Dissapointing results with HR-03 Plus and 8800 Ultra

MoJo · Post by **MoJo** » Fri Jul 27, 2007 12:17 pm

I was impressed with how quiet the stock cooling on an 8800 Ultra is, but decided to go for a HR-03 Plus anyway. Installed it with Arctic Silver 5.

The temps are very dissapointing to say the least. With a Panaflo 92mm fan at 5V it is silent, but hits over 100C in ATI Tool before I stopped it. Test Drive Unlimitied ran fine though.

Ramping the fan up to 12V I get approximately the same results as the stock cooler, but for the same noise. I am going to get a Nexus 92mm fan tomorrow and try that, see if it is any better.

The worst thing is, the stock cooler exhausts heat while the HR-03 Plus does not.

I double checked that the heatsink is making good contact etc. Is this reasonable performance to expect from an 8800 Ultra? These things just seem to be very, very hot no matter what. Considering it whines anyway when playing games, the stock cooler might be the best bet.

WR304 · Post by **WR304** » Fri Jul 27, 2007 12:42 pm

The only other 8800 Ultra temperatures using a quiet-ish fan I've seen were from falcon26's PC

viewtopic.php?t=41722

That had quoted load temperatures of 66c using a Nexus 92mm fan (at 12v?).

I'm not sure what ambient temperatures were or what was used to load the card though.

MoJo · Post by **MoJo** » Fri Jul 27, 2007 2:49 pm

Yeah... those temps tally with what I have found elsewhere on the net too.

I have double checked the fitting of the HR-03, and it seems to be making good contact. Then again, the whole system seems to be running a bit hot.

I am wondering if it's the fans I have. The CPU (Core 2 Quad OC'ed to 2.9GHz) hits about 70C in Speedfan (closer to 80 in CoreTemp) under load. Sythe Ninja with the stock fan at 7V. Maybe the Nexus fans are that much better?

It's in a P182 with front intake and rear outlet fan. I tried blocking the top vent but it made no difference. I double checked the CPU HSF too, it is making good contact.

Aeria · Post by **Aeria** » Fri Jul 27, 2007 3:35 pm

Try using the top vent with an intake fan and also try a 120mm fan on the hr-03.

christopher3393 · Post by **christopher3393** » Fri Jul 27, 2007 5:39 pm

Try using the top vent with an intake fan

Can you say more about why this might help?

MoJo · Post by **MoJo** » Sat Jul 28, 2007 1:18 am

Aeria wrote:Try using the top vent with an intake fan and also try a 120mm fan on the hr-03.

Thanks, on your advice I am going to try wedging a 120mm fan onto the HR-03.

One of the problems with both the Ninja and the HR-03 is that it is very hard to tell if they are making good contact with the chips. They look flush and are fairly tight. When I take them off, the thermal paste is evenly spread and "disturbed", but...

MoJo · Post by **MoJo** » Sat Jul 28, 2007 4:43 am

A little update.

I ramped the HR-03 fan up to 12V. Far too noisy of course, but I now get 67C load temps. Again, this seems quite poor compared to what other people get with similar set-ups.

I lapped the Ninja because I noticed that there was a mark left by the protective sticker. My temps went from 85 load down to 79 load (on one core, the lowest is 73). Still not anywhere near this HSF should perform I think. Can it simply be that I am using the stock Sythe fan? Surely not. I know it's a quad core CPU but...

I am wondering if the thermal paste I am using is okay. It's Arctic Silver 5. Wasn't there a bad batch of this some time ago? I suppose I could try the thermal paste that came with the HR-03.

WR304 · Post by **WR304** » Sat Jul 28, 2007 5:27 am

MoJo wrote:I lapped the Ninja because I noticed that there was a mark left by the protective sticker. My temps went from 85 load down to 79 load (on one core, the lowest is 73). Still not anywhere near this HSF should perform I think. Can it simply be that I am using the stock Sythe fan? Surely not. I know it's a quad core CPU but...

Intel Core 2 Quad CPUs (130w) are supposed to have roughly twice the heat output of a Core 2 Duo (65w) so you have to expect higher temperatures to some extent.

http://www.techreport.com/reviews/2007q ... dex.x?pg=1

So long as load temperatures are below 85c the CPU shouldn't be throttling.

