[Review] Indigo Xtreme vs. AS5, MX-2, IC Diamond, Shin-Etsu X23-7783D

[Vapor]

Vapor, another stellar review!

glad I use MX-2 for CPU block testing. The Indigo just boggles my mind, same with the liquid metal though.

Is the D5470 an industry standard for thermal resistence?

Again, well done on another set of lengthy tests and well written review!

Thanks

D5470 is the industry standard, I believe. Enerdyne (makers of Indigo Xtreme) and Innovation Cooling (makers of IC Diamond) both used it to test their TIMs and some of their competition (I got my D5470 data from both of their websites). The test measures both contact resistance and bulk resistance which, when combined, equal the overall resistance of a TIM at a specific bondline thickness (BLT). I really don't know the specifics of the test though

Wow. I'm very impressed with this. I can't believe how much time you invested in testing some TIM and this definitely sets the bar...

Really well done. I appreciate this very much. And this Indigo stuff is very impressive, I must say.

Thanks! I tried to think about and answer all the questions I had about TIM and use the tools available to me to test as best I could....I think it came out well Fortunately I could just leave the testbed running on its own and could get ~2 mounts done a day for ~2 weeks straight

I'm sure my CPU hates me....doing nearly 24hrs/day of Small FFT loads at nearly 1.5v with HT on is pretty rough on it

And I'm also really impressed by the Indigo...the performance of it really didn't set in until I came up with the last chart (which I didn't even think of until finalizing the review last night)....it's really, really close to being 'ideal.' On my testbed, it looks like there's about .3-.4C of headroom for TIM improvement. Just very foreign to think that some aspect of thermal performance is at it's limits....it's always been my mindset that "we'll always see something better around the corner" and that's just not physically possible any more--any improvements from a passive TIM (can there even be an active TIM??) will be immeasurably small on all but the most insane of heatloads.

I think my head just exploded. So many graphs!

Great review

Thanks I tried to keep the data vomiting to a minimum....I have ~35million data points in Excel spreadsheets here

Interesting endorsement of IX...do you know if other reviewers found the same results? I'd be interested to know if HS/mount affect temps at all. Would this be less pronounced for 775, or a different cooler?

I haven't seen any other reviews up yet....I think I'm first out of the gate (not sure what other sites are reviewing IX...but they approached us and made it sound like at least one other site was also getting it).

As for whether the HS effects it? TIM performance is dictated by heatload and contact...assuming contact remains constant (which I think is safe to say in this case), the difference between IX and the other TIMs scales linearly with heatload. There's small changes in heatload through the TIM (and into the cooler) depending on what actual temperatures are at (because of secondary losses through the board), but those are pretty minor difference considering the tight window of temperatures we keep our CPUs within. Basically, the cooler shouldn't make a difference aside from variations in contact (some are bowed, some are flat).

As for whether or not the socket effects it....that's actually a scenario where heatload and contact change. LGA775 IHSs are smaller, so the TIM performance is even more important at a given heatload. Granted, newer LGA775 chips are pretty miserly when it comes to power (and therefore heatload)....older chips like Kentsfield and early Conroes could show some really sweet gains with IX

Wow that's impressive!

Definately bang for the buck (Benefit/cost) winner compared to everything else.

I was always curious about curing timeframes, that's awesome work...thanks for sharing!!

I was really curious about curing times as well...I had seen early that if I did my waterblock flowrate vs. temp tests in a different order, the shape of the curve changed (so I did them all in the same order, shrug), so I figured MX-2 had some sort of curing time and confirmed it with these tests. AS5's curve is pretty insane--and it shows that AS5 losing by huge amounts (as it does in a few other TIM tests I've seen) is possible with really quick tests, despite being a solid option as a TIM for end-users.

Anyway, glad to share....this stuff is fun