• 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.

Those Crazy WU's

E

EvilAlchemist

2[H]4U
Joined
Jan 11, 2008
Messages
2,732
Over the last few days, been getting some weird WU's

My Quad cores are running projects:

3043 - 1440 Points
3052 - 1440 Points

My Dual Cores

3065 - 1837 Points

Usually my Linux boxes run 2605 and windows clients run 2653
But they have flipped this weekend and running the exact opposite.

I just the assignment servers have run out of the large point WU's ... for the moment

Anyone else had this happen?

 
EvilAlchemist said:
Over the last few days, been getting some weird WU's

My Quad cores are running projects:

3043 - 1440 Points
3052 - 1440 Points

My Dual Cores

3065 - 1837 Points

Anyone else had this happen?
Click to expand...

Which cores are they running FahCore_a1.exe or FahCore_a2.exe ?
Has your PpD altered ?

Still only getting 2605/2653 here.

Luck ............ :D
 
The quad core I have running started picking up 3052WU just recently.
Going back thru log file the previous WU was 3051
and right before that I noticed this:

Unit 1 finished with 74 percent of time to deadline remaining.
[05:13:33] Updated performance fraction: 0.736188

possiblity need to beat a certain performance fraction to get the larger 3052 wu?
 
My Q6600 box got a 3065 WU today and it is taking a lot longer per frame. It had been eating a 2605 a slightly less than 10 min per frame and now it is taking 12.5 min per frame. It's worth 1837 points. I haven't seen the FahCore_a2.exe yet and I have been giving the client 2 meg of ram to work with.

 
Tigerbiten said:
Which cores are they running FahCore_a1.exe or FahCore_a2.exe ?
Has your PpD altered ?

Still only getting 2605/2653 here.

Luck ............ :D
Click to expand...

All running FahCore_a1.exe
On the quads, PPD has not been affected.

ON the dual cores ... my PPD dropped by 40% on these new cores.
 
EvilAlchemist said:
All running FahCore_a1.exe
On the quads, PPD has not been affected.

ON the dual cores ... my PPD dropped by 40% on these new cores.
Click to expand...

:eek:

I wonder if they are running out of Gromacs? Or if they are redoing the point structure.
 
Kendrak said:
:eek:

I wonder if they are running out of Gromacs? Or if they are redoing the point structure.
Click to expand...

Or, as many times happens the project is done and they have started another. No one project goes on forever.;)
 
Still running 2605s on my quad and still running 2653s on my X2, so nothing new yet :D

 
There are many more work units than the 1760 pointers. Most of the rest of the SMP work units don't garner as many PPD. The difference seems to be even worse if you have a CPU with a smaller amount of cache. Generally speaking, anything with less than 4 meg of cache doesn't put out the same points as one that does.

I know that my X2 takes a huge hit to points if it's not running the 1760 pointers. It does around 1350 PPD with the 1760 pointers. Any of the other work units and it's closer to 800-900. My old E6400 took a similar hit and it only had 2 meg of cache. I haven't had anything but the 1760 pointers since I got my quads so I'm not sure how bad the hit is on them. From what I've heard, it's not as bad due to the 4 meg of cache for each two cores.

 
As of 9:30 PST, there were only 260 units remaining on the server that dishes out the P2653, P2651 and P2610 units. By tomorrow, everone will probably be folding something new.
 
Guess we will just have to wait and see what happens .....

I don't like the points dropping .... but if it affects everyone .... there will be no real difference ..... maybe the core will help speed the new WU up ...

Tomorrow we will see ....
 
Killer[MoB];1032115255 said:
As of 9:30 PST, there were only 260 units remaining on the server that dishes out the P2653, P2651 and P2610 units. By tomorrow, everone will probably be folding something new.
Click to expand...
Has there been any mention of new SMP WUs by Standford? If new WUs take longer to complete and have the same deadline it might affect people running borderline systems.

 
There seems to have been some projects added with lower point values

Here is a list of the Active SMP WU's

Project---------Points
3060 -----------2539
3065 -----------1837
2604 -----------1760
2605 -----------1760
2651 -----------1760
2653 -----------1760
1487 -----------1741
2610 -----------1523
3043 -----------1440
3050 -----------1440
3051 -----------1440
3052 -----------1440
3061 -----------1329
3064 -----------1329
3062 -----------1324
3059 -----------1167

It may just be Stanford has to download the completed WU's and upload the new ones.
I don't think the good PPD WU are gone ... lets just give them some time and keep an eye on the server status!

Also, Only 5 on this list have project descriptions. Most say "This Folding@Home project is new!"
I don't think we could have completed a project before Stanford had time to post about it.
 
The 2653's have had the "this project is new" message for as long as I can remember. I dontknow that they are updating that too much any more.
 
A couple of my clients at 4am CST grabbed new WUs and they were 2605s for 1760 points so I'm good to go on those clients for sure until later tonight.

 
Me too, 2605 and 2653 is what I'm always working on.
 
I'm running 5 2605s at the moment. This works for me.

 
No change here.
17x p2605's & 1x p2653.
Still chugging along at ~28,500 PpD.

Luck ............. :D
 
I am running:

[00:48:25] Project: 2653 (Run 21, Clone 107, Gen 60)

I picked it up Friday sometime. I went out of town Friday evening and my SMP VERY quickly decided it wanted the weekend off. Lazy bugger.


