Cooling a Q6600

[mnmsmith]

Plan to get a Q6600 when the G0 stepping is available. In the meantime I'm trying to determine what cooler I am going to install.

I'm going with either the Gigabyte GA-P35-DS3R or MSI P35 Neo2 board. I have an Antec 900 case. I'm not a gamer, I do video editing and such, and do plan to overclock.

I am leary about using the Thermalright Ultra-120 or Tuniq Tower 120 because of the size and do not want to have to mod the board to make it fit, so I am considering the Thermalright Ultra-90. From reviews I have read it looks like a solid cooler, just not sure about with a quad core system. Any input?

Thanks!

 

[Raider1284]

Are you planning on overclocking the quadcore? Are you worried about the noise from the stock heatsink? If you answered no to both these questions the stock heatsink "has to be alright".

 

[mnmsmith]

Yes, I do plan to overclock. I'm not an overclocking expert, so I am not sure how far I'll go, but the answer is yes. Noise from the stock heatsink was never something I thought about because it has always been a given that I upgrade the cooler when I buy a new CPU.

 

[BonzaiDuck]

Well, mnmsmith --

We're on the same trail here.

First -- there are some new developments, with ThermalRight releasing a yet-to-be-reviewed cooler named the "Ultima-90." Supposedly, the new cooler-entry provides for use of either 92mm or 120mm fans.

Here are some open issues, and gurus like AigoMorla here in our forum may have additional insights.

I want to create an expression called "thermal scaling." What I'm trying to describe is how a single-core processor that runs "hot" -- like the P4 Prescott "E" -- might have a TDP of 105W of peak thermal power, so that overclocking it by so many megahertz and so many hundredths of a volt in VCORE voltage would increase that thermal power to -- say -- 120 or 130W. But a dual-core or core-2-quad running at 105 or 110W at stock settings might behave differently. I'm sure there is some sort of deductive logic that would explain that such differences in "thermal scaling" exist between processors, or that would otherwise "explain it away."

AigoMorla seems to think that really "serious" over-clocking of a Core-2-Quad almost requires water-cooling. To take the meaning of "serious" to his own limit, he's most probably correct with that.

But so far, I've gleaned a few things from customer-reviews for the Q6600. First, the ThermalRight Ultra-120-Extreme is quite good enough for a Q6600 running at stock speeds, so since it's a better cooler than the stock cooler (by light-years), you should be able to squeeze extra from it.

Second, the performance boost you get from four cores at stock or mildly over-clocked speeds as opposed to an E6600 should give opportunity to temper one's over-clocking ambitions.

Third, with the first two points in mind, I have some good "intelligence" that you can do as follows with either the stock cooler (or better) -- the TR Ultra 120 or Extreme. You can drop the multiplier from 9 to 8 (or even 7), go forward with your plan to get high-performance, low-latency memory (if that was, indeed, your plan, because it is certainly mine), and then -- raise the FSB so the 2.4 Ghz Core 2 Quad runs at around 2.7 Ghz.

That's what I'm hoping to do. Maybe -- a bit better -- but I cannot tell until the retail Q6600 arrives on Thursday or Friday.

And -- of course -- we can always explore the "limits," because we can find for ourselves the "Chapter 5" Q6600 thermal limit specs at the Intel web-site.

 

[adairusmc]

I am using a Thermalright Ultra-120 Extreme on my new B3 Q6600 and it is working great. I still have not tested load temps with it, but so far it is running about 32-34C idle consistently (temps taken with speedfan and Everest Ultimate Edition).

 

[firewolfsm]

Do you think overclocking on stock volts is possible with the stock cooler?

What about an AC Freezer 7?

 

[BonzaiDuck]

Well, Firewolfsm, I'm trying to get a more precise handle on your question.

I still have to get the quad-core in my hands -- right now UPS is in transit.

Only speculating, I just guess there is some wiggle-room on the "Auto" vcore setting and using the stock cooler.

I don't think I ever wanted to even bother with a stock cooler though, while it's interesting that some here post their results. If I'm going to up the FSB at all, or even tweak the VCORE by a couple hundredths of a volt, I'd just as soon narrow the difference between the idle temperature and the load temperature. In fact, I'm predisposed to do that anyway if running everything at stock settings. . . . . "It's . . . . just . . . . me . . . . I guess. . . . "

Now, with mnmsmith's remark about waiting for the G0 stepping, I'm wondering about the limitations of what the UPS lady brings.

A BIOS-flash mishap on my Striker board a couple weeks ago made me worry about my E6600 (seems to be fine now), and when things like that stick in your head on the day people expected some big price drops, one can act . . . . impulsively . . . .