Did you lap the CPU as well? Intel Core 2 CPUs are often quite badly concave so it's worth doing.

Are you using the plastic socket 775 push pins to attach the cooler? Replacing them with a bolt through setup should help:

viewtopic.php?t=42101

When applying Arctic Silver did you follow the Intel Quad core instructions?

http://www.arcticsilver.com/pdf/appinst ... d_wcap.pdf

There does seem to be quite a large sample variance between Scythe heatsinks as well sometimes.

For the HR-03 Plus are you able to fit the cooler above the card? If you have room that should hopefully give slightly improved performance.

MoJo · Post by **MoJo** » Sat Jul 28, 2007 6:00 am

WR304 wrote:So long as load temperatures are below 85c the CPU shouldn't be throttling.

That is certainly true. Thanks for the reassurance.

WR304 wrote:Did you lap the CPU as well? Intel Core 2 CPUs are often quite badly concave so it's worth doing.

I didn't... It is a rather expensive part so I was reluctant to do anything to it. Looking around it seems like you only get a 2-3C improvement even with concave heatsinks. If it was 10-15C maybe...

WR304 wrote:Are you using the plastic socket 775 push pins to attach the cooler? Replacing them with a bolt through setup should help:

Yeah, I have never liked those 775 plastic pin things. In fact, to remove the HSF I have to take the mobo out of the case because there is so little room to get my fingers in I can't undo one of the pins.

Thing is, I already have a DangerDen RBX waterblock running in the system this one is supposed to replace. I wanted to move to 100% air cooling and a smaller case, but now I am thinking I might stick with the water cooling. I could get a block for the 8800 too.

As long as it's stable and quiet I don't mind a CPU up in the 70s, but the problem is this is with the Sythe fan at full speed. To get it silent I need to replace it with a Nexus at 7V, which is going to bump temperatures over 80 again.

To be fair, it's only when stressing all four cores with something like Prime95 or CPU burn that it gets that hot, but still... With a closed case and the GPU running at 80-90 with a Nexus fan it's gonna be tight. I might go back to the stock HSF on the GPU with Arctic Silver, just to exhaust the heat.

WR304 wrote:When applying Arctic Silver did you follow the Intel Quad core instructions?

Yes.

Many thanks for your suggestions. It's been really helpful.

WR304 · Post by **WR304** » Sat Jul 28, 2007 7:43 am

Something that could give you a small decrease in temperatures would be to try a different thermal paste such as Arctic Cooling MX-1.

According to this it should hopefully drop your temperatures by 2c compared to Arctic Silver 5.

http://www.xtremesystems.org/forums/sho ... p?t=135359

After lapping my Scythe Ninja rev A base and E6700 CPU heatspreader I got a 4c drop in load temperatures.

It took ages to do.

Putting the plastic cover back on the base of the CPU seemed to be enough to protect the chip.

What are your CPU temperatures like when not using synthetic stress tests? There aren't that many programs which are going to load all four cores fully in normal use.

There's usually a big gap between CPU stress test temperatures and the temperatures the CPU reaches in day to day use.

My Intel Core 2 Duo E6700 (10*300, @3hz) goes up to 74c under artificial load but hardly ever goes above 60c in actual use.

If your typical use keeps the CPU a lot cooler than 85c with the slower fan that could be good enough?

The Thermalright HR-03 Plus could be worth lapping too. Some of the newer Thermalright coolers (Ultra 120 Extreme mostly) have had bad press for uneven heatsink bases recently.

MoJo · Post by **MoJo** » Sat Jul 28, 2007 8:25 am

Well, a little update.

I put a 120mm Tricool fan I had spare on the HR-03. At medium speed (7V?) it stays around 86 under load. That's with all case fans on maximum, which is of course far too loud.

The CPU, on the other hand, is looking like a bit of a lost cause. With the case sides on and Prime95 maximum heat or CPUburn, it hits 89 on one core and 82 on the lowest. With a blend test in Prime95 it stays around 75. Again, all case fans on max, with them at more reasonable levels went over 90 before I stopped. Oh, and this is with the CPU undervolted to 2.8V instead of the stock 3.5V.

