Intel 980x stock tower heatsink and gummy compound

[spworley]

My new 980x is working well. It's notable that the 980x comes with a "Real" heatsink.. a quite nicely engineered tower. I didn't use it (I use a Nocua NH-D14) but it really is a change. It'd be interesting to see a performance comparison with Intel's new heatsink versus all the current modern towers.

I did notice the Intel heatsink compound that came with the cooler is nicely packaged in a syringe, clearly labeled, and has terrible instructions (basically "empty the entire syringe (0.5grams) onto the top of your CPU then place the cooler on top.") I also noticed that the compound is extremely rubbery, more like a gum than a paste. It made it very challenging to spread it evenly over the heatspreader. My usual method of using a business card to get a thin even layer was quite challenging and unsatisfying.

You probably ask "why did you use the Intel compound instead of Arctic Silver or the compound that came with the Noctua".. and that's my own question too. I used it because it was convenient but now I wonder if that was a mistake.

My 980x is running at 4.1 GHz on air, the CPU reports 70C during Prime95 load, but the heatsink is barely warm to the touch.. is it because the heatsink is awesome (it is!) or because the 980x is able to deal with the heat even with a badly seating compound?

My 4.1GHz speed matches well with all the 980x overclock results.. all of the sites seemed to get between 4.0 and 4.3 Ghz.

Anyone experience the Intel compound? Or have an aftermarket cooler on a 980x? What are your temps like and is your heatsink warm?

[axee]

Since I don't think there's alot of people with that kickass CPU (with kickass price!), I think YOU are one of a few people who should test that heatsink

[ces]

Why don't you send your extra cooler to MikeC and ask him to give it a test. It aught to be interesting. It will help SPCR traffic as well. An review on that would draw all kinds of traffic that would never otherwise come and get a taste of SPRC.

[Fayd]

every time i see a reference to comparisons of thermal compounds, i keep thinking of this.

[cordis]

I have a scythe mugen 2 on my 980x, I think the temperature is around 65, but I'm not sure, still haven't gotten the temperature measurement working in linux. I haven't overclocked it yet, so that's stock speed. But it definitely is not too hot to touch, although it's probably hotter closer to the chip. The intel cooler does look interesting, but I didn't even try it. I'd be happy to send it up, although it would have to wait until after I finish my taxes.

[kittle]

Definately keep us updated on what you find out.

Ive got my sights on a 980x system as well (compliments of doing my taxes), and Id like to know which HSF cools well and quietly.

[ces]

every time i see a reference to comparisons of thermal compounds, i keep thinking of this.

Nice article.

[ces]

I have a scythe mugen 2 on my 980x, I think the temperature is around 65, (

That seems pretty hot for such a good cooler on a chip running at stock speeds. Or is that fully stressed?

[cordis]

It's fully stressed, running folding 24/7. Edited to add, ok, I checked out some of the temps that were being reported, and one seems to correspond to cpu temp. Currently, again running at full speed, it's being reported at around 51 C. I'm not sure how reliable that is, I'll see if I can do some benchmarking to bios data.

[ces]

It's fully stressed, running folding 24/7.

cordis

Why don't you just get the Noctua D14.

Run it with the stock fans until MikeC gets done with their next round of fan reports. Then (a) determine how its doing on stock fans, and (b) assess whether to make changes in the stock fans.

Have you looked into the 2000rpm Thermalright fans? I don't think they are quiet, but together with the D14 I bet they would make a big difference.

spworley

Why don't you try 3 apache PWM fans. the D14 is supposed to respond well as the fan speed goes up. They handle high impedance well. And if you are not pushing a load, they should be reasonably quiet at 500rpm.

I believe that all cpu cooler feel cool to the touch, even with hot cpus. The problem isn't moving heat from the fins to the air, the biggest barrier is getting the heat to the fins.

Also consider ducting in cold external air right to the cooler. That might be the one thing that will produce the greatest result. No matter how fast you push the air past the fins, if it is warmed over air, it can only do so much.

Better to get some external air that is 10C cooler.

[cordis]

Well, I guess I could get a D14, but if the cpu is really around 51 C, then do I need a new cooler? I suppose I could OC it up and see how hot I can get it, but I have a few other cpus I want to OC first, and I'm a little worried about the affect that running it faster will have on my ppd/watt ratio. I used to have a Core 2 Quad that ran in the high 70s, it was in a small micro-atx case with a low-profile heat sink. I worried about it a little, but it seemed to run fine. I'm not sure when to worry about how hot a cpu is, but 51 C under full load doesn't seem too worrisome. Even spworley up at 70C probably isn't a big problem, as long as the readings are reliable. That's all I'm worried about, making sure my temp report is accurate. If so, then I may as well try to OC the crap out of mine and see what happens. Maybe I'll set the bios temperature warning low. Whatever I do, I don't see a big need to get a new cooler anytime soon.

[mentawl]

cordis - considering that's the CPU running at a decent load, 51c is absolutely fine and you have no need of a new cooler whatsoever unless you want to start getting silly with voltages. IMO anyways.

spworly - I've had my i7 920 up at near 90c on the core sensors during stress testing (I just can't get it to 4ghz with a reasonable amount of voltage), and at no time did my Scythe Kabuto heatsink feel anything other than lukewarm to the touch - even the little heatsink directly on top of the CPU wasn't particularly warm. Thermal compound type and application rarely makes more than 5c difference to load temps on a CPU, unless it's very poorly applied or the CPU or heatsink are mating very unevenly. You'll probably find that the extremely rubbery thermal compound softens very quickly when heat is applied, so I wouldn't worry about replacing it.

[ces]

but 51 C under full load doesn't seem too worrisome.

For some reason I thought you were running hotter than that.

[psyopper]

@cordis: what did you use to extract your temps under Linux?

[cordis]

Well, it's not too clear, I have lm-sensors running, and one of the temps (the anonymous 'temp2') seems to track cpu usage pretty well, so I'm going under the assumption that that's the cpu temp. For all the other cpus, it usually gives a cpu temp and temps for each core. It doesn't do that for the 980x, I probably need to update something, possibly the kernel. I don't really want to update the kernel, so I'll see if I can correspond it to the bios temps or something. Eh, I'll figure it out at some point.

Oh, and ces, sorry, I made an edit to my last message with the new temp, must've missed it.

[MikeC]

every time i see a reference to comparisons of thermal compounds, i keep thinking of this.

It's a perfect piece, a great counterpoint to the breathless TIM comparison reviews on so many tech sites.

For the sake of those who have not run across Dan before, here's the actual URL -- http://www.dansdata.com/goop.htm -- and let me just say it's often a great pleasure to read his stuff, he's about as good as you can get writing about technical stuff to a broad audience. And did I mention, absolutely hilarious too. A classic Aussie in some ways.

[ilovejedd]

Lol, Mike's quadruple-posting.