 

[BonzaiDuck]

On the AC Freezer 7:

I can't keep facts straight in my head about all the available coolers, but I look for performance comparisons all the time, like those in the Anandtech reviews between March and May of this year. I just look for the cooler that shows the lowest thermal resistance.

If you plod through my "Ducting" thread posted today, it is only implicit there that one would find additional factors besides pure performance in choosing a cooler. Personally, I'm not easy to compromise or deviate from getting the model with the lowest thermal resistance, but fitting it in the box, or how it fits with other components may be a valid consideration.

But even so -- if it's copper and has heatpipes, it would promise to be more effective than the stock cooler. I'm assuming, of course, that the Q6600 (or even the QX6850) still sports those minimalist aluminum-copper heatsink designs that have come up through more than one generation of Intel processors. I'd be absolutely confident to say the Freezer improves over the E6600's heatsink and fan, and (if I remember correctly) the Freezer 7 has had some very favorable reviews.

I was just looking at the last ULTRA-120-Extreme comparison review here at Anandtech, and I don't find the Freezer among the 20-some coolers evaluated. You'd need to search for some more reviews and see if you can get a handle on the thermal power generated by the review bench configurations. Then either look at the idle and load temperatures in comparison to those of the U-120-Extreme or any of the others -- just to see how it stacks up. IF there is some basis for matching the thermal power of an unreviewed cooler against a test-configuration in a comprehensive comparison review, then you should get an idea based on those temperatures.

However, since I see that Anandtech shows cooler idle-load performance at stock settings, any other review of the Freezer test configuration at stock settings should give you some idea in cross-comparison. Just make a note from the AMD or INtel web-sites about the TDP spec in their "Thermal Considerations" of spec sheets.

 

[SolMiester]

Hi all, I too am looking to do Q6600, but with the Zalman 9700 which is only rated for C2D.

If the 9700 fits inside the Wind Tunnel of my TJ06 like so, do you think it would be cool enough ?

 

[BonzaiDuck]

Yo, AdairUSMC!

Assuming you have an air-conditioned refuge from any "Baghdad summer" temperatures -- (I'm only accounting for some probability of your location -- not the certainty of it) -- you might fire up Everest and run their stress-test. It is milder than ORTHOS, but it would give you (and us -- while we wait for the UPS lady) some idea of a load temperature value at stock settings.

At around 75F, my E6600 -- that's the Core 2 Duo -- will show idle values around 28 to 30C. So? Your Q6600 is running just a few degrees hotter at idle. But this doesn't give us an idea of the idle-load temperature spread. At near-stock settings, the E6600 runs up to a load value somewhere in the vicinity of the high-30's Celsius. [I'd need to check again to see if it even breaks 40C.] I'd expect a Q6600 to go beyond that, but obviously, and given the design of the stock cooler . . . . well . . . . we could form a betting pool about whether it even comes close to 50C . . . or 45C at stock settings. The thermal limit on these processors is way, way beyond that.

Someone here who is running a Q6600 has to know, even if they use a different cooler. At least there, you could go to Anandtech's May Ultra-120-X review, and if the "different cooler" is shown by comparison, we'd have a very good idea.

For me, I'll know sometime between Thursday and Monday.

 

[BonzaiDuck]

SolMeister --

That's very interesting! I can almost imagine, like some ship in a bottle, how the Zalman "copper flower" could fit inside that duct!

You WILL get lower temperatures with that duct and the Zalman than would show in the May Anandtech comparison with the TR Ultra-120's. See my thread on ducting. It's a slam-dunk. So even if the TR coolers outperform the Zalmans, you may probably erase the shortfall in the Zalman's performance.

 

[SolMiester]

Originally posted by: BonzaiDuck
SolMeister --

That's very interesting! I can almost imagine, like some ship in a bottle, how the Zalman "copper flower" could fit inside that duct!

You WILL get lower temperatures with that duct and the Zalman than would show in the May Anandtech comparison with the TR Ultra-120's. See my thread on ducting. It's a slam-dunk. So even if the TR coolers outperform the Zalmans, you may probably erase the shortfall in the Zalman's performance.

Bonzai, I'm sure some of that went over my head....Its reaccering to know you think it will be enough to cool the Quad, not sure what to go for with a OC yet, might have to sit back and watch you guys 1st.
Anyway just to ensure no wind will sneak around the flower, I thought I might stick a band of foam inside the duct and along the bottom to seal any escaping air. Should prove to be a nice quiet pull/push/pull system. The 92mm behind the HDD is the only system exhaust though.
I'll prolly only get a 88GTS to go with it so the 2nd & 3rd chambers shouldn't get too hot...

