Last night on another thread,

http://setiathome.berkeley.edu/forum_thread.php?id=59804

VW Bobier suggested I turn my GTX 295 fan up to 100%. I had never thought to do that as with the three fans in the side panel of my case, the GTX 295 temps were running at about 60°C. Even while testing with Furmark, I was only getting a maximum temp of 70°C.

I kicked it up to 100%, and within a few minutes the GTX 295 temps had settled ariound 51°C. I was happy about this, as it gave me even more headroom. The strange thing is that my CPU temp went from 58°C to 46°C, and my water coolant temp went from 15°C to 11°C.

I thought it was a fluke, so I rebooted, and let it set over night. The same thing is true this morning. Those of you that have been following my posts will know that my GTX 295 has been throwing -1 errors back to the SETI servers. So far today there aren't any errors. My RAC is actually a bit lower than it was last night, but I'm not worried about that.

What I don't understand is why would the CPU temp reduce so much when all I did was increase the fan speed on my GTX 295. The CPU doesn't even have a fan, just water. I am very happy with the result, but I don't understand why it happened.

My CPU is an i7 980 running at 4.2 GHz, and to be running full out at only 46°C means I have a huge amount of headroom to clock it faster. My GTX 295 did blank out the screen and recover once while I was typing this, but it doesn't seem to produce errrors. I will just sit tight for a while to see where things settle out, but if anyone knows why this happened I would love to know.

Steve

---

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

Steve,
I see that happens on my 295´s rigs, and they are on air, just because less heat is inside the case.
My guess is as your 295 is on the same water system as you cpu, and if you leave the 295 fan to auto, it will do a small job, because 295 sensors will be happy with temperatures around 70°C (thanks to you water system, an easy thing), thus spinning really low.
When you manually set the 295 fan to 100%, it will blow off your system much more heat, thus leaving a much more easy work to your water system to cool down your cpu much more effectively.
And it seems the deffective 295s will apreciatte low low temperatures, as im learning now with my own and the tips for you guys on the other topic :)

---

It even gets better! I was just informed by Danger Den that the water blocks for the GTX 480's were in stock, so I ordered two with a copper base and nickle plating on the top. Things are coming together. My second radiator that holds four 140 mm fans has shipped. SabrePC also just informed me that my second GTX 480 has shipped! I will have my dedicated 30 AMP line and 2200 VA UPS system installed my tomorrow evening, so that with the 1250 watt PSU, and I will be ready to begin my dash to the finish line for my mega cruncher build! Happy Happy Joy Joy! The only things I still need to do is order a second water pump, fans, and 140 mm fan grills to keep the cats from sticking their paws into the fans, and I am set to boogie!

Steve

---

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

Well, the errors are back and my CPU temps returned to normal. I have no idea what actually happened, but it was certainly unusual. That an air cooled video card could affect a water cooled processor was very strange to me. At least the DTX 480's are on the way, along with the hardware to water cool them.

Steve

---

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

Well, the errors are back and my CPU temps returned to normal. I have no idea what actually happened, but it was certainly unusual. That an air cooled video card could affect a water cooled processor was very strange to me. At least the DTX 480's are on the way, along with the hardware to water cool them.

Steve

Did Yer fan speed go down? As that's why the Riva tuner is/was set to full even on a reboot, That way It over rides the stock setting for fan speed, Otherwise the stock fan speed will come back on a reboot of the PC.

Oh and I'm glad I could be of help. :D

Fan speed is still at maximum. I am using the EVGA Precision tool, and I made sure it was still at max. This puzzles me. I was very pleased with the results, but I waited until this morning to be sure before posting. Now, with everything back to normal I am even more confused. The GTX 295 is still running at 51 deg. C.

Steve

---

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

Yes... tricky pieces of hardware that 295 things.... some are very hard to understand.... I can only compare them with some girls.... guess they are all broken, in the end.....

---

Yes... tricky pieces of hardware that 295 things.... some are very hard to understand.... I can only compare them with some girls.... guess they are all broken, in the end.....

I had to share that with my wife, and we both laughed.

Steve

---

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

Yes... tricky pieces of hardware that 295 things.... some are very hard to understand.... I can only compare them with some girls.... guess they are all broken, in the end.....

Dunno....I have had no trouble at all with either of my EVGA GTX 295's....but I am also only running 1 card per rig.

