Project 3060 - WUs worth 2539 each!

[scottv67]

Has anyone else put a hook into these whopper work units for project 3060? I just noticed that one of my Q6600s is working on a WU that is worth 2539 points. Wow!!! I'll watch the elapsed time on that baby to see if this WU takes any longer than the 1440 point WUs that seem to be the main part of my Folding diet.

I wouldn't mind finding a "school" of these 2539 point WUs. 😀

I think we're gonna need a bigger boat...😉

-s :beer:

 

[Markfw]

3050 and 3043, but they are only worth 1440, I hate them The take almost 24 hours on a 2800 quad running ubuntu, 1500 ppd. I get 2200 out of a 2653.Both of those numbers are running 2 smp clients on ubuntu.

 

[Insidious]

I've got a couple 2653s ready to finish....

/me crosses fingers

C'mon 3060 😀

 

[Markfw]

well, with 11 machines going, and 15 units (4 have 2 smp units) I have so many going all the time, no 3060 on any ,,,,

 

[Uppsala9496]

Only 2653's for me so far. I'm curious to see how much longer the 3060's take, so please kep us posted.

 

[Insidious]

all @ 2653 here

 

[natethegreat]

Originally posted by: Insidious
all @ 2653 here

Ditto

 

[nefariouscaine]

why not give SoB a try - the WU's are far more understandable and you get a pretty little 17 in your icon bar 😛

just trying to do some shameless promotion 😉

 

[scottv67]

Uppsala9496 wrote:

Only 2653's for me so far. I'm curious to see how much longer the 3060's take, so please kep us posted.

This is a rough estimate: The Q6600 that is currently crunching the 3060 WU can turn-around a WU from project 3043 (1440 points) in under 12 hours. This is running Fedora 7 and one copy of the F@H SMP client for Linux. Watching FahMon today, it looked like the 3060 WU was going to take 17 hours to complete when running one copy of the F@H SMP client for Linux. Don't use these numbers to compute PPH for the respective work units, these are just rough numbers.

One thing that I am fairly sure of is that running one Linux SMP client on that box (crunching the 3060 WU) was giving around 3500 PPD. If I run two Linux SMP clients on that box (crunching 1440-pt WUs from Project 3043), I was getting 3600 PPD. Right now, FahMon shows that box is running at 3780 PPD (it is crunching one 1440 point WU and one 2539 point WU).

The Project 3060 WU is "heavier" (system resource and wall clock-wise) than the Project 3043 WUs and it looks like the reward for the 3060 WUs is proportional to the increased "weight".

This is on Fedora 7 on a Q6600. Your Mileage May Vary.

I hope this helps. If I come up with something more scientific, I'll post it.

-s :beer:

 

[Insidious]

Thanks Scott!

I'll be curious to see if a windows box gets those. Sounds like the scoring in terms of PPD is pretty much in line with the other WUs... well except maybe the boner 2652 ones.

Cheers :beer:

-Sid

 

[CupCak3]

Originally posted by: Insidious
Thanks Scott!

I'll be curious to see if a windows box gets those. Sounds like the scoring in terms of PPD is pretty much in line with the other WUs... well except maybe the boner 2652 ones.

Cheers :beer:

-Sid

I will also be interested to see this.

I wish I could go back to linux on my c2d to squeeze out the extra PPD... oh wells. Maybe in a couple months I'll finally get a quad or two (thought most likely I'll be waiting until something good in the 45nm variety gets cheap)

 

[xpblackdragon]

starting mdrun 'p3060_BBA5_1'
20000000 steps, 40000.0 ps.

[16:46:35] Protein: p3060_BBA5_1
[16:46:35] Writing local files
[16:46:35] Completed 13547770 out of 20000000 steps (67 percent)
[16:46:35] Extra SSE boost OK.
[16:46:35] ompleted 13547770 out of 20000000 steps (67 percent)
[16:46:35] Extra SSE boost OK.
[16:52:32] Writing local files
[16:52:32] Completed 13600000 out of 20000000 steps (68 percent)
[17:08:32] Timered checkpoint triggered.
[17:14:51] Writing local files
[17:14:51] Completed 13800000 out of 20000000 steps (69 percent)
[17:30:51] Timered checkpoint triggered.
[17:37:04] Writing local files
they take a long time to complete
Q6600@3500
running 2 SMP clients on Ubuntu

 

[GLeeM]

Originally posted by: xpblackdragon
starting mdrun 'p3060_BBA5_1'
20000000 steps, 40000.0 ps.

[16:46:35] Protein: p3060_BBA5_1
[16:46:35] Writing local files
[16:46:35] Completed 13547770 out of 20000000 steps (67 percent)
[16:46:35] Extra SSE boost OK.
[16:46:35] ompleted 13547770 out of 20000000 steps (67 percent)
[16:46:35] Extra SSE boost OK.
[16:52:32] Writing local files
[16:52:32] Completed 13600000 out of 20000000 steps (68 percent)
[17:08:32] Timered checkpoint triggered.
[17:14:51] Writing local files
[17:14:51] Completed 13800000 out of 20000000 steps (69 percent)
[17:30:51] Timered checkpoint triggered.
[17:37:04] Writing local files
they take a long time to complete
Q6600@3500
running 2 SMP clients on Ubuntu

Looks like about 1662 ppd for this client.

What's the other one doing?

 

[Markfw]

That not good... I get 2400 each on my 3420 mhz quad and 2500 each on my 3610 quad, for 4800 and 5000 ppd total. That would only be 3300...I will take the 2653's..

 

[biodoc]

Originally posted by: xpblackdragon
starting mdrun 'p3060_BBA5_1'
20000000 steps, 40000.0 ps.

[16:46:35] Protein: p3060_BBA5_1
[16:46:35] Writing local files
[16:46:35] Completed 13547770 out of 20000000 steps (67 percent)
[16:46:35] Extra SSE boost OK.
[16:46:35] ompleted 13547770 out of 20000000 steps (67 percent)
[16:46:35] Extra SSE boost OK.
[16:52:32] Writing local files
[16:52:32] Completed 13600000 out of 20000000 steps (68 percent)
[17:08:32] Timered checkpoint triggered.
[17:14:51] Writing local files
[17:14:51] Completed 13800000 out of 20000000 steps (69 percent)
[17:30:51] Timered checkpoint triggered.
[17:37:04] Writing local files
they take a long time to complete
Q6600@3500
running 2 SMP clients on Ubuntu

What's the final word on these new WUs?

Scott & XP?

I haven't seen one yet.

 

[Insidious]

I haven't seen one. I just keep getting 2653 with a very rare 2652 tossed in (and then out as soon as I see it) of the mix.

-Sid