It does not appear to be throttling at these speeds, at least not according to CPUz. I am wondering if CoreTemp is accurate, as SpeedFan shows temps about 10-14C lower. I could probably live with the SpeedFan temps.

The ASUS util shows another temp, higher than SpeedFan but lower than CoreTemp (closer to SpeedFan). Intel TAT crashes so I don't know what that says. Try Everest maybe?

What really gets me is that I just can't see what I have done wrong. I am a computer technician with two and a half years experience, I know how to fit heatsinks. I followed the instructions carefully, tried various different combinations. Either I screwed up repeatedly on both heatsinks, or the HR-03 Plus is only just up to the task and certainly no better temperature wise than the stock HSF, which is not what reviews tend to say, and the Ninja is just broken.

Then again, these are theoretical tests we are talking about. In general gaming and 3DMark, I don't see any problems temperature wise. Even at full tilt, Prime95 finds no errors and ATITool sees no artifacts. Am I just paranoid?

Take a look at this:
http://www.xtremesystems.org/forums/sho ... stcount=19

I am thinking of getting an Ultra-120 Extreme anyway. I know it's a slower CPU, but the temperature difference is over 40C at load.

MoJo · Post by **MoJo** » Sat Jul 28, 2007 8:36 am

WR304 wrote:After lapping my Scythe Ninja rev A base and E6700 CPU heatspreader I got a 4c drop in load temperatures.

Sounds worth it then. I'll have to buy some more sandpaper.

WR304 wrote:What are your CPU temperatures like when not using synthetic stress tests? There aren't that many programs which are going to load all four cores fully in normal use.

See my above post, written while you were posting

Well, load temps are find, and as I mentioned I don't get any errors. The GPU is a good 10C lower in 3DMark and games as well.

Actually, it was an XFX 8800 Ultra XXX, which comes factory overclocked. It would freeze in ATITool with the stock cooler, but was okay with 3DMark and Test Drive Unlimited (not installed any other games yet). I dropped it back to stock speeds before testing though. I didn't even notice it was the XXX model when buying it, as I had no intention of overclocking really.

WR304 wrote:My Intel Core 2 Duo E6700 (10*300, @3hz) goes up to 74c under artificial load but hardly ever goes above 60c in actual use.

I suppose 85C shouldn't really surprise me with load then, considering there are four cores. Since it is multiplier unlocked, I simply upped the multiplier to 11 to get 2.9GHz for testing.

As I said, I am not sure if CoreTemp is accurate anyway. Theoretically, the CPU should be throttling, but does not appear to be.

WR304 wrote:The Thermalright HR-03 Plus could be worth lapping too. Some of the newer Thermalright coolers (Ultra 120 Extreme mostly) have had bad press for uneven heatsink bases recently.

Yeah, and missing washers from what I read

I think I might just get an Ultra-120 (since the Ninja is covering a memory slot anyway and I still don't think the base is making as good contact as it perhaps should, and comparisons can't hurt anyway) and try it with some Nexus fans.

Thanks ever so much guys.

You know how it is. Easy to panic when your temps are worse than everyone elses. Perhaps watercooling and lower power systems made me too used to low temps. This is the first very high performance machine I have tried to cool on air (or at least tied to cool quietly.) You have reassured me somewhat.

MoJo · Post by **MoJo** » Sat Jul 28, 2007 9:04 am

Found another interesting post:

http://www.xtremesystems.org/forums/sho ... stcount=26

This is what I am getting - an 8-9C difference between cores. Suggests that the Ninja really is not seated properly. Thing is, I can't do much about it short of changing to a solid mounting bracket, which means I may as well just get the Ultra-120 and stick a 2p coin in it.

WR304 · Post by **WR304** » Sat Jul 28, 2007 9:22 am

MoJo wrote:I am wondering if CoreTemp is accurate, as SpeedFan shows temps about 10-14C lower. I could probably live with the SpeedFan temps.

Older versions of Speedfan show inaccurate Intel CPU temperatures.

It sounds like you need to update to Speedfan 4.32.

Speedfan 4.32 Core 2 Duo Temperatures

The main thing missing from most people's discussions of PC temperatures is what the ambient room temperatures are.

