I see computers with the same or lesser specs than the ones I have processing average units in much greater quantities than my stats show. What is the magic in achieving this I wonder? For exampleI have one unit with the lastest 980 extreme chip and 2 HD5800 series gpu, and see another user with the same chip and 1 gtx470 gpu. My computer shows an average of 2065 works units since the system came back up, and his shows over 8600. I see others with computers that have lesser specs than ones I have outperforming my systems. I have not done anything other than download and run BOINC on each system. I have set the preferences to let it use 100% of the processor usage.

---

I see computers with the same or lesser specs than the ones I have processing average units in much greater quantities than my stats show. What is the magic in achieving this I wonder? For exampleI have one unit with the lastest 980 extreme chip and 2 HD5800 series gpu, and see another user with the same chip and 1 gtx470 gpu. My computer shows an average of 2065 works units since the system came back up, and his shows over 8600. I see others with computers that have lesser specs than ones I have outperforming my systems. I have not done anything other than download and run BOINC on each system. I have set the preferences to let it use 100% of the processor usage.

OCing and/or optimized apps are probably the difference.

---

"Freedom is just Chaos, with better lighting." Alan Dean Foster

I see computers with the same or lesser specs than the ones I have processing average units in much greater quantities than my stats show. What is the magic in achieving this I wonder? For exampleI have one unit with the lastest 980 extreme chip and 2 HD5800 series gpu, and see another user with the same chip and 1 gtx470 gpu. My computer shows an average of 2065 works units since the system came back up, and his shows over 8600. I see others with computers that have lesser specs than ones I have outperforming my systems. I have not done anything other than download and run BOINC on each system. I have set the preferences to let it use 100% of the processor usage.

Plus the Fermi cards can process 2-3 units at the same time. I don't know how far up I can go with my budget cruncher

CPU type	        AuthenticAMD AMD Athlon(tm) II X4 630 Processor [Family 16 Model 5 Stepping 2]
Number of processors	4
Coprocessors	        NVIDIA GeForce GTX 460 (767MB) driver: 26063

---

Optimized apps, overclocking, and fermi are the main differences.

---

Traveling through space at ~67,000mph!

Yes replace all your AMD/ATI gpu's with CUDA and just make sure you let them crunch 24/7.

Yes replace all your AMD/ATI gpu's with CUDA and just make sure you let them crunch 24/7.

I would not go that far, the 5800 series cards will do good at Milkyway and DnetC and the 5700 series cards will do good at Collatz and Dnetc.

---

I just had a bit of an experiment.. increased % of processors to use from 90 to 100%.. and the GPU time TRIPLED. Taking it back to 90% increased it back.

So balancing out your system can really pay off. But you gotta play with it to find your sweet spots.

---

Janice

I just had a bit of an experiment.. increased % of processors to use from 90 to 100%.. and the GPU time TRIPLED. Taking it back to 90% increased it back.

So balancing out your system can really pay off. But you gotta play with it to find your sweet spots.

Soft^spirit, You raised a good point, so I checked my preff page. I take it you mean setting the use at most % of CPU time to 100?

I have this i7 set at 100% and my times for GPU are about 23 to 30 minutes per WU.

Now my suspend work if CPU is above 95% is what I have set. Not sure what setting you tweaked to find your sweet spot.

---

[/quote]

Old James

Still tweaking James... I will try to keep you posted what my sweet spot ends up being. of course your mileage may vary.

at 100/100, gpu units were shooting up to 2hrs +, well above my average cpu core speed. that made me go AAAAAGGGGHH!!!

I was running 3 cpu cores, I now have it at all 4 (90% vs 100% of processors)
But did take the "use at most" back down to 90%. That helped. More dialing in to follow. I am also wondering if the opp apps .04%of cpu+gpu is adequate..

But right now.. it seems my gpu is taking around 50 minutes for a typical work unit. That still seems high... Lots to figure out and it will take me some time.

---

Janice

Ok thanks soft^spirit. I did some tweaking back when i fist got the i7 But dropped back to the 95 after seeing whacky stuff happening on screen. I think I will just leave mine as is.

---

[/quote]

Old James

Still tweaking James... I will try to keep you posted what my sweet spot ends up being. of course your mileage may vary.

at 100/100, gpu units were shooting up to 2hrs +, well above my average cpu core speed. that made me go AAAAAGGGGHH!!!

I was running 3 cpu cores, I now have it at all 4 (90% vs 100% of processors)
But did take the "use at most" back down to 90%. That helped. More dialing in to follow. I am also wondering if the opp apps .04%of cpu+gpu is adequate..

But right now.. it seems my gpu is taking around 50 minutes for a typical work unit. That still seems high... Lots to figure out and it will take me some time.

The .04% is just fine. It really doesn't take much. Use me as an example.

Steve

---

Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website

Still tweaking James... I will try to keep you posted what my sweet spot ends up being. of course your mileage may vary.

at 100/100, gpu units were shooting up to 2hrs +, well above my average cpu core speed. that made me go AAAAAGGGGHH!!!

I was running 3 cpu cores, I now have it at all 4 (90% vs 100% of processors)
But did take the "use at most" back down to 90%. That helped. More dialing in to follow. I am also wondering if the opp apps .04%of cpu+gpu is adequate..

But right now.. it seems my gpu is taking around 50 minutes for a typical work unit. That still seems high... Lots to figure out and it will take me some time.

The .04% is just fine. It really doesn't take much. Use me as an example.

And the figure in app_info doesn't actually change any processor behaviour - it's just an accounting device to help maintain resource share. The only time it makes a difference is if you have multiple cards, and set a high percentage - if you have say three GPUs, and declare that each of them requires 40% of a CPU, then the total is more than a core, and accordingly BOINC will stop scheduling a core to leave space.