 4-core mac pro 2.66, running the smp in leopard. this is also my web server, with priceless info on it. so i just bought an i7 pc that will fold, and be a backup for the mac pro (couldn't afford a new mac pro).

right now on the pc i have the GPU running in vista64 (geForce GTX 260) and i have 2 VMs folding with notfred's images (i followed a tutorial on another folding forum). now according to FahMon, the gpu is getting over 6500 PPD, while the 2 VMs are getting around 1800 PPD each. so total is around 10300 PPD for the machine.

but i according to others, i should be getting around 3000 PPD for each VM, or even more, and i should get around 15000 - 17000 PPD out of this machine.

any tips?

I do fold with my 4830. It's not the speed that a nVidia card can turn out but it's fast enough for my.

oh ok. i didn't know that you were folding. glad you are though!

[Eidorian]

The majority of my folding for Team MacRumors has been off of my GPU. :p

[twoodcc]

The majority of my folding for Team MacRumors has been off of my GPU. :p

well right now my GPU client is killing my 2 CPU clients put together. but i don't think it's supposed to be that way ...

here is a pic of my FahMon :

[SciFrog]

I read GPU clients with Nvidia chips don't use much CPU...

[twoodcc]

I read GPU clients with Nvidia chips don't use much CPU...

the problem is that the 2 smp clients that are running in 2 different VMs aren't using all my CPUs

[Eidorian]

well right now my GPU client is killing my 2 CPU clients put together. but i don't think it's supposed to be that way ...I don't see much of a reason to run the CPU client anymore when I can churn out many more units for less power on my video card.

[twoodcc]

I don't see much of a reason to run the CPU client anymore when I can churn out many more units for less power on my video card.

right. but i'd like to be able to run the CPU to the max when i'm not using the computer.

[SciFrog]

First bigadv unit is done. 41.5 mins per frame, 2.05 faster than deadline resulted in 51.6k points thus 17.6k PPD. Not bad eh...

It looks like the Harpertown and Nehalem can do better per Ghz...

Now we need more team members to try them out.

[twoodcc]

well i'm at 85% now on my first one. just over 9h left! i should finish in about 2.5 days total. hopefully it'll be done when i wake up tomorrow!

[Eidorian]

nVidia hasn't released a GPU worth buying to get more ppd over my ATI one. :(

[twoodcc]

nVidia hasn't released a GPU worth buying to get more ppd over my ATI one. :(

really? not even the GTX 295 or whatever?

[Eidorian]

really? not even the GTX 295 or whatever?I don't have a use for such a monster. You're better off getting multiple GTS 250 or 9600 GSO for folding.

[twoodcc]

I don't have a use for such a monster. You're better off getting multiple GTS 250 or 9600 GSO for folding.

oh ok. i have a GTX 260 i believe. i've never been able to keep up with video cards that much

[Eidorian]

oh ok. i have a GTX 260 i believe. i've never been able to keep up with video cards that muchHere is what you'd build (http://www.bit-tech.net/hardware/buyers-guide/2009/09/15/what-hardware-should-i-buy-september-2009/5) if your life is folding.

[twoodcc]

Here is what you'd build (http://www.bit-tech.net/hardware/buyers-guide/2009/09/15/what-hardware-should-i-buy-september-2009/5) if your life is folding.

i don't know. i really like my i7. based off of that, if i just added another GTX 260 to my machine, i'd be set.

well my first bigadv unit has been completed! just waiting to see what my bonus will be

[SciFrog]

Maybe my estimate is too slow, someone posted this:

Project #: 2683
Average time/frame: 35m 2s
CPU: Xeon E5520 @ 2.26 GHz
# of CPU sockets: 2
# of cores: 8
# of fahCore_A2 processes running: 16

RAM installed: 10GB
RAM used by FAH: 5.20GB
OS / Linux kernel: Mac OSX 10.5.6

That would be 22k PPD.

[twoodcc]

Maybe my estimate is too slow, someone posted this:

Project #: 2683
Average time/frame: 35m 2s
CPU: Xeon E5520 @ 2.26 GHz
# of CPU sockets: 2
# of cores: 8
# of fahCore_A2 processes running: 16

RAM installed: 10GB
RAM used by FAH: 5.20GB
OS / Linux kernel: Mac OSX 10.5.6

That would be 22k PPD.

now that looks better. i wonder if the number would be even better if running 10.6?

[twoodcc]

well, 55,378 was what i got for my 1st one. that's over 22k PPD. i'm happy with it!

[SciFrog]

Great, now if we get three more crunchers we will double the team's output...