With low airflow PCs in particular it makes such a massive difference that it's difficult to safely compare temperatures without knowing it.

In that post you linked the poster has 5x120mm fans in the case plus the 200mm fan of the Antec 900 case running too.

That's going to produce far more airflow than the fans in your P182.

I wouldn't discount that your Scythe Ninja heatsink could be faulty either.

viewtopic.php?t=40522&postdays=0&postorder=asc&start=0

It's not unknown for visually identical heatsinks to work differently also.

A few weeks ago I did a comparison between two (unlapped) 2005 Scythe Ninja rev A heatsinks. The CPU speeds (10*300, 3ghz) and cooling were identical between the two cases.

The room temperature then was 24c approx.

At that time my CPU idled at 41c approx.

Under 100% load on both cores using Intel TAT tool the temperature of my CPU rose to 78c (as shown by Coretemp).

The temperatures for my brother's E6700 CPU at the same clockspeeds were significantly lower though. (The CPU was purchased May 2007)

At a room temperature of 24c his Core 2 Duo idled at 31c whereas mine idled at 41c approx.

Under the same 100% TAT load on both cores his E6700 CPU stabilised at 62c.

Possibly it's just a more efficient processor as it was purchased recently whilst my E6700 was purchased last summer.

That 16c difference in load temperature was why I decided to try lapping my Ninja and CPU in the first place.

I'd want to try the bolt through mounting method before ditching the CPU cooler though.

Here's some good pics showing why you'd want to lap the Thermalright CPU cooler:

http://www.xtremesystems.org/forums/sho ... ght=lapped

MoJo · Post by **MoJo** » Sat Jul 28, 2007 10:37 am

WR304 wrote:It sounds like you need to update to Speedfan 4.32.

That's what I'm on I'm afraid. Freshly downloaded and everything. Maybe Intel changed things on the quad cores again?

WR304 wrote:The main thing missing from most people's discussions of PC temperatures is what the ambient room temperatures are.

That is a very, very good point. What worries me is that it's not even particularly hot right now (weather in England is not very summer like, in fact half of England is under water)

WR304 wrote:I wouldn't discount that your Scythe Ninja heatsink could be faulty either.

There is certainly something not quite right about it I think. The stock cooler, while louder, is considerably better vs. the Ninja with stock Ninja fan at 12V.

WR304 wrote:I'd want to try the bolt through mounting method before ditching the CPU cooler though.

I suppose I could get some bolts tomorrow. It says a lot though that the Ninja is apparently not putting enough pressure on, and not evenly. I am wondering if I can bend the clip somehow to make it sit better.

I'll order an Ultra-120 I think. Aside from anything, the clip is much better and the biggest known problem with it can be solved with a coin.

WR304 · Post by **WR304** » Sat Jul 28, 2007 12:21 pm

Were you able to try the HR-03 Plus mounted above the graphics card?

With the Antec P182 case that should give you more airflow over the cooler when compared to fitting the HR-03 Plus below the card.

MoJo · Post by **MoJo** » Sat Jul 28, 2007 1:32 pm

WR304 wrote:Were you able to try the HR-03 Plus mounted above the graphics card?

With the Antec P182 case that should give you more airflow over the cooler when compared to fitting the HR-03 Plus below the card.

I could try it tomorrow I guess. It would be very near the CPU... although, it might be possible to put it in the lower PCI-e slot maybe. As it is, the front 120mm intake fan is blowing over it.

Tomorrow I'm going to get some Nexus 120mm fans. Put one in the lower chamber, one on the CPU and one as an exhaust at the top of the case. I will also have one on the HR-03 Plus. That's a total of seven 120mm fans - top, rear, front, lower, CPU, GPU and PSU. Four Nexus/Yate Loon, two Papst and whatever is in the PSU. My current water system has only five - 92mm front and rear, 120mm on HDDs, rad and PSU.

I made a divider to keep heat from the GPU away from the CPU as well, but so far I'm not sure it really makes any difference.

The issue with non-flat Ninjas and Intel IHSs seems to be common. I think the major difference with the Ultra-120 is that it can put enough pressure on to even things out. After all, it's just metal, it will bend a bit with pressure. If you compare temps of a Ninja when vertical and horizonal, even though it's not that heavy it easily makes >5C difference.

