I didn't want to bring the whole "my win 7 system only sees 3/4 GPU" thing into this thread, it seems to be a non-fixable problem unless you're willing to go to win xp 64bit.

Look this:

I'm a little confused. WIN 7 recognizes ALL cores
http://setiathome.berkeley.edu/forum_thread.php?id=59371

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

I will work on the power supply thing. I have a coolmax 1000w
http://www.newegg.com/Product/Product.aspx?Item=N82E16817159116
It "should" be able to handle 2x gtx295. Also, maybe it isn't the extra videocard. I've been getting errors ever since I reverted the drivers back to 191.07 I wonder if it would do better with the newer drivers? Before I was running 700mhz 24/7 with zero errors on the one card. I can't even run stock speeds with these drivers. But... still crunching mega numbers with the reverted drivers. I'm tempted to bump it back up now that I have the rescheduler. Maybe that really was the slow-problem. I was trying to crunch the slower WU on my GPU.

Also, my motherboard doesn't support quad-sli, which I assume is what is needed to utilize all 4 cores. With SLI disabled BOINC only sees one of the two on my single card. With SLI enabled it sees both. With a second card added in I cannot go Quad-sli because it's not supported. Perhaps when I have some cash I can get a new mobo and try that out. I've been looking at the MSI BIG BANG FUZION with the hydra chip. I'm not sure if the Hydra would be "Boinc" friendly.

---

What are the temperatures of the 2 (or 4) GPUs?

Check with GPU-Z
http://www.techpowerup.com/gpuz/

or EVEREST
http://www.lavalys.com/


Do you have random crashes of non-graphic programs?
Check all the fans in the computer.
Do you overclock CPU and RAM (timings)?

Look:
Temperature Monitoring Programs
http://setiathome.berkeley.edu/forum_thread.php?id=59292

.

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Umm, my main card is watercooled, ~45-50* under load. The 2nd card when added in there hits 70* with BOINC, and 85-90* with vantage.

I don't get random crashes at all. I do get the video driver crashing only when BOINC is running and I'm doing other stuff like surfing the web also. Or if I have BOINC and the 3 GPU's working at the same time. Again, I never had this problem before reverting the drivers to 191.07 I could re-update them, but I'm afraid of losing my current crunch speed.

I do overclock, got my i7 860 @ 3.8ghz, stable with prime 95 for 12 hours or so. Can't quite get it stable at 4ghz. I have my memory actually underclocked, its 2133mhz memory running at like 1440 or so.

---

Umm, my main card is watercooled, ~45-50* under load. The 2nd card when added in there hits 70* with BOINC, and 85-90* with vantage.

I don't get random crashes at all. I do get the video driver crashing only when BOINC is running and I'm doing other stuff like surfing the web also. Or if I have BOINC and the 3 GPU's working at the same time. Again, I never had this problem before reverting the drivers to 191.07 I could re-update them, but I'm afraid of losing my current crunch speed.

I do overclock, got my i7 860 @ 3.8ghz, stable with prime 95 for 12 hours or so. Can't quite get it stable at 4ghz. I have my memory actually underclocked, its 2133mhz memory running at like 1440 or so.

Here's one out of left field, then. Try dropping back the overclock to (say) 3.5ghz just to see if it has any effect. Mine is a totally different environment but I did find, several months ago when I was having GPU problems, that skimming 10% off the CPU overclock cured them. And trying is this is much less effort than most other things available to you.

F.

---

I'll give that a shot. Probably not tonight, but in the next few days. Then I'll report back.

---

So, I backed it down to 3.5ghz and enabled hyperthreading. Creates more heat.

I upgraded drivers and get the nvidia driver failure less. I actually only get the error when I'm doing stuff with google chrome while crunching. If I'm not doing background stuff no errors. Maybe it's a software thing messing it up? Or overloading the GPU's? I dunno. Either way I'm not worried about it. I had to add an A/C unit in my room because it was getting so hot, lol!

---

So, I backed it down to 3.5ghz and enabled hyperthreading. Creates more heat.

I upgraded drivers and get the nvidia driver failure less. I actually only get the error when I'm doing stuff with google chrome while crunching. If I'm not doing background stuff no errors. Maybe it's a software thing messing it up? Or overloading the GPU's? I dunno. Either way I'm not worried about it. I had to add an A/C unit in my room because it was getting so hot, lol!

Only Google Chrome has this effect?
What about Opera, Firefox, IE, Safari?

Maybe it's Adobe Flash Player (YouTube videos, Flash games) involved?

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Yea, it might be adobe flash because it only happens when it's a very busy webpage. I'm not too worried about it.


Just figured out how to keep my GPU clocks up! woohoo, almost doubled my productivity! a constant 400mhz core clock up to a constant 612 is a good jump! Almost like adding another card in there. DIfference between Nvidia's 2d clockrate and 3d clock rate

---

Just thought I would add this into the mix. I was running a 2nd gtx295, only using one of the two extra GPU's, as BOINC wouldn't recognize the fourth.

I updated to the new 25x.xx drivers and BOOM, 4 cuda cores working in windows 7, with SLI off.

*note* I turned on quad-sli and got the blue screen of death. Had to reboot in safe mode to turn SLI off, and it's crunching numbers now on 4 cuda cores. Too soon to see if I get a bump in performance with WU/day

---