[twoodcc]

Great, now if we get three more crunchers we will double the team's output...

yeah you're right. which is kinda crazy actually

[mikes63737]

How much bandwidth do these WUs use? I have a 10GB rolling weekly cap, and if it doesn't use too much (less than... say ~700MB/week), I'll gladly add my Linux box to the team. It's a Core i7 920 w/ 12GB DDR3 RAM.

[SciFrog]

I think you'd be under 300mb a week...

[mikes63737]

OK, that sounds good then. I will add my PC to the MR team.

[twoodcc]

OK, that sounds good then. I will add my PC to the MR team.

alright! should put up good numbers. i also have an i7 920. running ubuntu

[mc68k]

my bigadv unit should be done today. 90% done as of now. each step was taking about 41-45 mins on a 3GHz x 2 MP 2,1. only concern is the passkey WUs, even though ive completed a ton of these w/o the passkey i just started the passkey thing friday. i should have the 10WU passkey minimum for the bonus. 25-50K WUs make WUs <1K pts seem like a complete waste of energy

also i think the reason it was rebooting after every ctrl-c before was the beta executable from the ~kasson link. a switch to the normal 6.24.1 has been flawless so far *knock on wood*

[twoodcc]

my bigadv unit should be done today. 90% done as of now. each step was taking about 41-45 mins on a 3GHz x 2 MP 2,1. only concern is the passkey WUs, even though ive completed a ton of these w/o the passkey i just started the passkey thing friday. i should have the 10WU passkey minimum for the bonus. 25-50K WUs make WUs <1K pts seem like a complete waste of energy

also i think the reason it was rebooting after every ctrl-c before was the beta executable from the ~kasson link. a switch to the normal 6.24.1 has been flawless so far *knock on wood*

sounds good. i'm almost done with my second one. i should be able to squeeze in a third just before i move back to my apartment, so i'll be down for a few days then.

i will have to test out ctrl-c soon after this one finishes (91% now)

[SciFrog]

So I can keep the #1 cruncher spot for a few more days? ;)

[twoodcc]

So I can keep the #1 cruncher spot for a few more days? ;)

haha. you will at the end of the week. i should jump up tonight though

[mc68k]

if i could use all these i would

http://img84.imageshack.us/img84/581/screenshot20090921at524.png

[SciFrog]

Just the 8 cores would be enough ;)

[twoodcc]

if i could use all these i would

http://img84.imageshack.us/img84/581/screenshot20090921at524.png

now that would be nice.

finally got done uploading 2nd bigadv unit. this time it took 2d12h24mn. a little longer i think b/c i was using the machine some today

oh, and it took like 36 min to upload to stanford

[twoodcc]

my bigadv unit should be done today. 90% done as of now. each step was taking about 41-45 mins on a 3GHz x 2 MP 2,1. only concern is the passkey WUs, even though ive completed a ton of these w/o the passkey i just started the passkey thing friday. i should have the 10WU passkey minimum for the bonus. 25-50K WUs make WUs <1K pts seem like a complete waste of energy

also i think the reason it was rebooting after every ctrl-c before was the beta executable from the ~kasson link. a switch to the normal 6.24.1 has been flawless so far *knock on wood*

so i noticed that you finished yours, but you didn't get the bonus. do you think it's b/c of the passkey thing, or did you not meet the deadline?

[jmann]

I am having such difficulty with this. It will run for a while and then it will say something about the cores not working, and how it is going to sleep and then I have to restart terminal to get it going again. And how do check how much/what percentage of these WUs I have done?

[Tower-Union]

if i could use all these i would

http://img84.imageshack.us/img84/581/screenshot20090921at524.png

Why can't you? Just curious how your in a position to "see" all those computers, but can't touch them.

[SciFrog]

You need to have folded 10 units under the passkey to get the bonus. You also need to send back at least 80% of your units before the deadline, which everyone reading this likely does.

jmann: is your machine stable? Does it sleep or gets restarted?

Read the log file to see where you are, or use the FAHWudget...

[SciFrog]

Why can't you? Just curious how your in a position to "see" all those computers, but can't touch them.