Am I going to make deadline?? Where do I look such a thing up? In the log?

 
Wheresatom said:
I am running:

[00:48:25] Project: 2653 (Run 21, Clone 107, Gen 60)

I picked it up Friday sometime. I went out of town Friday evening and my SMP VERY quickly decided it wanted the weekend off. Lazy bugger.

Am I going to make deadline?? Where do I look such a thing up? In the log?
Click to expand...
That all depends how much was completed and your system specs. You probably have about only a half day left or less, so it seems doubtful unless it's close to completion.

 
I am at 84 percent right now. It's taking about 28 minutes per percent.

I should be done soon as long as my internet doesn't go screwy and mess up the SMP.
 
I am working on two P3064 WUs and one P3062 WU. My systems DL them earlier today. These WUs are worth considerably less than the P2653, being worth only 1329 and 1324 points respectively. More importantly, they run a lot slower than the older WUs, between 15-20% slower from the looks of things. I don't understand Stanford's point allotment scheme. The new WUs take longer to process and they are worth less. :confused:

Anyone else get these?
 
Killer[MoB];1032130675 said:
3065 here....it sucks by comparison.
Click to expand...
Yeah, I just can't wait for my older systems to get some of these new units. :rolleyes:
 
I just got a 3052 and it killed my PPD, It's running 870 PPD on my main rig, see sig for normal opp. :mad:

 
Stanford only benches at 200 PpD per Ghz.
1760 PpD running at 2.2 Ghz on 4 cores.

So if your running faster than that its a bonus, not a right.

Luck ............... :D
 
See .. you all thought I was crazy when i started this thread .....

ON the new work units worth 1440 .. my quads are tearing them up ... it is great

The dual cores struggle on those but luckly, i am still getting the project 2605 on those

Some of you running dual VMware on quad might have to stop if you want your ppd to stay up
 
EvilAlchemist said:
See .. you all thought I was crazy when i started this thread .....

ON the new work units worth 1440 .. my quads are tearing them up ... it is great

The dual cores struggle on those but luckly, i am still getting the project 2605 on those

Some of you running dual VMware on quad might have to stop if you want your ppd to stay up
Click to expand...

But why did my poor X2 get the new 3052 worth 1400?!

 
Tigerbiten said:
Stanford only benches at 200 PpD per Ghz.
1760 PpD running at 2.2 Ghz on 4 cores.

So if your running faster than that its a bonus, not a right.
Click to expand...
Tiger,

The only thing I'm wondering about is Stanford's point allocation for different WUs, not their methodology. They're inconsistent, not only across clients but across projects for the same client. I have been saying this since 2001. Nothing has changed in regards to the points scheme as far as consistency is concerned.

Actually, I wouldn't care if I was forced to retire my old systems. Truth be told, I've been looking for a good reason to do it because they aren't all that productive to begin with. I run them because they still manage to make the SMP deadline in about 3 days, so why not have them do something useful?

That being stated, I can't understand why slower processing WUs are worth less than quicker WUs that have been longer established. My dual quad is down 1500PPD because of these longer-running, lower valued WUs. I really don't comprehend the reasons behind the evaluation system, that's all.
 
APOLLO said:
See what I'm saying? The distribution seems random but the point evaluation is simply illogical.
Click to expand...

From Stanford's SMP FAQ ...............
How do you decide the credit value of SMP work units?

Points are determined by the performance of a given machine relative to a benchmark machine, similar to the CPU client benchmark process. Before releasing any new project (series of work units), we benchmark it on a dedicated Macintosh Pro with 2 - 2.33 GHz Dual Core Xeon processors. (more specifically, 2 Woodcrest 5140 processors with 4 MB cache (each), 5 GB FBDIMM Memory (667 MHz DDR2), 1.33 GHz Bus)

We plug the results of this benchmark test into the following formula:

points = 1760 * (daysPerWU)

where daysPerWU is the number of days it took to complete the work unit.

Please note the very concept of a reference machine will mean that some WU performance will vary from the performance on your machine. Even between various Xeon processors, there are significant differences in architectures. Moreover, there are variations between WUs within a given project which can lead to speed differences.

Our goal is consistency within a given definition of a reference machine setup (described above), but beyond that, the natural variation from machine to machine and WU to WU will never allow any point system to perfectly predict what you get on your machine.
Click to expand...

The major difference I see between work units is how well they scale from 4 cores down to 2 cores.
The slow protien that are getting more common don't scale that well, hence getting 220 PpDpGhz on a 4 core vs 250 PpDpGhz on 2.
The quick protiens scale a lot better, 250 PpDpGhz on 4 vs 330 PpDpGhz on 2.
Thats where the major variation in PpD happens.
But again anything over 200 PpDpGhz is a bonus.

Luck ................. :D
 
I've been getting some 2653s lately that aren't folding as well with my dual-SMP setup. My overall PPD drops by about 500 when I get these units in the mix. Also, I've noticed that my 3.0 GHz Q6600 struggles with 2605 units. I think they like the memory bandwidth on my 3.5 GHz rig. When I get the slow 2653s and 2605s together, it hurts.


 
You must log in or register to reply here.
Back
Top