[mordak]

I'm getting 1200-1500 ppd on an i5-750 overclocked to 3.5 GHz. What gives!?

I'm using SMP2, 100%, didn't complete 10 units yet so no bonus points...but still, this seems very low! How come?

I followed this guide exactly:
http://www.overclock.net/overclock-n...t-windows.html

Also 1 more question...it says set the checkpoint interval to 5 minutes. Why not 3, 15, or 30 min? Does this have any impact on ppd?

(Project P6023), 475 point project

 

[nckid4u]

Try this setup and run -smp 4 flag and see if it gets better. Almost sounds like you are not folding on all four cores. Can you confirm 100% usage on all four cores.

 

[mordak]

Yes in task manager folding is using 95-100%

Fahmon reports 1200 ppd...HFM.net is reporting 4800 ppd, so I guess that's better...

 

[Mikecdm]

What are using to monitor your ppd? If you are using fahmon, that could be the reason for low ppd. The smp2 projects give good ppd due to their bonus points. Fahmon doesn't calculate the bonus points.

You can either use the smp2 calculator, which is based on TPF, or switch over to HFM.net

 

[k4m1k4z3]

Quote:

Originally Posted by Mikecdm What are using to monitor your ppd? If you are using fahmon, that could be the reason for low ppd. The smp2 projects give good ppd due to their bonus points. Fahmon doesn't calculate the bonus points. You can either use the smp2 calculator, which is based on TPF, or switch over to HFM.net

Quote:

Originally Posted by mordak Yes in task manager folding is using 95-100% Fahmon reports 1200 ppd...HFM.net is reporting 4800 ppd, so I guess that's better...

So FahMon was showing the lower points because it doesn't calculate the bonus, but still I think you could be doing a little bit better than 4800 from what I have heard others say they get.

Do you have a lot of other things running that could be slowing it down?

Until you complete the 10 units, you will only be getting the points that FahMon reports you getting. After the 10, you can count on the points HMF shows you.

 

[mordak]

I read somewhere here that windows 7 smp2 is just as fast as vmware, so vmware is useless now.

True or not?

Also my question regarding checkpoint intervals, does it affect ppd if you set it to 3 min vs 30 min?

Finally, I'm getting 8k ppd now on a 2811 point Project P6023.

Guess 475 pointers just suck!

 

[ablearcher]

SMP2/A3 Windows is just as fast as linux. Running a VM now wastes memory and CPU time (guest OS overhead).

HOWEVER... the Windows client (when out of A3), reverts to A1 ewww...
The Linux client defaults to A2, which is much better.

Nothing you can do about it, except wait it out.

 

[zodac]

Switch to -smp 3, then let the SMP client update by 3%. Check your PPD now.

 

[mordak]

zodac, what does let SMP client update by 3% mean?

Also what does the -smp 3 flag do?

 

[F1ForFrags]

My i5 got about 12,000 PPD, but Fahmon always showed me as getting around 1700 PPD because it doesn't calculate the bonus points. But right now my i5 is pulling 0 PPD due to my dead motherboard and/or PSU.

 

[zodac]

Quote:

Originally Posted by mordak zodac, what does let SMP client update by 3% mean? Also what does the -smp 3 flag do?

HFM and Fahmon both monitor PPD based on the last 3 frames. So even if the issue is fixed, the monitor won't show that until you complete another 3% of the WU.

And the -smp 3 flag means you'll be using the SMP client on only 3 cores. You'd then need to set the affinity of the GPU2 client FahCore_11/FahCore_14 file (in Task Manager> Processes) to the free core.

That allows your GPU to fold on a free core and stop it taking resources away from the CPU.

 

[mordak]

Never said I was doing GPU

ATI 4890 sucks for that...
I'll keep the smp -3 tag off then.

Nobody knows the question about the checkpoint interval 3-30 min whether it affects ppd or not? Because 3 min seems best if it doesnt affect ppd

 

[zodac]

It doesn't affect PPD, so go right ahead with 3 mins if you want. I just thought 5 looked better than 3 (no, seriously, that's why).

 

[PCCstudent]

What parameters allowed a 10,000 ppd HFM report? My stock clocked i7 reports 10,000 ppd but only on certain a3's, typically it's much closer to 9000

 

[kup]

Quote:

Originally Posted by ablearcher SMP2/A3 Windows is just as fast as linux. Running a VM now wastes memory and CPU time (guest OS overhead). HOWEVER... the Windows client (when out of A3), reverts to A1 ewww... The Linux client defaults to A2, which is much better. Nothing you can do about it, except wait it out.

I have actually found my Linux -Advmethods to revert to A1 when there are no A3 as well as reverting to A2.

 

[ablearcher]

Quote:

Originally Posted by kup I have actually found my Linux -Advmethods to revert to A1 when there are no A3 as well as reverting to A2.

yeah, I have learned, now:

Linux A1, A2 (bigadv is a superset), A3
Windows A1, A3

At least Linux has a chance of A2 over A1

thank you!