Many thanks WR304. Just a shame I have to wait for the Ultra-120 now, I want to crack this.

WR304 · Post by **WR304** » Sat Jul 28, 2007 2:49 pm

MoJo wrote:That's a total of seven 120mm fans - top, rear, front, lower, CPU, GPU and PSU. Four Nexus/Yate Loon, two Papst and whatever is in the PSU.

Maybe sticking with the watercooling would be an idea.

Having just one fan running is bad enough.

That's basically the point of this thread too:

viewtopic.php?t=39804

With most new-ish Socket 775 motherboards there should be enough space to fit a HR-03 Plus and fan above the graphics card.

It's only if you have an older one or a massive northbridge cooler (like some of the MSI motherboards) that there coud be an issue.

This pic shows the amount of clearance with a Asus P5K (non deluxe) motherboard and thermalright HR-03 (non plus) cooler.

Suspended fan in Antec Sonata case cooling Intel E6700@3ghz/ Nvidia GF700GTX

MoJo · Post by **MoJo** » Sat Jul 28, 2007 3:16 pm

WR304 wrote:Maybe sticking with the watercooling would be an idea.

It had crossed my mind, but the main issue is space. My current case is a large server model, basically two large ATX cases stuck together. It's brilliant for water cooling, but I am moving to a smaller house and want to save space.

Might just have to go with it though. Or should I say, wait until DangerDen produce an 8800 Ultra compatible waterblock. Since all my kit is DD and all 1/2" ID tubing, there is really no other option. The two new blocks would be around Â£120 total.

WR304 wrote:Having just one fan running is bad enough.

I don't think it's necessarily a problem to have lots of fans. I mean, as I said my current system has three 120mm fans and two 92mm fans, all at 5V. The system is basically silent - push the power button and a light comes on, otherwise nothing.

In theory, you it should be possible to run the system on the same fans with air alone. Water cooling helps concentrate the heat of the two hottest components on the radiator where a single 120mm fan cools them both easily, but in the end it's the same amount of heat overall.

Currently I am running a X2 4200+, idling at 38C, and a 7800GT idling at 46C. Neither gets above 50 under load.

WR304 wrote:That's basically the point of this thread too:

Yeah, I read that one too. The people arguing that air should be as good as water must not have actually tried water I think. Sure, the theory is there, but water is so much easier to cool quietly.

WR304 wrote:With most new-ish Socket 775 motherboards there should be enough space to fit a HR-03 Plus and fan above the graphics card.

Presumably the fan is pulling down. Wouldn't this just pull air off the CPU, which is going to be slightly warmer air? Also, with the heatsink on top, it will be out of the direct path of the front intake fan.

I will try it anyway. To be honest, the stock cooler may be the way to go. It is pretty quiet anyway, and can be quieter with some tweaking. Whenever the GPU ramps up I will have speakers/phones on anyway, and the P182 is amazingly good at dampening noise anyway. The major advantage is, it exhausts out the back of the case.

Someone needs to make something like a Ninja for a GPU, with the fan vertical and blowing out the back. The heatsinks could be massive and cool everything, just like a stock cooler. Four or five bays taken up is fine (that's what the HR-03 uses anyway and most people only need one PCI slot for a soundcard) and a built in shroud would ensure all heat is sent straight out the back of the case. Maybe I should try to make something, it looks doable with heatpipes:

http://www.geocities.co.jp/SiliconValley-Oakland/5070/

WR304 · Post by **WR304** » Sat Jul 28, 2007 3:50 pm

MoJo wrote:Presumably the fan is pulling down. Wouldn't this just pull air off the CPU, which is going to be slightly warmer air? Also, with the heatsink on top, it will be out of the direct path of the front intake fan.

In that pic the fan is blowing down onto the graphics card cooler. There isn't an intake fan.

In use with a Nvidia 7900GTX card and E6700 Core 2 Duo CPU the graphics card cooler is always far hotter than the CPU cooler.

It seemed to make more sense having direct airflow onto the hot graphics card cooler.

It might be different with the Core 2 Quad though.

