• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Impressive PPD Output for a single user

Axdrenalin

Axdrenalin

[H]ard|DCer of the Month - Nov. 2009
2FA
Joined
Jan 28, 2004
Messages
6,301
Does anyone know anything about PDC who is folding for the Pande Lab team? He / She is currently pushing 3.7 million PPD a *day* average, practically carrying the whole team up the charts. :eek:

I think it's awesome that they can contribute to folding, but I was also wondering what kind of setup this guy / gal has got at their disposal to push that kind of data through on a daily basis. They've had a massive ramp up just in the past four weeks that they've contributed (started on June 25th 2009).

Just freaking AMAZING!!!
 
More then that, they are dropping units worth over 50K EACH... Must be running on beta SMP or GPU units that aren't in the wild yet. Largest unit I can find is a 25K SMP unit, whatever PDC is running on is at least twice as big as that, closer to 55K per unit.
 
PDC is Pande Lab themselves.

They get time on legit super-computers, and fold under that name when they do.

Hence the crazy 8 core WUs.
 
So if I have an 8 core box and run linux SMP I can get 55k WU's?
 
Probably not. ...But I believe if you contact them, I think they can set you up.
 
Also for these Real Supercomputers

That run some sort of linux right?
What Linux are they running? Is it just like Beowulf or is customized per setup?
I'm guessing you cant just install the linux SMP client can you?

Reason I ask is I was looking at Eka...world 4th fastest supercomputer.
And the hardware is nothing you cant buy from newegg.
All it is are Dual Quad Xeons....they are running dual 5340's with 16gb of FB-dimms, and 20gb/s SAS interconnects, with Gig ethernet failover, and another gig ethernet for Admin.
Granted they are running 1800 of those......and have an insanely fast storage backend.
But my point is they are not using proprietary hardware like Blue Gene.

If I wanted to spend the money and time I could physically build this using Retail stuff.
 
nitrobass24 said:
Contact who?

I mean I have a 16core rig in the making so im quite interested.
Click to expand...

The Folding@Home team members, I'm sure you can PM them on Foldingforum.org.
And by 8 cores they mean like 2 Quad CPUs. Hyperthreads don't count.
 
Kendrak said:
Me thinks the upload would choke a FIOS line.
Click to expand...

:D

My DSL line laughs at me when all SMP clients are trying to upload. Hell if I'm trying to do at most 3 of them at the same time, it's pretty damn ugly.

55k point WU must be an INSANELY large file.
 
capreppy said:
:D

My DSL line laughs at me when all SMP clients are trying to upload. Hell if I'm trying to do at most 3 of them at the same time, it's pretty damn ugly.

55k point WU must be an INSANELY large file.
Click to expand...

get a faster connection.. :p the smp WU's were never a problem for me.. thanks to comcasts 5mbit upload speed boost.. :p
 
I've got a server available that I have nothing else running on since I've done a fresh format and CENTOS 5 install, so we'll see how it goes after I change the -smp flag to 8 cores. It's a dual quad Xeon with 12 gigs of RAM, and I usually get 7 minute frames when running the -smp 4 flag.
 
Axdrenalin said:
I've got a server available that I have nothing else running on since I've done a fresh format and CENTOS 5 install, so we'll see how it goes after I change the -smp flag to 8 cores. It's a dual quad Xeon with 12 gigs of RAM, and I usually get 7 minute frames when running the -smp 4 flag.
Click to expand...

Must post screen shots!
 
Which Xeon's? With my dual e5420 box running -smp 8, frames finish in about 3-4 min.
 
tkam said:
Which Xeon's? With my dual e5420 box running -smp 8, frames finish in about 3-4 min.
Click to expand...


That's what I would expect. My QX9300 (C2Q @ 2.53 in my laptop) runs 6.5 minute frame times. I am running VMWare Workstation with an Ubuntu 8.04 Guest
 
tkam said:
Which Xeon's? With my dual e5420 box running -smp 8, frames finish in about 3-4 min.
Click to expand...

Yeah, looks about right. I'm getting 3:28 - 3:30ish frame times with the -smp 8 flag. I went ahead and added the same to my other two dual quad servers as well, to essentially double production out of them. I'm hosting VM's on those two for work related purposes, but it doesn't seem to bog them down at all using all 8 cores to fold with under Linux SMP.

Xeons are Intel® Xeon® processor E5345 (2.33 GHz, 1333MHz, 80W)

Just a quick camera shot, as I'm not proficient enough to figure out a screen shot in Linux yet. :rolleyes:
CentOS5.jpg
 
nitrobass24 said:
Contact who?

I mean I have a 16core rig in the making so im quite interested.
Click to expand...

They've said in the past that the only thing stopping them from making more exotic clients (mega-SMP, beowulf clusters, etc) was the lack of any such systems available to them. As was said earlier, try the folding forum. I believe it was kasson (sp?) who expressed interest before.
 
nitrobass24 said:
Also for these Real Supercomputers

That run some sort of linux right?
What Linux are they running? Is it just like Beowulf or is customized per setup?
I'm guessing you cant just install the linux SMP client can you?