Work ;) :eek:

[mc68k]

I am having such difficulty with this. It will run for a while and then it will say something about the cores not working, and how it is going to sleep and then I have to restart terminal to get it going again. And how do check how much/what percentage of these WUs I have done?

yeah the default location if you're not using -local is /Users/yourname/Library/Folding@Home

in there you should unitinfo.txt which you could open with TextEdit, vi, pico, cat etc

[mc68k]

Why can't you? Just curious how your in a position to "see" all those computers, but can't touch them.
yeah like scifrog said it's work. i have admin access but some are laptops which would fry and i would have to take care of it. some are workstations of people that have been using unix since before i was born so they would know. a lot of people keep all their apps open and never reboot (windows or ignorant mentality) which uses all the memory. some are servers we have (mail, backup, webdev/serving, storage) that i wouldn't touch. for me it's utilizing resources that aren't being used and wouldn't have an impact on the user. some wouldn't be worth the effort. if a macpro is worth 25k-50k every 2.5-4 days (now) then a ppc based comp is worthless (points wise for me) for folding. keeping 5 ppc or early intel machines up and going would be way too much work. running large memory demanding units on machines 24/7 can cause intermittent problems with daily use. imacs already run hot with just web browsers and M$ Office open

folding for me is just a hobby, it's not worth stressing over, breaking machines, or getting in trouble

[mc68k]

second WU i got this-- what a bunch of *****

Making 1D domain decomposition 8 x 1 x 1
starting mdrun 'SINGLE VESICLE in water'
6500000 steps, 26000.0 ps (continuing from step 6250000, 25000.0 ps).
[02:38:46] pleted 2500 out of 250000 steps (1%)
[03:23:18] Completed 5000 out of 250000 steps (2%)
[04:07:24] Completed 7500 out of 250000 steps (3%)
[04:52:30] Completed 10000 out of 250000 steps (4%)
[05:38:28] Completed 12500 out of 250000 steps (5%)
[06:23:18] Completed 15000 out of 250000 steps (6%)
[07:08:39] Completed 17500 out of 250000 steps (7%)
[07:54:30] Completed 20000 out of 250000 steps (8%)
[08:41:30] Completed 22500 out of 250000 steps (9%)
[09:26:36] Completed 25000 out of 250000 steps (10%)
[10:12:11] Completed 27500 out of 250000 steps (11%)
[10:57:36] Completed 30000 out of 250000 steps (12%)
[11:42:50] Completed 32500 out of 250000 steps (13%)
[12:28:07] Completed 35000 out of 250000 steps (14%)
[13:12:41] Completed 37500 out of 250000 steps (15%)
[13:59:11] Completed 40000 out of 250000 steps (16%)
[14:45:56] Completed 42500 out of 250000 steps (17%)
[15:31:24] Completed 45000 out of 250000 steps (18%)
[16:17:03] Completed 47500 out of 250000 steps (19%)
[17:02:18] Completed 50000 out of 250000 steps (20%)
[17:46:59] Completed 52500 out of 250000 steps (21%)
[18:31:10] Completed 55000 out of 250000 steps (22%)
[19:17:03] Completed 57500 out of 250000 steps (23%)
[20:04:04] Completed 60000 out of 250000 steps (24%)
[20:50:31] Completed 62500 out of 250000 steps (25%)
[21:36:58] Completed 65000 out of 250000 steps (26%)
[22:24:21] Completed 67500 out of 250000 steps (27%)
[23:11:16] Completed 70000 out of 250000 steps (28%)
[cli_4]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[cli_7]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[cli_3]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[cli_6]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[cli_0]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[cli_5]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[cli_2]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[cli_1]: aborting job:
Fatal error in MPI_Sendrecv: Error message texts are not available
[23:56:28] CoreStatus = 1 (1)
[23:56:28] Sending work to server
[23:56:28] Project: 2681 (Run 10, Clone 7, Gen 25)
[23:56:28] - Error: Could not get length of results file work/wuresults_05.dat
[23:56:28] - Error: Could not read unit 05 file. Removing from queue.
[23:56:28] - Preparing to get new work unit...
[23:56:28] Cleaning up work directory
[23:56:28] + Attempting to get work packet
[23:56:28] - Connecting to assignment server
[23:56:28] - Successful: assigned to (171.67.108.22).
[23:56:28] + News From Folding@Home: Welcome to Folding@Home
[23:56:29] Loaded queue successfully.
[23:57:15] + Closed connections

[SciFrog]

This is exactly what happened to me on my third unit when my network crashed... again...

[mc68k]

i think i'm gonna do this weekend only. there's too many variables that seem to cause it to wedge. maybe that's why the program is labeled beta ;)

[SciFrog]

Well you need to have a stable network, maybe manual IP adresses would help? I hope my 1.5 year old Airport Base is not going down in flames.

[twoodcc]

well luckily i haven't had that error yet. i'm on my third one now, but running with the -oneunit flag b/c i will be leaving out of here tomorrow. i'm at 65% done on the third one, so it should be done before i leave. also, i have had my system freeze twice on this WU for some reason. granted, i was using the system while folding, but still. it didn't crash before.