Have a look at these figures for a Thermalright HR-03 Plus mounted above (position #1) and below the card (position #2) in an Antec Nine Hundred case.

http://www.madshrimps.be/?action=getart ... rticID=557

If you look at the fanless figures they're quite a bit better with the cooler mounted above the card suggesting it could work better in that configuration.

I've never actually tried both layouts myself though. The HR-03 on that 7900GTX has always been mounted above the card.

I found this link about Core 2 Quad and Speedfan 4.32 temperatures also:

(See the "required software" section)

http://forums.speedguide.net/showthread.php?t=225919

.

MoJo · Post by **MoJo** » Sun Jul 29, 2007 2:44 am

WR304 wrote:It might be different with the Core 2 Quad though.

Well, yes... I am starting to wonder. If the CPU hits, say, 75C at load, and the graphics card hits 85 load, there isn't actually that much difference between them.

I did try it that way up but it made no appreciable difference.

WR304 wrote:Have a look at these figures for a Thermalright HR-03 Plus mounted above (position #1) and below the card (position #2) in an Antec Nine Hundred case.

I simply cannot understand how they got these figures. I know the card is a GTS, but they are not exactly cool, right? 48C load at 7V. I'm going to experiment more I think.

I just keep thinking maybe it's me doing something wrong. As I say, I have plenty of experience. I was personally repairing 80-100 computers a month, and I've never had this amount of trouble with heatsinks. Even on P4s which have similar TDPs to the Core 2 Quads.

One thing I have noticed is that the reviewer on the Madshrimps site installed the HR-03 incorrectly. When he has the HR-03 under the graphics card, he has the bolts on the GPU (lower) side. The official instructions say put them on the top side away from the GPU, which makes sense as it's a lot easier to tighten them that way.

WR304 wrote:I found this link about Core 2 Quad and Speedfan 4.32 temperatures also:

That matches what I get too. How do you find this stuff man? I searched around for ages on the top, trying to figure out which set of figures to take.

It's a shame SpeedFan does not support my mobo (P5N32-E SLI) properly yet, as I could do with some fan control.

I am seriously considering getting a couple of waterblocks. Â£140 total but it's pretty much 100% certain to work. Will have to live with a massive case though, which could be a problem. The alternative is to get an Ultra-120 Extreme (Â£45) and maybe a Zalman 900CU (Â£30).

On the plus side I found some springs and bolts from an old RBX waterblock I had for Socket A. If I just get some bolts and washers for the bottom of the mobo I should be able to fit them to the Ninja. Since I'll have to wait for the Ultra-120 anyway, might as well try.

MoJo · Post by **MoJo** » Sun Jul 29, 2007 9:28 am

After some experimenting today, I have moved a bit further on.

I decided to get some Nexus fans, since that was all I could do today and I needed some anyway. I now have Nexus 120mm fans on the CPU, GPU, rear and front grills of the case. On the top vent and in the lower chamber, I have the Papsts.

The CPU fan speed control does not seem to work (BIOS and ASUS PC Probe II controls do nothing) so I installed a Zalman resistor to drop it down from 12V a bit. Runs at a little under 1000 RPM. The lower chamber fan runs at 60%, around 800 RPM, and the rest of the fans are at 80% which gives a little under 1000 RPM.

I lapped the Ninja, the CPU, the HR-03 and the GPU. Temps dropped 3C on both. I applied the Thermalright thermal paste instead of the Arctic Silver. The TR paste is more like liquid, and dropped idle temps about 5C. I guess Arctic Silver only works well when both surfaces are pretty flat. Even after lapping it seems that the GPU and CPU are a bit concave.

Noise levels are acceptable but not as good as I would like. I am finding it hard to tell how good they are. My room is not that quiet, because there seems to be constant noise from water pipes in the house. It's a soft hissing, and does not bother me. My current rig is quieter than that. I'm not sure if the new one would be when under my desk, I will have to give it a try. The new house should be even quieter though.

Temps are not brilliant. CPU hits 92 in Prime95 on all four cores, but does not appear to throttle (how can you tell? CPUz shows no drop in clock, and I have had temps go higher. Max temp is 100C according to CoreTemp). On the other hand, at least now all the cores are within 2C of each other, where as before lapping and with Arctic Silver there was 8-10C difference.

