Q6600 G0 HOT?

[terencek]

Originally posted by: BonzaiDuck
Hey, guys!

Terencek -- Woodbutcher gave me a heads-up to post here -- suggesting I might have something to offer.

I'm stunned that you can't get the idle temperature for a G0 quad-core below what you've shown. I wasn't able to read through all the other posts here -- I'm "keeping busy" -- but I took a look at the Cosmos case.

I don't think the Zalman 9500 stood up too well in the May 2007 Anandtech comparison review focusing on the ThermalRight Ultra-120-Extreme, and later -- with the Ultima 90 review. But your temperatures should still be well below what you've shown.

When I first saw the Cosmos several months ago, I thought it was promising, but I took a closer look.

It's just my opinion, but I see one rear exhaust fan 120mm and two top fans (which could only be exhaust, I would think, the way they shipped that case.) That leaves only a single intake fan at the bottom.

There are myriad opinions about this, but I stand firm from a few years of case-modding and air-cooling experience: my recommended strategy is to overwhelm the case interior with exhaust CFMs and pressure, seal the rest of the case, duct the air over the hot components and force it out through one or more exhaust fans. This case doesn't do that.

I'm still running a B3 Q6600. I've got it OC'd to 3.204 Ghz at the 9 multiplier and a Vcore of 1.4125V. The TCASE temperature at today's current room ambient of 70F is about 25C, and my idle TJUNCTION Core temperatures are 37,36,40,46 (Everest seems to load up Core#0 (46C) with all its sensor reading for the fans, voltages, temperatures, etc.)

I still have some ducting refinements to my case, and I've added a second intake fan BEHIND my hard disks -- that lowered my RAID5 disk temperatures by another 5C across the board.

You could keep the Zalman, but you could do better with this. IC Diamond thermal paste is $5 -- that's worth about 2C or so. You can lap your processor and heatsink base. Using ducts in that case wouldn't be terribly complicated, but I'd like to see more intake air coming into it -- a lot more.

If you don't want to mod that case, at least to figure out how to add one extra intake fan -- maybe even block off and remove the top fans -- you really might want to consider water-cooling.

On the UP side, if you can keep your core temperatures below 70C at load, the G0 is still immensely over-clockable. You have to decide what you're going to do with that case.

For instance, I've taken to adding bottom intake fans (I know you have ONE there already), and using double-wheeled casters to lift the case off the floor for air intake. An extra fan on the bottom might do it. But I just think the Cosmos has too many exhaust fans -- unless I'm mistaken and the top fans are really intake fans.

Well, I am glad that you posted this!! On Sunday night, I was thinking that there wasn't enough intake as well and I installed a spare 120mm fan that I had. There wasn't a place to install it under my DVD-RW drive so what I did was used zip ties and suspended it under the drive blowing to my CPU. Now what I may do from wood's suggestion would be to duct it straight to the CPU. That is what you were suggesting, right wood? I also have a 80mm fan that is mounted on the wall next to the HDDs blowing on the Vid Card.

No, the top fans ARE exhaust. I am looking to Water Cool soon, just need to get the $$. I just dropped a lot of money to get the 8800, Q6600, PSU, and the mobo, so I am a little drained, but there's always Christmas bonuses!.

As it stands, I have the CPU OC'd to 3.0GHz with vCore at 1.200v. The ambient temps lately because of the nice cold front that has moved in (I live in Tampa Bay Area, FL) are around 72-75F. The idle is 36, 38, 40, 42 and when I run Prime95 or OCCT I'm at 55, 57, 58, 60. Now, that's with the side of the case open. Everytime I close the case, the temps are up about 2-3C, so I generally leave it off. I'll try ducting that new fan that I put in and post back with the results.

Thanks again for the intel!!

 

[BonzaiDuck]

As it stands, I have the CPU OC'd to 3.0GHz with vCore at 1.200v. The ambient temps lately because of the nice cold front that has moved in (I live in Tampa Bay Area, FL) are around 72-75F. The idle is 36, 38, 40, 42 and when I run Prime95 or OCCT I'm at 55, 57, 58, 60. Now, that's with the side of the case open. Everytime I close the case, the temps are up about 2-3C, so I generally leave it off. I'll try ducting that new fan that I put in and post back with the results.

Really -- that's not too bad. If you're stressing all four cores with ambients 72 to 75F, thoses temperatures are about right. Adding 3C to those is still about right.

[Keep in mind I live in So-Cal and we try and keep the AC bill down.] This summer, we had some marathon threads going with people who'd bought the B3 stepping of the Q6600. We were trying to measure room ambients to within 1F degree, and there were various reports of temperatures around 75F ambient that were just under and a few C-degrees over 70C among the cores. With ducting, I was showing about 64C to 65C at room ambients of 78C and my over-clock at 3.2Ghz/1.41Vcore. Since I started using the Chassis_FAN plug on the mobo to control my CPU exhaust fan, my highest load temperature has been about 3C higher than it was -- normalized for room ambient.

If you're running at 75F +- 1F-degree and showing a closed-case high-core value of 63C, that's pretty good. Take a look at your Intel retail box and find the VCore "Maximum" -- with the B3-stepping, it is 1.35V.

