Announcement: SMP2 Cores

I switched to the WInSMP client last week from VMs on my main rig and haven't had any GPU slow downs since. If this will boost CPU output and keep the GPUs running at full steam that would be great.
 
The_Heretic said:
I switched to the WInSMP client last week from VMs on my main rig and haven't had any GPU slow downs since. If this will boost CPU output and keep the GPUs running at full steam that would be great.
Click to expand...

Yea im getting full speed on my GPUs but only if i use a process manager.
 
Excellent. I will switch over my Phenom II to the new WinSMP client when I get home tonight. Everyone remember to use the -oneunit flag so your old client finishes and turns in it's WUs. The ones that never come home can be pretty hard on the science output from what I've been reading.
 
[H]ugh_Freak;1035236369 said:
Ok so if you don't have a passkey setup you're gonna get screwed for points?
Click to expand...
If you don't set up and use a passkey, you will get the base credit for every SMP2 workunit you submit, but you will not receive any bonus points.
 
Some numbers from my 5530's with HT

Project ID: 6015
Core: GRO-A3
Credit: 484
Frames: 100


Name: CC SMP
Path: C:\Folding\SMP0
Number of Frames Observed: 100

Min. Time / Frame : 00:03:20 - 15,426.1 PPD
Avg. Time / Frame : 00:04:02 - 11,589.9 PPD
Click to expand...
Project ID: 6013
Core: GRO-A3
Credit: 380
Frames: 100


Name: CC SMP
Path: C:\Folding\SMP0
Number of Frames Observed: 200

Min. Time / Frame : 00:02:38 - 15,427.6 PPD
Avg. Time / Frame : 00:03:05 - 12,176.6 PPD
Click to expand...
Project ID: 6012
Core: GRO-A3
Credit: 470
Frames: 100


Name: CC SMP
Path: C:\Folding\SMP0
Number of Frames Observed: 115

Min. Time / Frame : 00:03:17 - 14,954.1 PPD
Avg. Time / Frame : 00:03:43 - 12,416.6 PPD
Cur. Time / Frame : 00:03:26 - 13,002.9 PPD
R3F. Time / Frame : 00:03:40 - 12,175.4 PPD
All Time / Frame : 00:03:32 - 12,634.9 PPD
Eff. Time / Frame : 00:03:58 - 11,254.6 PPD
Click to expand...
 
Nitro, is that a dedicated system? If it is your main desktop, the PPD variations are normal.

As the A3 core continues to develop, performance will get even better.

It is very nice the roll out has gone so smooth. No Users reporting EUEs and very few client issues.
 
EvilAlchemist said:
Nitro, is that a dedicated system? If it is your main desktop, the PPD variations are normal.

As the A3 core continues to develop, performance will get even better.

It is very nice the roll out has gone so smooth. No Users reporting EUEs and very few client issues.
Click to expand...

Its my main desktop.

I dont have any truly dedicated folders.
 
Well I'll give this a shot on my q6600. It's not clocked high enough to try bigadv, so do I need a passkey? I might do it anyway just so that when I finally get my i7 I can jump right in.
 
Yes you need a passkey if you want to receive the bonus
 
So, it looks like people are seeing around 6,000 to 8,000 ppd from Core2Quads judging by a few posts here. Does that seem reasonable?
 
I'm seeing 1700ppd on my q6600 at 3.8ghz w/ 80% usage. If I let it go 100% system is completely useless. Also, even at only 80%, it tanks my first gpu client down about 2200ppd, so the cpu ppd isn't worth it. Running vista 64 here. Still not worth it for me, but that's probably still due to it having pulled a stupid a1 unit. Quitting cpu folding again till this is fully deployed and only a3 units are out there.
 
jebo_4jc said:
So, it looks like people are seeing around 6,000 to 8,000 ppd from Core2Quads judging by a few posts here. Does that seem reasonable?
Click to expand...

I was getting about 9,000 on the P6014 WU but yeah. It's a pretty decent improvement from my VM which got about 5,500-6,000.
Posted via [H] Mobile Device
 
okay i am running the new beta here, Q6600 at 3.4, 8800GT stock, why are my SMP points so low, much lower than if i run notfreds VM.

SMPbetatime.jpg
 
the snake said:
okay i am running the new beta here, Q6600 at 3.4, 8800GT stock, why are my SMP points so low, much lower than if i run notfreds VM.
Click to expand...
Because you are using FAHMon. FAHMon does not know how to calculate the bonus points. You have to use something like HFM.net or FAHSpy to see your true numbers. Also, you had to have done 10 SMP A2 or A3 units with a passkey to qualify for the bonus.
 
arnemetis said:
I'm seeing 1700ppd on my q6600 at 3.8ghz w/ 80% usage. If I let it go 100% system is completely useless.
Click to expand...
Since you are still running A1, did you upgrade the client to 6.29?

On the "useless" comment, This is a known issue with 2.14 A3.
These is a fix in the works that will pay nicer with the windows scheduler.
That will makes the system feel more "responsive" when F@H is at 100% on the cpu cores.
 
Tobit said:
Because you are using FAHMon. FAHMon does not know how to calculate the bonus points. You have to use something like HFM.net or FAHSpy to see your true numbers. Also, you had to have done 10 SMP A2 or A3 units with a passkey to qualify for the bonus.
Click to expand...

got it now, i added a passkey and use HFM.NET and all is good, about 5500- 6000PPD on the CPU now, thanks :)
 
EvilAlchemist said:
On the "useless" comment, This is a known issue with 2.14 A3.
Click to expand...
Ahh, I didn't realize Stanford rolled this public beta out with v2.14 core. This would explain some of the oddities many here are seeing with priority management on non-dedicated machines.
 
So can we expect a more civil a3 core sometime soon? 2.14 is not playing well with others on my and several other's mixed GPU and SMP systems.
 
flogge said:
So can we expect a more civil a3 core sometime soon? 2.14 is not playing well with others on my and several other's mixed GPU and SMP systems.
Click to expand...
I've asked Dr. Kasson about this. Not much I can say at this point without violating my NDA as a private beta tester. I will let you know what I hear from him.
 
I've asked Dr. Kasson about this. Not much I can say at this point without violating my NDA as a private beta tester. I will let you know what I hear from him.
Click to expand...

No problem. That tells me that the problem/quirk is known and that there are better cores in the works, the only question is when will they be released. Thanks.
 
flogge said:
....the only question is when will they be released.
Click to expand...

Sometimes, testing does not go well < --> Sometimes it does. SMP2 was easy and painless 98% of the time.There are other times, we test <--> EUE <--> new version <--> repeat.That can go on for weeks and months sometimes.

I did not understand the idea of never giving a time frame before I started Testing for F@H. Now, i totally get it. Getting a piece of software to run on one system is easy.
Getting it to run on 10,000's systems, all with different Operation systems and conditions is another.

We test, and test and retest, so you never have to deal with what we do! :rolleyes:
 
You must log in or register to reply here.
Back
Top