Holy Work Unit, batman!

P

poddo

2005 [H]orde|Bowl Champion
Joined
May 1, 2005
Messages
464
I just got a 414 pointer on my dual core pentium, and its taking almost an hour per frame! EM3 says it will take 88 hours to finish it. Thats a lot of time for a 414, don't you think?
 
what's your CPU?


Is the protien a "p2102_lambda_5way_melt_0_00000"? that's the only 414point protein I see on Stanford's Protien summary page.... It's a Gb Gromac... could take awhile depending on the CPU..... I don't think I've folded any of those yet....... I'll check when I get home


Keep on Folding!! For the [H]orde!!

 
yes, it is one of the lambda melters. I got one on my work computer yesterday too. This comptuer here is a pentium D 820, with a gig of ram. the 2nd core is doing a tinker.
 
Yeah, I just got one of the 414 point GBGromacs. On my P4 2.8 it was taking around 50-55 minutes per frame. Not the greatest PPD I've seen before.

 
I've got one of those on my work box as well, and as its stands, it's working out to about (and mind you, this is real rough) ~6.4 PPH. If someone wants to double check the math, it's the same lambda 5 way melt, 414 points, processing for 13 hours and is 20% complete. That is on a P4 3.6 at stock.

 
I guess they're just slow folders. Oh well. Lots of points when the finally drop.
 
I've got a couple of these buggers going as well...one on an athlon and one on an opteron. I'll post stats when I get home tonight on them.

 
doing about 68-69 minutes/frame on my P4 2.4C @ 3.0..... but the other core has a 113 pt GbGromac...


Keep on Folding!!

 
Yeah my 170 has one on each core and they're going at about 31 mins a frame. Kinda slow, but more exciting than the dumb small gromacs.

 
I just got one of those as well on my A64 @2.4GHz. It's taking about 42 minutes per frame if I'm not playing games or anything. It's a pretty big mofo.

 
OSUguy98 said:
what's your CPU?


Is the protien a "p2102_lambda_5way_melt_0_00000"? that's the only 414point protein I see on Stanford's Protien summary page.... It's a Gb Gromac... could take awhile depending on the CPU..... I don't think I've folded any of those yet....... I'll check when I get home


Keep on Folding!! For the [H]orde!!

Click to expand...

That's the one. I've got one going on an Athlon XP 2400+ w/ 256MB ram running OpenBSD. It's cranking at 111PPD and 4.65PPH. It's 9 frames in and will take another 80 hours to complete. Ouch.
 
Yeah I got a couple too. My P4 3.6 is doing 1% every 41 minutes. My 3.2 is doing the same in 45 minutes. This is about 6.05 pph and 5.52 pph respectively. EMIII doesn't even know what it is. Is there a way to update EMIII with new wu project types?

-joeb0b
 
Ive got one on my work computer here.

p2102_lambda_5way_melt_0_00000

Work computer is a P4 3.0HT w 512 ram.

Currently sitting at about 50 min per frame, running at approx 5 points per hour.

Not terribly impressed, specially considering thix box WAS running those yummy
QMD's


 
My 2.4 XeonHT is running at 61 min per frame here at work. I miss the QMDs, I don't think it had ever received anything else until now.
 
joeb0b said:
Yeah I got a couple too. My P4 3.6 is doing 1% every 41 minutes. My 3.2 is doing the same in 45 minutes. This is about 6.05 pph and 5.52 pph respectively. EMIII doesn't even know what it is. Is there a way to update EMIII with new wu project types?

-joeb0b
Click to expand...

http://www.em-dc.com/

Click on downloads, and grab the latest data file. Follow the instructions on the page.
 
yeah, I just had to download the protien update thing to get EM3 to make it work.
 
Four of my boxes grabbed this one, and are working on them.

(2) P4 2.4C w/512 ram at 96 and 94 hours total.
(1) P4 2.66 w/512 ram at 84 hours total.
(1) A64 3700+ w/1GB ram at 63 hours total

Gonna have a nice point drop this weekend.

 
It's a chewy one for sure. PpDpGHz sucks too.

~53 min per frame on a XP @ 1733, 512 DDR running Win2K.

Edit: After further review, the PpDpGHz is about mid range for current work units excluding the 364 and 600 point gromacs.
 
I got one of the 414 pointers today also. protein 1258 - p2102_lambda_5way_melt_0_00000

It's taking 37:20 per frame on one core of my dual 3800 (oc'd to 2.5g).

By comparison, the other core is running a 586 pointer, which is only taking 26:25 per frame. (protein 920 - p2364_Protein_A_wt_Sosnick).

Doesn't seem right.

meisterbrau

 
doesn't seem right indeed, but we're all having the same results. I guess its just a beast. *shrug*
 
p2102_lambda_5way_melt_0_00000
Celeron 1.8 GHz
Time per frame: 1:34:37

p2102_lambda_5way_melt_0_00000
P4 3.0 GHz OC
Time per frame: 50:23

At least my big processors got the Big DGromacs. Some what closer to the previous PPD that I was getting with QMDs. Sorta... :(
 
pageian said:
Approx. 63 minutes a frame on an Athlon XP 2.1 GHz.
Click to expand...

You may want to check on this box. That's a pretty good bit slower than it should be.
I'm getting 00:52:50 to 00:53:05 on a XP @ 1733.
 
meisterbrau said:
I got one of the 414 pointers today also. protein 1258 - p2102_lambda_5way_melt_0_00000

It's taking 37:20 per frame on one core of my dual 3800 (oc'd to 2.5g).

By comparison, the other core is running a 586 pointer, which is only taking 26:25 per frame. (protein 920 - p2364_Protein_A_wt_Sosnick).