 

[adairusmc]

Originally posted by: BonzaiDuck
Yo, AdairUSMC!

Assuming you have an air-conditioned refuge from any "Baghdad summer" temperatures -- (I'm only accounting for some probability of your location -- not the certainty of it) -- you might fire up Everest and run their stress-test. It is milder than ORTHOS, but it would give you (and us -- while we wait for the UPS lady) some idea of a load temperature value at stock settings.

At around 75F, my E6600 -- that's the Core 2 Duo -- will show idle values around 28 to 30C. So? Your Q6600 is running just a few degrees hotter at idle. But this doesn't give us an idea of the idle-load temperature spread. At near-stock settings, the E6600 runs up to a load value somewhere in the vicinity of the high-30's Celsius. [I'd need to check again to see if it even breaks 40C.] I'd expect a Q6600 to go beyond that, but obviously, and given the design of the stock cooler . . . . well . . . . we could form a betting pool about whether it even comes close to 50C . . . or 45C at stock settings. The thermal limit on these processors is way, way beyond that.

Someone here who is running a Q6600 has to know, even if they use a different cooler. At least there, you could go to Anandtech's May Ultra-120-X review, and if the "different cooler" is shown by comparison, we'd have a very good idea.

For me, I'll know sometime between Thursday and Monday.

I will give that a try as soon as I get home, and post the results.

 

[adairusmc]

Well, I did some testing with Everest Ultimate as BonzaiDuck suggested, and here are the results.


Here is a screenshot of CPU-Z, Speedfan, and Everest Ultimate (the light blue sidebar utility) at idle, about an hour after I booted the machine up.

Idle

Here is a second screenshot I took, with the Everest Ultimate stress test window open along with the others in the previous screenshot, showing the load results. Note that I ran the load test for just over an hour before taking the screenshot.

Load

These last two images are pics of my setup, to give you an idea of what kind of airflow I have going on with this setup. Note that the part of the window directly above the Ultra120 Extreme has been cut out - I did that a while back so that a little more air could move through the side of the case, it did improve temps when I did that.

Computer Pic 1

Computer Pic 2


The idle temp before starting the test shows 31C in both Speedfan and Everest Ultimate. After an hour of the torture test, the highest temp I saw was 45C, but it did spike to 46C for a brief second once. The temperature of my house during the testing was around 74F according to the thermostat.

I am sure the Everest stress test may not test as hard as some others, but for an hour of the test, I consider that load temp to be pretty damn good for air.

Specs of the system are in my sig.

 

[eelw]

I'm using the ThermalTake V1 on my Q6600 OCed to 3.05GHz currently. Full load temps reach 63C reported by CoreTemp.

 

[BonzaiDuck]

Well, that's not bad, because AdairUSMC shows a load temperature range of 59 to 63C among his four cores, and he's using the TR U-120-Extreme.

The TT V-1 cooler was reviewed against the Zalman and a few others in Maximum PC sometime in the last few months. I was peeved at Max PC, because they pre-select the manufacturer choices in comparison reviews, and it always seems evident that the "Kick-Ass 10" award goes to a company that's purchased a full-page-color ad with them. At least (I hope, anyway) -- they don't phony up the test results.

I thought it was an interesting design for a TT cooler. Fact is, the smaller they can make these things, the better -- at least for me.

You can probably get a handle on TT's standing with the TR Ultra Extreme by looking at Max PC's comparison review with the Zalman CNPS-9700, and cross-referencing the results (including bench-configuration thermal power) with the Anandtech review of May 5 that included the Zalman and TR coolers.

I'm also wondering -- if TT provides the cooling fan -- if it might turn out that more throughput CFM using a different fan-choice might even show better results.

I'm preparing to install my Q6600 right now. For the benefit of Solmeister, I'll post my settings for a lower multiplier and pumped-up FSB. Provided, of course, there aren't setbacks and mishaps.

I suppose I'd better switch on the house AC after that sucker is running in my Striker board.

This might all prove interesting as much for me as for you other posters. AdairUSMC's rig is impressive -- I think that's a TT case? If I show lower load values at same or higher clock settings, it would be either for the case differences or the motherboard. I usually get lower-to-mid-range motherboards, but I went for the Striker because it has "8-cycle voltage regulation," "Stack-Cool" mobo design, and capless power-regulation. Like my old P4P800SE, it seems to run cool. And I heard that the eVGA (nVidia 680i) runs hotter than the Striker.