The GPU hits 90C too in ATITool, although the CPU drops to about 87 because ATITools uses some CPU time. Again, no artifacts or detectable issues. 3DMark is fine. HDD temps stay under 40.

I experimented a bit with fan speeds. Even ramping them all to 100%, the CPU only got down to 86.

I think everything would probably be fine if SpeedFan worked on my mobo. I could set it to ramp the fans up when the CPU or GPU gets really hot, but generally keep them slow. SpeedFan can only control one fan on my mobo (not the CPU fan sadly) but it can run it as low as 40% (around 5V?) before it stops where as the BIOS and ASUS Probe only go as low as 60%. I have a fan controller than can provide 5V and I am tempted to go that way. If I could control CPU and intake fan speed with SpeedFan...

One thing I have noticed is that fan noise is a bit smoother with the fan controller. My guess is that's because it puts out true 5V DC, where as the mobo headers are PWM.

I'm tempted to try and live with it. Problem is that because I'm moving, things are a bit up in the air. I was going to ditch the old case before going, but now I'm thinking I might need to keep it somewhere in case I need to go for water cooling. Either that or try and squeeze all the water stuff into the P182. The main thing is I would need a smaller rad, and I wonder if a it would be up to the task compared to my current heatercore (which is a bit bigger than a 120mm fan). Oh, and add another Â£40 for the rad, basically looking at Â£200 in water cooling gear.

Well, the other option is to try an Ultra-120 Extreme. It's a shame there are no more after-market coolers for the 8800 as well. I am tempted to go back to the stock HSF, because it's fairly quiet under normal conditions and blows air out the back of the case, which would probably drop case temps quite a lot. Time to go experiment perhaps.

I am just amazed no-one has come up with a VGA cooler which blows air out the back like the stock one, but which uses a better fan. Arctic Cooling would be the usual place to look, but they currently don't have any 8800 compatible gear. Even then... What is needed is a 92mm or 120mm fan, either at the end of the card under it at a 30 degree angle, with a shroud and massive fin/heatpipe arrangement.

WR304 · Post by **WR304** » Sun Jul 29, 2007 12:37 pm

According to this article about the QX6700 Core 2 Quad processor it will start throttling at 81c.

(Edit: See my next post down also as this isn't correct.)

http://www.digit-life.com/articles2/cpu ... core2.html

Also have a look at this thread about the new Q6600 G0 processors and their load temperatures:

http://www.hardforum.com/showthread.php?t=1212009

With the graphics card have you tried it using the stock cooler and a single slot PCI blanking plate for better airflow?

It could make the stock cooler a bit quieter in use due to less air turbulence.

You might be able to just swap the PCI plate from the 8800 Ultra with the one from your old 7800GT card?

For the fan control have you considered something such as a mCubed T-Balancer? The software features would give you quite a bit of control over fan speeds.

http://www.t-balancer.com/english/index.htm

.

MoJo · Post by **MoJo** » Sun Jul 29, 2007 1:44 pm

WR304 wrote:According to this article about the QX6700 Core 2 Quad processor it will start throttling at 81c.

I tried Throttlewatch but it didn't detect anything at over 81C.

WR304 wrote:You might be able to just swap the PCI plate from the 8800 Ultra with the one from your old 7800GT card?

That's an excellent idea, I will try it!

WR304 wrote:For the fan control have you considered something such as a mCubed T-Balancer? The software features would give you quite a bit of control over fan speeds.

It looks good, and I must say I am tempted.

I have been playing around a bit more with the system. Under my desk in my room I can't hear it with all fans on 60%. Temps are not brilliant but... I suppose the question becomes, how reliable is this configuration? I suppose I should run 3DMark in a loop for a few hours.

Hopefully the stock cooler and an Ultra-120 can bring things under control. If I can get to a max of 75C for the CPU and anything under 80 on the GPU at full load at the current noise levels, I'll be happy. I think I can push the fans a bit more, currently the two WD5000AAKS HDDs are the loudest things in the system.

WR304 · Post by **WR304** » Sun Jul 29, 2007 2:11 pm

It looks like the Digit-Life.com article I posted had the 15c issue with their monitoring software. If you add 15c to 81c the resulting 96c is about the same as other links are saying.