See, I think you could probably bump up the VCORE even a 10th of a volt and still be within the retail warranty range -- even if you invalidate the warranty. They determine those "maximum" specs to reduce the probability of warranty returns to zero -- or near zero. Imagine the probability distribution of CPU failure with the lowest tail of the distribution at the retail-box maximum VCORE. If you go even 5 to 7% above that maximum, the probability of failure is likely still low, so if you stay just within that maximum, you're only worry is heat.

I think you'd get some improvements with the mods I and WoodButcher suggested -- ducting gets you some more, and diamond thermal paste -- as I said -- is worth 2C. If you go after every last grain of rice [a Japanese QC expression] -- I'm betting you could take the clock to 3.2 Ghz or as far as 3.4 and still be well within the safety range of the Q6600's thermal limit for the TJunction core temperatures. You might even show a net decrease overall in CPU and VGA temps if you take pains with those "last grains of rice," even with a net increase in VCore and bus-speed.

If you replace the Zalman with either the Ultima 90, the Ultra-120-Extreme, or the Sunbeam Tuniq -- that's quite a handful of rice-grains -- alone.

 

[WoodButcher]

Damn Duck, all this talk of rice and no gravy!
Yes tk, I meant to duct through to the cpu if you can, this will prevent those top fans from exhausting all that lovely room temp air before it does any work. I was thinking the clear file folders with some clear packaging tape would do the trick and not look to "butchered", pardon the pun but I can be a real hack when I'm tired.

 

[BonzaiDuck]

I really need to write up my "project" to build the "Chrome Lightning" from the old '94 ProLiant Server case. The drive cage that came with the ProLiant was huge, and I modded two Lian-Li hard-drive "expanders" -- an EX-23A and EX-34A -- allowing a total of 7 drives in the ProLiant cage which originally only had four HD trays. These things slide in and out of the heavy-metal ProLiant cage on rails, using two of the original trays for the large SCSI drives equipping the old Compaq server. I have pictures, but I'm in a bit of a hurry now.

The original Compaq setup had intake fans at the rear of the cage -- pulling air across the hard-disks in front of them. So I used some PCI fan-brackets to put oversized 140mm Sharkoon fans behind the ProLiant cage. They are currently pulling a mixture of internal air and air drawn through the cage, so ducting those fans to draw air exclusively from the cage will make a bigger improvement.

But -- to the point -- I'm a long-winded blowhard. You mentioned using plastic file-folders or something to cover the unwanted top-side blow-holes in the Cosmos case.

When I added the second fan to the rear of the drive cage, I wanted to keep the number of fans constant in the system, but I didn't want to remove a fan for the new one I installed.

So I unplugged the Sharkoon in the lower case front, cut a Lexan square with holes to fit my silicon-rubber fan-mounts on the chassis exterior, and just popped the square over the fan hole -- the fan still installed behind it for some future experiment or unforeseen usage.

 

[aigomorla]

Heres a teaser what high end water can pull.

Ambients Cold! 16C
3.73 ghz 100% Loaded Temps doing her WCG run:

http://i125.photobucket.com/al.../aigomorla/Temps-2.jpg





mmmm.... those like like idle air temps.

 

[BonzaiDuck]

I won't argue with that at all, but only to say that 1.45V Vcore is just a tad more than I'd care to run. Remember -- no matter how close to room-ambient you can pull those load temperatures, you still increase the risk of "electron-migration" over the long-haul when you bump up the Vcore.

Still -- if the G0 still has the retail-maximum of the B3 at 1.35V, your voltage increase is less than 10% above that figure.

I'll start shopping for WC-parts when I know the tenant in my house back-East has caught up on the rent owed me. Freaking lease agreement is supposed to be a "contract," and people are not supposed to break contracts. The worst possibility -- I may have to evict -- and that will delay my desire for spending dough on water-cooling.

I suppose if I really want to do it right, I should set aside five Franklins for it . . .

 

[ArchAngel777]

I don't think increasing Vcore is much of a problem. If you look at the specification for the max a C2D can run, it states from .95 to 1.5 (IIRC) and then factor in a 5% margin of error and you easily have safe running at 1.575. And this isn't to argue or suggest that running high Vcore is good, I just don't believe it is going to kill your chip within the average 2-3 years that you will be using it... Heck, I don't doubt it will last several years beyond that.

Additionally, with the use of CEIST and Speed Step, your Vcore will never see those values unless you run 100% load all day and keep your PC on. I happen to shut mine off whenever I am not using it and run 1.56 BIOS (1.47 under load) and 1.38 in windows, this is at 3.6Ghz. Temps are very cool. (water).

Even more additional is that Tom's Hardware ammended their thought that 1.5v was the max for a C2D G0 stepping in their E6750 review... They decided that 1.6v was safe, because Dell, a major OEM was selling 3-5 year warranteeded systems at that setting. Now, one could argue that a major OEM doing that is taking a risk OR they just happened to know a lot more than the common enthuesiest in regards to what it safe for Vcore. Having close ties with Intel would give them far greater access to the truth of what these chips can handle and their average chip life. Remember, damage happens to the circuits even at stock setting (called wear and tear) so your chip is constantly degrading even at stock voltage and speed, just at such a slow rate that it realisticly is never an issue within 20+ years. But, who here in THIS thread keeps their CPU for more than 3 Years? Maybe one or two.. how about 5? Maybe 1... How about 10? No one. What is the value of your chip after 2 years? It is pocket change... I say, overclock, get the performance now, and if it happens to fail, replace it with a like kind for dirt cheep, or a better chip and have at it.