I had to back off the RAM speed a little bit on the Frozen Nehi's 295, as the weather has been rather warm lately and GPU 2 kept falling back to a slower clock. But cutting the RAM speed a little bit seems to have stabilized it again.

The i7 920 rig's 295 actually clocks a little faster than the other, and it has not missed a beat yet.

---

"Time is simply the mechanism that keeps everything from happening all at once."

And for what it's worth...
My GTX 295 overclocks are as follows.

Nehi 295...684/1512/1026
i7 920 295...702/1512/1188

Core/shader/RAM.

---

"Time is simply the mechanism that keeps everything from happening all at once."

Boy I wish I had no troubles with my 295. The errors are still -1's, and my clock speed is 620. It really has no effect on the errors what ever I do, so I may as well do as much science as I can until the 480's arrive. I am hoping that I can use the 295 to power the monitor, and the 480's to crunch. I don't know, but am perfectly willing to play guinea pig. Once I have the system installed and set, I will be willing to test anything. I had company today, which delayed me getting the 30 circuit in. Now I started drinking beer, so the power run will have to wait until tomorrow. I love beer.

Steve

---

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

Your CPU temps probably came down due to the GTX 295 sucking all the hot air out of the case :)

---

Steve,

I also have GTX295 and GTX480.
I have no problems with gtx295 using the latest nvidia-driver and using the normal seti-program 6.09 (same speed as lunatics-version!).
You cannot use at this moment the gtx480 for crunching seti.
With which computer do you have the problems?

Skoll/proost (for the beer)!

Ton (ftpd) Netherlands

I thought about the air being drawn out of the case, but I have four fans blowing air into the case, and in addition to the one fan blowing air out, I have wide open areas for air to escape. I can even feel the exit draft 7 feet away from the tower. There is a huge air circulation in the tower. Since the temps returned to normal with no changes, I was even more confused. It stayed in that condition about 12 hours, then corrected itself.

The fan spped on the GTX 295 is still at 100%, and it is still running at 51 Deg. C. My 295 does work with the standard apps without causing errors.

ftpd, do you have your GTX 480 in the same rig as your GTX 295? I know it won't crunce SETI yet, but it will crunch SETI beta. I would like to use the GTX 295 strictly for video, and turn the 480's loose on crunching. Even once I get the build complete, it would be nice to be able to crunch with either the 480's or the 295. That way when crunching regular SETI is ready, I can just switch.

The computer has the following hardware:
Rampage 2 Extreme motherboard
i7 CPU 980 OC'ed to 4.2 GHz water cooled
12 Gig Corsair DDR3 2000 RAM
four 1 Terabyte SATA hard drives
1250 Watt PSU, from BFG.
three 120 mm fans
one 140 mm fan
one smaller exit fan
GTX 295
X-FI PCIe X1 sound card
Klipz 400 watt THX 4.1 speakers
1/5 horse power aquarium chiller
EHEIM 1408 pump
APC 2200 VA UPS system
Blue ray DVD player/recorder
Canopus video converter
Logitech force feedback steering wheel
Logitech rumblepad
22" widescreen flat monitor
HP printer/scanner/coppier combo

being added over the next week or two:
2 GTX 480's
secondary GTX 480 cooling system coonsisting of
two, GTX560 black ice radiators, each with 4 140 mm 91 cfpm fans
EHEIM 1408 pump
tygon tubing with 1/2 id diameter

All this running Windows 7 64 bit ultimate with
Visual Studio 2008
SQL Server 2008 Developer edition
LabVIEW 2009 developer suite
Starry Night Pro astronomy software

I will eventually monitor and control my house with my computer. I will also be able to point a telescope from my computer, and look at the stars and planets from my desk. Not only is this going to be built to crunch some serious science, but it is designed to do anything I throw at it.

Steve

---

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

Steve,

Do you use two computers now? According the specs = yes!

The gtx480 is in another computer. It works OK for Collatz only now.
Milkyway new version 3 is also working OK, but not yet available.
Dnetc is also not working OK with the new cards.

The gtx295 works OK for all gpu-jobs (gpugrid.net special OK)
GPU-temp = 75 C.
PCB-temp = 66 C.
Fanspeed = 79% = 2838 rpm.
Memory used = 292 MB.
GPU-load = 86%

Use nvidia-driver 197.45 and boinc-manager 06.10.50 (beta)!

Ton (ftpd) Netherlands

Steve,

Do you use two computers now? According the specs = yes!