Doesn't seem right.

meisterbrau

Click to expand...

Hmm I woulda thought the 3800 core at 2.5 would be doing better than 37:20. My Opty at 2.63 is doing them in 33 flat. Wonder if the cache is effecting it at all?
 
My Opteron 175 at 2.2 is getting about 40min/frame for 148 PPD while my Sempron 3000 at 2.0 is getting about 49 min, 20sec/frame for about 120PPD.

My Sempron averages about 110 PPD normally checking my benchmarks, but has been as high as 160 with GBGromacs and as low as 90 PPD with regular Gromacs, so it's right in the middle.

As for my Opteron, it gets usually 100-150 PPD topping out at 190PPD with tinkers, so 150 PPD is good for it (per core - so anywhere from 200-400 PPD normally for both cores).

 
Imitation said:
Hmm I woulda thought the 3800 core at 2.5 would be doing better than 37:20. My Opty at 2.63 is doing them in 33 flat. Wonder if the cache is effecting it at all?
Click to expand...

I was doing some other things with my machine at the time (like reading this forum), so it wasn't a pure test. (btw, these new forum ads do seem to up the cpu usage for opera quite a bit). When I stop the non-folding activity I cut about a minute off of the times for both cores, so the lambda protein is coming in around 36 and change. That still sounds a tad slower than the frequency difference would explain I think, so maybe the cache does make a difference. I'm also not running my system memory at 1:1 in order to overclock it, but I've been told that wouldn't make much of a difference with AMDs.

meisterbrau
 
meisterbrau said:
I was doing some other things with my machine at the time (like reading this forum), so it wasn't a pure test. (btw, these new forum ads do seem to up the cpu usage for opera quite a bit). When I stop the non-folding activity I cut about a minute off of the times for both cores, so the lambda protein is coming in around 36 and change. That still sounds a tad slower than the frequency difference would explain I think, so maybe the cache does make a difference. I'm also not running my system memory at 1:1 in order to overclock it, but I've been told that wouldn't make much of a difference with AMDs.

meisterbrau
Click to expand...

Yeah I don't think running 1:1 makes that big a diff. My mem isn't at 1:1 but its still running well over stock. A divider will let you take that 3800+ as high as it will go. Unless you have it in a sucky mobo that can't do more than about 270-280ish fsb.
 
Hmm, I have a bigger work unit on my GF's pc (a folder I built her)

p2353_lambda_repressor

541 Points

25:54 per frame (with a few apps running)
12.553 Points per hour
300.792 Points per day

The PC is a A64 3400+ Newcastle @ 2.4ghz.

There is hope without the QMD's!
 
Imitation said:
Hmm I woulda thought the 3800 core at 2.5 would be doing better than 37:20. My Opty at 2.63 is doing them in 33 flat. Wonder if the cache is effecting it at all?
Click to expand...

the times I'm seeing now make more sense when compared to your Opty -

34:55 for the lambda and 24:39 for the 586 pointer. that's with only folding going on.

sorry for the mixup - next time I'll make sure to capture folding only times.

meisterbrau

 
sandmanx said:
http://www.em-dc.com/

Click on downloads, and grab the latest data file. Follow the instructions on the page.
Click to expand...

Actually It looks like it figured it out. It must have updated itself. Thanks though Sandmanx I'm sure I'll need to do that in the future...
 
Killer[MoB] said:
You may want to check on this box. That's a pretty good bit slower than it should be.
I'm getting 00:52:50 to 00:53:05 on a XP @ 1733.
Click to expand...

I saw that and took a look, it's my work box so it's constantly in use but still should be getting most of the cpu. I checked the cpu usage yesterday and it was getting approx. 99%, checked again today after updating Firefox to 1.5.1 and it's about 80% with Firefox taking up most of the rest. :( What the hell? Between the memory usage and now this I might have to switch to Opera.

 
pageian said:
I saw that and took a look, it's my work box so it's constantly in use but still should be getting most of the cpu. I checked the cpu usage yesterday and it was getting approx. 99%, checked again today after updating Firefox to 1.5.1 and it's about 80% with Firefox taking up most of the rest. :( What the hell? Between the memory usage and now this I might have to switch to Opera.

Click to expand...

Well if you're looking at any webpages that have flash ads on them it will take a fair amount of cpu time like you said. Minimizing doesn't really help I don't think.

 
Imitation said:
Well if you're looking at any webpages that have flash ads on them it will take a fair amount of cpu time like you said. Minimizing doesn't really help I don't think.

Click to expand...

Yeah, Firefox seems to use quite a bit more resources than some others though..... maybe it's not a good browser to use on a folder. Something to think about anyway.

 
OSUguy98 said:
what's your CPU?


Is the protien a "p2102_lambda_5way_melt_0_00000"? that's the only 414point protein I see on Stanford's Protien summary
Click to expand...

Thank you for that link. I never actively searched for a link to the point allocation, but I knew something was around.

 
pageian said:
Yeah, Firefox seems to use quite a bit more resources than some others though..... maybe it's not a good browser to use on a folder. Something to think about anyway.
Click to expand...

Firefox consistantly uses more CPU time than IE. I tested it a about a year ago by opening the same web page in IE and FF and then minimizing them to the task bar. Come back in a hour or so and pull up task manager and FF will have used more than double the time IE did.
 
On the topic of big workunits, one core of my Opty 165 is chugging on a 722 pointer, p2363_mAcP_mutant. I just finished a round of gaming (then bluescreened and rebooted!!) so I'm not sure how accurate the time-per-frame is, but EMIII is reporting 40:56. I'll repost/edit if it changes, but that would be ~254ppd.
 
You must log in or register to reply here.
Back
Top