Reason I ask is I was looking at Eka...world 4th fastest supercomputer.
And the hardware is nothing you cant buy from newegg.
All it is are Dual Quad Xeons....they are running dual 5340's with 16gb of FB-dimms, and 20gb/s SAS interconnects, with Gig ethernet failover, and another gig ethernet for Admin.
Granted they are running 1800 of those......and have an insanely fast storage backend.
But my point is they are not using proprietary hardware like Blue Gene.

If I wanted to spend the money and time I could physically build this using Retail stuff.
Click to expand...

More and more of the top supercomputers now are just room-fulls of blade servers running a tweaked version of Linux. You'll probably still see a lot of exotic networking configurations, but the rest is pretty much standard hardware.
 
Well Im gonna try and fake it on my laptop using Linux and some Virtualization to try it out for compatibility.
The performance will be absolute shit, but it should give me a idea of what is going to be possible when I get my farm back up and running.
 
Wow, that is some amazing PPD :eek:

I need to really start picking up production, and then we'll see about getting some 50k wu's :D
 
nitrobass24 said:
So if I have an 8 core box and run linux SMP I can get 55k WU's?
Click to expand...

Here are your answers ....

http://foldingforum.org/viewtopic.php?f=24&t=10697

We are preparing for the public release of a new work unit category: extra-large advanced methods work units. Some background information is provided here, keep watching this thread for more details.

  • Why a new work unit category?
We have some specific projects where we 1) have large simulation systems and 2) want to get results fast. As multi-core processors get more powerful, we can perform calculations on Folding@Home that previously required supercomputing clusters.

  • What's different about these work units from the donor perspective?
These work units are special SMP work units that have larger upload and download sizes, shorter deadlines, and require more memory and CPU resources. That's why we've created a new category.

  • Is there any points incentive for running these work units?
The base value of these work units corresponds roughly to what an SMP work unit using the A2 core would yield on an equivalent calculation.
However, because fast completion is a scientific priority for these work units, we are doing a **trial** of a new bonus scheme where faster WU completion yields a points bonus.

  • What systems can run these work units?
Right now, only Linux and OS/X systems can run these work units, and they require 8 or more cores. We prefer 8+ *physical* cores, although fast Core i7 machines that are dedicated folders have proven sufficient during the testing process. The points incentives are designed to match appropriate resources to points value; if your machine is marginal for the extra-large work units, you're probably better off running standard SMP.

  • Does this have any relation to the large-points value work units and recent high-scoring users?
Yes. The initial projects are 2681 and 2682, valued at ~25K points base. Although these point values seem high, the work units are correspondingly larger, so the base PPD (points per day) value is roughly comparable to standard SMP.

A collaborator has donated a large amount of compute time to this project; those clients were initially running under username Anonymous/team 1. To give proper credit for the donation, we have changed the username to PDC, team 1. During the period of this donation, there are at any time between 100 and 400 8-core clients running under this username (800-3200 cores total).

Please stay tuned for further details regarding the upcoming release.

-----END QUOTE FROM FF ------
 
Last edited:
Evil, thanks for posting some details on this. I checked again last night and noticed that the PDC was up to a daily average of over 6 million, so this certainly explains it. 100 to 400 8 core clients? Awesome computational power there...I'm hoping that I can get in on some of this with one of my servers when they go public with it all.

EDIT: Posted at 3:04 in the morning? Man, don't you ever SLEEP? :D
 
DAMN!!!!!

I've got my i7 running @ 4.2 under WC. I have been contemplating building a Dual i7 for a DB Server. I may just have to build 2 :D :D :D :D

 
Unfortunately, these new SMP units are just not good for me to run despite having several 8-core systems available. They seem to require too much RAM, are too large to upload for my crappy ISP cap, and are best not run with the GPU client. Moreover, the minimum clockspeed stated was something in the 2.4GHz range, and I don't think I can OC my slower systems through software in a pure Linux environment. So, it seems they're off the table for me this year unless someone can show me otherwise.

The fact Stanford stated they're not significantly better in PPD makes me wonder why anyone would prefer to run them instead of a multiple VM A2 setup in combination with a GPU client or more. A single GPU client alone will grant at least ~4000 PPD providing it's a half decent card. To spend the money on 8-core or higher hardware for a nominal bonus in SMP units and no GPU clients seems a bit like going backwards to me. If one has the hardware already and for one reason or another cannot run a single GPU client, I can see the need but other than that...
 
EvilAlchemist said:
A collaborator has donated a large amount of compute time to this project; those clients were initially running under username Anonymous/team 1. To give proper credit for the donation, we have changed the username to PDC, team 1. During the period of this donation, there are at any time between 100 and 400 8-core clients running under this username (800-3200 cores total).
Click to expand...