The gtx480 is in another computer. It works OK for Collatz only now.
Milkyway new version 3 is also working OK, but not yet available.
Dnetc is also not working OK with the new cards.

The gtx295 works OK for all gpu-jobs (gpugrid.net special OK)
GPU-temp = 75 C.
PCB-temp = 66 C.
Fanspeed = 79% = 2838 rpm.
Memory used = 292 MB.
GPU-load = 86%

Use nvidia-driver 197.45 and boinc-manager 06.10.50 (beta)!

Ton (ftpd) Netherlands

I use two computers, but only one is described on my last post. I am on my laptop now, and all it is used for is web surfing and email, with an occaisional game of solitair. Never for crunching. My desktop is for anything and everything.

GPU temp = 51 C.
Processor temp = 58 C
GPU fan is at max
Other fans are either uncontrolled or at what ever the system sets. There will be two separate closed loop water cooling systems. One for the CPU, and one for the two GTX 480's

Steve

---

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

I was going to suggest that the card is now dumping heat out of the box more quickly, hence it doesn't get the chance to bring up the ambient temps inside as a result. The heat's "gone" too quickly in a sense.

Just for comparison, one cruncher with three 285s is presently running temps of 81C, 88C, and 78C. The one in the middle is a bit of a surprise as usually it's more up around 90 - 91C.

Just took a closer look - how's this for coincidence....two of the GPUs just finished tasks as I was writing this and were idle for a few moments while new WUs were piped up to them.

Another cruncher I have runs a 9800 GX2 and pushes nearly 100C with the side off the box as well. I am seeing a few artifacts once in a while with that one now so I anticipate it is nearly the end of its life. Would love to get some good news on the 480's question you've been following.

Regards,

---

I was going to suggest that the card is now dumping heat out of the box more quickly, hence it doesn't get the chance to bring up the ambient temps inside as a result. The heat's "gone" too quickly in a sense.

Just for comparison, one cruncher with three 285s is presently running temps of 81C, 88C, and 78C. The one in the middle is a bit of a surprise as usually it's more up around 90 - 91C.

Just took a closer look - how's this for coincidence....two of the GPUs just finished tasks as I was writing this and were idle for a few moments while new WUs were piped up to them.

Another cruncher I have runs a 9800 GX2 and pushes nearly 100C with the side off the box as well. I am seeing a few artifacts once in a while with that one now so I anticipate it is nearly the end of its life. Would love to get some good news on the 480's question you've been following.

Regards,

Thank you for the reply. Could it be that my card is running too cold? That's a strange possibility. I do have a massive amount of air flow in and out of the case.
I have seen this term used several times, but I don't know what it is. What are artifacts, and how do I recconize them? I just got the 30 Amp circuit installed, and my new UPS is running nicely. It has tricky settings, so I'll have to get used to it, but at least I have clean power, and plenty of it.

Steve

---

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

Steve, just read thru the thread, and one idea came up to maybe explain what happened. Is there a chance that whatever WU's you were working on were working the system less stressfully, and when they were thru things went back to 'normal'? Just tossing it out there...

---

Artifacts are where the GPU is no longer rendering the images sent to your monitor correctly in all respects.

What you see is basically collections of pixels that are just junk. The image on the screen (like say how this webpage looks at the moment) is missing bits and pieces of itself in random locations. It can get even worse and the whole screen image just freezes with "snow" across it (remember the old days when you would get poor reception on the old tv tube?!).

I'm seeing the old card looking like it might well be on its last legs and I wouldn't be surprised. It's been running flat out on SETI for just over a year and cranked out over a million doing so. I haven't been able to cool it any better than its doing now - a poor cooling design for a dual GPU card essentially.

But then Nvidia likley didn't envision some SETI junkie running the card 24/7 as their target market either eh!

[edit] Forgot to answer your question! No - colder is good. I've been in many data centres with the air conditioning howling around me and cold is always better. [/edit]

Regards,

---

Steve, just read thru the thread, and one idea came up to maybe explain what happened. Is there a chance that whatever WU's you were working on were working the system less stressfully, and when they were thru things went back to 'normal'? Just tossing it out there...

It could be, but the coincidence was that it was the only time it has ever happened, and it was at the exact time I set the GTX 295 fan to 100%. Everything has been normal all day. I'm glad I didn't have time to try over clocking it further, and end up with a temperature problem.

Steve

---

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