The Dreaded Thermalright Bulge - (L)ap, (R)MA or (I)gnore?

[Voldenuit]

Hi all,

Just received my thermalright Ultima 90 heatsink in the mail from jab-tech.

Unfortunately (and what appears to occur very frequently with thermalright 'sinks on forums these days), the base of the heatsink is convex (see picture).



How badly is this going to affect the performance of the cooler? With IHS caps these days, does it even matter if the base is not flat? Will the convexity squeeze the IHS into a matching shape? Or is this cooler Fxxx'ed?

My options as I see them right now are to lap the dang thing (I have sandpaper, a mirror, and some mineral oil standing by). This will definitely wreck the nickel plating on the base, in addition to voiding my warranty.

Another option is to RMA it, which I'm not so keen on as it might take an interminable amount of time to actualy get back another unit, with no guarantee of getting a flat sample, anyway.

Lastly, if the performance degradation is negligible, I could just install the cooler as-is.

I'll be running an E8400 (on preorder), with a mild overclock target of 3.6 GHz. I'm hoping to be able to run a Scythe Kama 92mm PWM fan at low RPMs, wihch may not be possible if the heatsink isn't actually conducting heat away from the cpu properly.

Thanks for any advice!

Cheers,
V

[Michael Sandstrom]

I would install it as is. Your CPU temps will probably be OK. If temps are too high you can always lap and reinstall. I have an XP-90 that I lapped resulting in a slight increase in temps.

[Conroy]

Thermalright says that the convex base is intentional, to match up with the concave surface of the CPU. It seems unclear whether it's actually better that way or not..

For example, here's one link I found that quotes a Thermalright email
http://hardforum.com/archive/index.php/t-1239985.html

Thermalright has stated and continue to state that the design of the base is meant to be convex, because the heat spreaders on most Intel CPU's is concave, thus matching both at the most important point. The only other company that does this is Swiftech on a few of their CPU water blocks, but they also use a seal around the block that tends to hide this fact.

People are used to the old days when cooler bottoms had to be flat and polished because the application was directly on the core of the CPU, but that is no longer the case today, due to the use of heat spreaders, but most people resist change and still want it the way it was.

Personally, I got my ultima 90 lapped, but that was mostly because I have been too restless waiting for the e8400 and started thinking too much. I hope it doesn't cause problems

[seraphyn]

I'd say: Install the bugger and see how it goes, if it functions poorly, RMA it.

[skiddy]

If you care enough to overclock, why limit your oc and reliability by wimping out on lapping?

It will make a significant difference, and if you do it right, you'll see equivalent temps in all cores.

(If readers disagree that it will make a significant difference in OC [not true!], then I'd suggest that any OC is also insignificant.)

[Michael Sandstrom]

If you lap the HS before testing it's performance stock you will never know whether lapping had any effect. Sometimes lapping can be detrimental. After I lapped my XP-90 temps actually went up rather than down (a fact that I knew because I first installed the HS stock).

[Voldenuit]

After corresponding with jab-tech, I have decided to RMA the unit to exchange it for a (hopefully) better sample.

I still don't entirely buy the "convex by design" theory, as matching curvature is nigh impossible, and there's no guarantee that a mismatched curve is any better than a mismatched plane.

I haven't heard back from Thermalright, so I'll go ahead and return it to jab-tech. Fortunately, my experience with the staff there has been awesome - they've been very speedy, helpful and accomodating in the affair.