These numbers boggle my mind...
 
elleana said:
These numbers boggle my mind...
Click to expand...

I know.... Since July 1 the group has produced over 96 million points, and the PDC account was responsible for 93 million + of them. :eek: I've noticed them being within 1.1 years of overtaking the [H]orde...
 
Axdrenalin said:
I know.... Since July 1 the group has produced over 96 million points, and the PDC account was responsible for 93 million + of them. :eek: I've noticed them being within 1.1 years of overtaking the [H]orde...
Click to expand...
Yes, I don't believe I've seen an overtake this close since I don't even remember when. In late 2007, MPC was in an upsurge and they had us overtaken in about a year and a half if memory serves, but this is cutting it really close.

We lost ROC, and that's hurting us significantly. PDC may not be putting out this kind of production in the future, so it could very well be temporary, yet this year has seen a few upstarts vying for the #1 position. Bottom line, I foresee more big challengers in the future and what this means is we better be in shape.
 
We need to get Kyle to buy a Couple of the SuperMicro 2uTwin i7 systems.....to host HardOCP stuff on
4 nodes in a 2u enclsure.
Each Node should be good for around 25k PPD with current LinSMP
SMP2 or these 50k WU's might yeild different results tho.
 
nitrobass24 said:
We need to get Kyle to buy a Couple of the SuperMicro 2uTwin i7 systems.....to host HardOCP stuff on
4 nodes in a 2u enclsure.
Click to expand...
In the past, Kyle stated that he didn't want F@H to run on the [H] servers because of the increased heat. Any additional heat will likely translate to more down time, and the forums see enough crashes as it stands. On top of that is the increased wear and tear, which in the long run will probably mean higher running costs. I very well understand his stance on this after all I've gone through with my systems.
 
some server hosts dont like it either.. one of my friends works for a data center in southern california.. if the heat output on a server gets to high they just shut it off until the owner emails them to have it turned back on..
 
sirmonkey1985 said:
some server hosts dont like it either.. one of my friends works for a data center in southern california.. if the heat output on a server gets to high they just shut it off until the owner emails them to have it turned back on..
Click to expand...

Wow, not what I would call stellar service from my ISP. I figure if I'm paying my bills for them to host the content on, then it's their responsibility to make sure the hosting environment stays cool. :rolleyes:

But back on topic, I'm hoping that we can get some solid folders out there that will help contribute to the team. I just think its incredible how many PPD that they (PDC and team Pande) are shoveling out. The only way we're going to keep from getting mowed down eventually is if this is just a temporary push on their part. :eek:
 
Axdrenalin said:
But back on topic, I'm hoping that we can get some solid folders out there that will help contribute to the team. I just think its incredible how many PPD that they (PDC and team Pande) are shoveling out. The only way we're going to keep from getting mowed down eventually is if this is just a temporary push on their part. :eek:
Click to expand...
They had a similar surge a while ago and it didn't last more than a month or two, but it was nothing of this magnitude. I have no idea how long they will be having use of these servers. Probably someone else on the forum closer to Stanford would know.
 
Maybe we could get a bunch of the [H] folders here and try to collect donations $10 per person, and then around I dunno next May when the CC2010 is on we can throw that money at renting a supercomputer or cluster for like 2 days.

Longshot, but it would be pretty kickass.
 
nitrobass24 said:
Maybe we could get a bunch of the [H] folders here and try to collect donations $10 per person, and then around I dunno next May when the CC2010 is on we can throw that money at renting a supercomputer or cluster for like 2 days.

Longshot, but it would be pretty kickass.
Click to expand...

Unfortunately I don't think we could rent a supercomputer for 2 days. Maybe like 30 minutes with the amount of money we'd raise.
 
1: Get 4x 6 core AMD CPUs
2: Linux SMP, 3 instances of 8 cores each
3:
4: Profit.
 
nitrobass24 said:
Maybe we could get a bunch of the [H] folders here and try to collect donations $10 per person, and then around I dunno next May when the CC2010 is on we can throw that money at renting a supercomputer or cluster for like 2 days.

Longshot, but it would be pretty kickass.
Click to expand...

It would be better if that money could be used to fuel something for the long term rather than instantaneous gain. The CC is supposed to motivate us to ramp up our farms, not to find short term expensive ways to up PpD. What's 1M PpD for 14 days when for the same price you could get a machine that could do 30M in a year?
 
crazjayz said:
It would be better if that money could be used to fuel something for the long term rather than instantaneous gain. The CC is supposed to motivate us to ramp up our farms, not to find short term expensive ways to up PpD. What's 1M PpD for 14 days when for the same price you could get a machine that could do 30M in a year?
Click to expand...
Besides, following our lackluster performance this year, I would think future [H] CC participation is cast into serious doubt. It's just too difficult to mobilize a team with this many members. Especially when we're competing with teams that have mega-folders who are like a whole CC subteam all by themselves. So, I agree a long term focus is a far better strategy.
 
You must log in or register to reply here.
Back
Top