Of course -- Max PC chose the eVGA over the Striker board in their 2007 "Dream Machine" system. They say the eVGA has an edge over Striker for OC'ing. But that depends on a lot of things, although it appears they used water-cooling for the DM rig this year.

 

[adairusmc]

Originally posted by: BonzaiDuck
Well, that's not bad, because AdairUSMC shows a load temperature range of 59 to 63C among his four cores, and he's using the TR U-120-Extreme.


This might all prove interesting as much for me as for you other posters. AdairUSMC's rig is impressive -- I think that's a TT case? If I show lower load values at same or higher clock settings, it would be either for the case differences or the motherboard. I usually get lower-to-mid-range motherboards, but I went for the Striker because it has "8-cycle voltage regulation," "Stack-Cool" mobo design, and capless power-regulation. Like my old P4P800SE, it seems to run cool. And I heard that the eVGA (nVidia 680i) runs hotter than the Striker.

Yep, that is a Thermaltake Aluminum Armor case I am using there.

Update on overclocking and temps with my setup. I played with it for a few hours yesterday, and got it up to 3 Ghz (1334 FSB) stable last night (11 hours of running two copies of orthos, no errors at all). I had to set the Vcore to 1.26 and the CPU VTT to 1.35 to get it stable.

The idle temp at that speed is now around 37C, with the same room ambient temp as my previous post. The load settings this morning after running 11 hours of Orthos overnight was around 55C, with the individual cores showing 65,65,63,67 according to Everest Ultimate and Speedfan.

I don't think I am going to try and push it any harder than that, I like where my temps are with this overclock, and it is just damn fast anyways.

 

[BonzaiDuck]

AdairUSMC --

I finally found the "Main Rig" link on your sig, which I overlooked last time I went there. I think you did this right. The mobo -- I THINK -- has all the features that I wanted with my Striker board.

According to Maximum PC, you may have gone "over the edge" with the 4GB of RAM unless your OS can really use it, but they did the same thing with earlier Dream Machines. I don't know what to believe anymore from that magazine, even so.

My Q6600 came via UPS earlier than the expected delivery date -- they must have left it on the doorstep yesterday evening while I was sitting down to the linguine and clam-sauce. Not enough sleep the night before and just a tad too much Pinot Grigio had me catching some Z's on the couch after the dinner, and it wasn't until maybe 2AM when I went outside that I discovered the package.

I may not get to the CPU swap until tonight or tomorrow. I need to finish running these tests on my ducting mod -- some posts will be there before lunch today.

 

[ultra laser]

I suggest you look into the Zalman 9700. I have the same board you plan on getting (the DS3R) and installation was a breeze. You simply screw on the backplate and bolt it down. Also, the Zalman comes pre-lapped to a mirror finish, which the Ultra 120 certainly does not. Another pro is that it's a bit lighter - only 765g - as opposed to the 1000+g ultra 120 and Tuniq tower. Hope this helps.

 

[BonzaiDuck]

Well, Ultra -- you have a couple valid points which I grant you -- with some caveats.

It's about preferences for taking time and trouble, about aesthetics, and trade-offs against performance.

On pure performance, the Ultra-120-Extreme trumps the Zalman 9700 significantly -- with or without lapping the base. The thermal resistance on the Ultra is just plain lower than that of the Zalman, and the review benchmark numbers at various OC settings bear it out. That may put some more at ease on the "time and trouble" versus performance trade-off.

On the aesthetics, the Zalman is prettier, and comes with its own fan. But with the right case, the right exhaust fan, and the right placement of an intake fan -- you don't need to hang a fan on the TR Ultra-120-Extreme, and you can still get the same load temperatures as might be ideally obtained by hanging a fan on the cooler!

Sadly, the better fans may have heavier motors, and a 120x38mm fan can weigh in at between 150+ grams and 250+ grams. A good Panaflo is in the quartile that tops out the 250+-gram-weight.

But, like all human progress, either TR or some other company will come up with an even more effective heatpipe cooler . . . . . Maybe. For now -- I got da time, I got da tools, and I got da Ultrah-120-Extreme . . . . .

[When God created the Universe, I bet he used a Dremel, a tap-and-die kit, and an Xacto knife. Just kidding. Just kidding. Forgive me, Lord . . . . ]

 

[SolMiester]

Bonzai, what stepping is your Q66?, I am having to bend the wifes arm and spoil her rotten for a least 3 months before I can go ahead with my build. I need new mobo,ram,cpu and video so it's gonna cost me.....Maybe by End of Oct there will only be G0 stepping around, and I wont have to worry too many about temps!.