Version 2.25 of RMClock released April 2007 updated the max temperature settings for Intel Core 2 Quad processors.

http://cpu.rightmark.org/

There are lots of threads on google saying that their Core 2 Quad processors don't throttle until just under 100c.

"Intel specs the Q6600 B3-stepping to 100C, which means most chips actually don't throttle til somewhere between 90C and 100C reported in Intel TAT or Coretemp software (mine hits 96C as the throttle point-- the way those programs dervive the reported temp is imperfect, which explains the variation)." Continuum

http://episteme.arstechnica.com/eve/for ... 724831/p/3

http://forums.anandtech.com/messageview ... erthread=y

I'm sure it's in this documentation somewhere but there's quite a lot of it to wade through...

http://developer.intel.com/design/core2 ... tation.htm

.

MoJo · Post by **MoJo** » Mon Jul 30, 2007 1:56 am

WR304 wrote:It looks like the Digit-Life.com article I posted had the 15c issue with their monitoring software. If you add 15c to 81c the resulting 96c is about the same as other links are saying.

Very interesting. That certainly bears out what I am getting.

Searching around, my temps actually don't look that bad. For example:

http://forums.nvidia.com/index.php?show ... pid=204576

He says a 10C drop from high speed (76CFM) fans. That makes one core 80C under load with quiet fans and the Everest System Stability Test. Everest probably doesn't load the CPU as much as cpuburn or Prime95 though, as it is most likely hammering the memory as well.

Water would probably get me to 70C. I think I can get lower than I am now though by pushing the GPU air out the back of the case. I have been trying to find a way to install some kind of heatsink to do this, but it looks like the stock cooler is the only option.

I think I will just have to live with the temps. They are scary, no doubt, but apparently not a problem. I mean, if Intel only start throttling at ~96C...

WR304 · Post by **WR304** » Mon Jul 30, 2007 6:04 am

Using the stock 8800 Ultra cooler you should be able to fit the "classic" cardboard exhaust duct which would help exhaust heat out of the case.

The vents on the 8800 Ultra don't look any different to the standard 8800GTS/GTX cooler.

Fan Duct on 8800GTS 640mb card

Something else to consider is trying the new Arctic Cooling MX-2 thermal compound also:

http://www.arctic-cooling.com/further_prod2.php?idx=140

.

MoJo · Post by **MoJo** » Mon Jul 30, 2007 6:12 am

That's the plan. I have ordered an Ultra-120 Extreme and am going to take a look at the graphics card now.

MoJo · Post by **MoJo** » Tue Jul 31, 2007 7:46 am

Installed the Ultra-120 Extreme with a 5 yen coin (has a hole in the middle do it stays on centre) and my temps are 65C on the hottest core after 1 hour in Prime95!

That's 25C better than the Ninja! Same fans, same speeds, same case, same everything.

Was tempted to chuck the Ninja in the bin. The Ultra-120 is much more solid, feels much better built, is smaller, and apart from the clamping issues which are easily fixed out-performs both the stock cooler and Ninja by a considerable margin. Installation was easy too, better than the stock cooler in fact and streets ahead of the Ninja.

Saying that, I will probably keep the Ninja. With a lower power CPU it should be able to cool quietly to a reasonable degree. I am just shocked at how poorly constructed it is. I'm sure that's the problem with it - it probably works brilliantly when you get a well made one with flat base and good hold-down clips. The 775 clips are rubbish though, and the fins don't seem that well attached to the heatpipes either. The fan mount is very poor compared to the solid, vibration dampened Ultra-120 design.

cmthomson · Post by **cmthomson** » Tue Jul 31, 2007 10:18 am

I'm surprised nobody else has mentioned mounting pressure of the Ninja.

A few months back, this was a much-discussed topic. Many of the push-pin Rev B Ninjas just don't have sufficient pressure against the CPU IHS for good heat flow. In some instances this was because the pins weren't fully inserted (apparently they click twice, and it's recommended that you take out the motherboard and visually inspect the back to ensure the pins are engaged and fully spread). In other instances the springs are simply inadequate.

Some people overcame this with the copper insert that comes with the universal retention kit. Others went to homebuilt bolt mounts.

The way you describe your U120 installation makes me suspect the primary difference from the Ninja is the mounting pressure.