You guys have it soooo good over there. Some really nice rigs....Sometime (just sometimes) I wish i was single again!!...<sigh>

 

[BonzaiDuck]

It's very wise that you are so selective, SolMeister.

I have the time now, and proceed slowly with many things, although this was certainly an impulsive choice. I think I explained the "trigger" for it.

So I hunted it down at the Intel site. It's an "SL9UM" -- stepping B3.

This is just my personal view on your "better half's discipline." I'm trying to guess when you built your AMD64 3500+.

I built my last system -- my "MOJO" -- beginning July '04. It was a socket-478 P4, and I started "planning" it in January '04. I went from a Northwood 2.4C to a 3.4E Prescott when the prices on the latter were so low that you had trouble finding an OEM-wrap for a while. I'm still tweaking it.

I started planning this C2D around March '06. I was watching the mobo offerings develop, and knew they'd release the Conroe. I just didn't figure on getting a C2Q so soon.

Five months of sheet-metal-work and painting, and expenses I never thought I'd incur. It's been about three years for me since the Northwood build.

In your case, I'd sure keep that Silverstone case, and hold on to the Audigy card. Flip the rest at Ebay when you can spring for your new parts. I guess my implied message is that this all goes in cycles between my geek friends and me. I think it's like surfing: you try and catch the wave of the technology at a balanced moment that trades off currency, "Currency," desires, needs, ambitions and objectives.

At least one of my friends has been about 180-degrees out of phase with my own "cycle," and has tried to catch too many waves. I don't want to be too specific about his flirtations with the technology, because he might be reading this. But -- he went through parts like a gorilla goes through bananas on a food binge . . .

 

[BonzaiDuck]

SolMeister -- They're selling the G0 stepping. I just thought it was a surge in quantity-demanded at the newer price that caused them to buck up the Q6600 price to $345 -- now about $319 -- after it dropped to around $298. Even so, as I understand, MSRP as of July 22 was lower even than that.

But I may have made the mistake here that my friend made when he bought his Presler, and then a new stepping was released that ran 25 Watts cooler.

Next paragraph for ADAIRUSMC:

Could you do me just a brief favor? You've apparently OC'd your system to an FSB setting of 1,333 or 1,334 and 3.0 Ghz. Could you set it back to stock values and run your double-ORTHOS for maybe a half hour? I'd like to get your CoreTemp load temperatures as a reference. You could either get a text-file log of it, or just report the maximum in the range for all cores.

 

[adairusmc]

Originally posted by: BonzaiDuck


Next paragraph for ADAIRUSMC:

Could you do me just a brief favor? You've apparently OC'd your system to an FSB setting of 1,333 or 1,334 and 3.0 Ghz. Could you set it back to stock values and run your double-ORTHOS for maybe a half hour? I'd like to get your CoreTemp load temperatures as a reference. You could either get a text-file log of it, or just report the maximum in the range for all cores.

No problem, here you go -

Overall - 48C

63,60,57,59

This is according to Everest Ultimate though, with two copies of Orthos running for about 35 minutes. I tried to load Coretemp, but I am running Vista x64, and all I got were about 100 error message windows saying the driver failed to load. Sorry about that.

 

[BonzaiDuck]

Great! I very much appreciate your help. I don't know if you were following my thread about motherboard ducting, but I almost came unraveled in my enthusiasm at the discovery that I could get the same speeds at lower VCORE on the E6600 -- and with lower temperatures -- for all the arts-and-crafts foam-board tedium.

Now I've got the Q6600 installed, up and running, and SolMeister has me kicking myself, because customer reviews at NewEgg show that they're selling the G0 stepping all of a sudden. I got the B3 -- maybe that's what you have also.

I can't see how Everest Ultimate would report inaccurate temperatures, but this is just a baseline check to see how I'm doing.

Here's the frequency distribution of my 2xORTHOS temperatures for 1 hour, 16 minutes:

Q6600 @ 2.4 Ghz, 1.28V VCORE, baseline temperatures under 2xORTHOS load and 73F room ambient

This is the entire distribution of my temperatures sampled every 8-seconds over that time-period. Your range upper-bound gives me some encouragement for all the little pieces of foam I have to sweep off the carpet now . . . .

I guess I can continue to put water-cooling on the back burner for a while.

Did you notice a wider variation in the load temperatures while you ran ORTHOS at the higher clock settings? I'm going to bump it up to 1,333 Mhz before dinner -- probably without even dropping the multiplier to see how that goes.