AS5 vs. toothpaste vs. mustard vs. ketchup vs. Turnip rape oil

[Calmatory]

I did some fast testing today. I have heard myths about someone who had to use toothpaste instead of aftermarket thermal compound, and he claimed toothpaste works better. I wanted to test this and go a bit beyond.

Testing rig in the sig, except that CPU was lapped and running completely @ stock with lapped stock HSF.
Temperatures were measured with the S&M CPU Stresstesting tool, which can be found from here. Testing was made with S&M aswell. Ran L1 cache, L2 cache, CPU Integer and CPU FPU tests, all with 100% CPU load and "short". I took the maximum temperature from the run as measurement. Ambient was around +22C.



All feedback is more than welcome. I am eager to test some more stuff between the CPU and the heatsink. Any ideas?

I have to mention that ketchup/mustard/toothpaste can't be used 24/7, as it already dried out in 10 minutes of running. It was real pain in the ass to clean it up.

[Nanometer]

you can use the ketchup and mustard for short benching sessions

[Scubar]

Ketchup seems to work well

how about some brown sauce, or you could try some yoghurt.

[DAK1640]

Mustard looses it's coefficient after time

[Calmatory]

I will be testing around with different kinds of oils in few weeks, I think.

The variation between compounds (is that a right word?) might be possible to explain with the thickness of the layer. When the layer thickness halves, the thermal conductivity doubles, AFAIK. So, there is very small variation, and it might be caused by the thichness of the layer, at least it surely affects it.

Thats why I want to redo the test in nearby days, maybe even tomorrow.

The drying compounds can be mixed with some oil to make them last longer, though thermal conductivity is probably going to suffer. But it might be that with oil it is easier to apply and remove them. They are quite sticky, especially toothpaste. And the smell after mustard-run... Just don't try that at home.

I will be testing with overclocked CPU next time, so variation should be a tad bigger.

Edit: Completely forgot.. The CPU and HSF are lapped.

[zsamz_]

was it heinz ketchup?

[hersounds]

very nice ," va como loco! "

[RPGWiZaRD]

OMG, I'm a big ketchup consumer myself (eat it with almost anything) but now I like it even more.

Anyways interesting test results.

[it_burns_when_i_pee]

interesting but funny

how would the ketchup hold though? its like over 40% carbs - fruit sugar and added sugar. yes tomatoes are fruit!

that sugar heating up for a length of time or repeatadly over a few days surely wouldnt last long

what about testing it with deepheat and then that menthol cooling stuff you can get?

[Bbq]

Don't forget water. Dansdata did one a long time ago with water and vegemite.

[NickS]

Vaseline?

[STEvil]

ketchup will corrode the IHS and heatsink.

Best results i've achieved so far are with nothing or extremely thin oil. This requires lapping of the IHS (both sides) and heatsink or you may as well go with a bowed base and whatever's handy.

[p0tter]

kY @!

[ChaosMinionX]

why dont we just fuse the HSF to the IHS with molten copper

[Omastar]

What the hell is turnip rape oil?

[Calmatory]

What the hell is turnip rape oil?

How should I know? :P Rather hard to explain, so I don't even bother trying.

Yeah, the biggest "Bottleneck" is the back of the IHS. I could try to remove IHS off completely and run the tests naked, which would make bigger variation between compounds used, at least I think so.

Will be redoing some tests today, also going to test some "regular ketchup" vs. the one you get from McDonalds.

[YukonTrooper]

Try Vaseline... that sounds like a good one. It seems like it would have longevity as well. Try some Aloe Vera to and various hand creams.

[Omastar]

Try Vaseline... that sounds like a good one. It seems like it would have longevity as well. Try some Aloe Vera to and various hand creams.

Are we talking about thermal compounds or something else?

[hipro5]

HA, HA, HA.......You crazy dudes.....

Yeep......Ketchup seems doing well.....

[MikeB12]

What the hell is turnip rape oil?

LOL!

This whole thread is hilarious!


here's some more:

best used during flu season-vapor heat


The best for your stick.

[metro.cl]

i once used blisteX and Toothpaste but didnt measure temps

[YukonTrooper]

Are we talking about thermal compounds or something else?

Haha, looks like I nailed two birds with one stone! I didn't even realize until after you pointed it out. Haha, that's hilarious.

[Calmatory]

Today I will test with shampoo and some shower gel. I will repeat the testing with ketchup, I can not believe the results even by myself!

Too bad I ran out of AS5, can't include it in the tests. Also, current AS5 measurement ain't valid later on, since I will use new measuring method.

I found out that I took the temps from the core#0 only. So, now I will measure both cores MAX. temp in the S&M run (L1, L2, CPU Int, CPU FPU and 100% CPU load and test lenght set to "short".) and take the average value of them.

Also, the right way to apply the compound: Don't just put a small tip of compound to the middle of IHS and put HSF on, at least not with sticky compounds. Instead, try to make as thin and even layer of compound which covers the whole IHS. I have heard the first method being used alot, which I found out not working so well. The layer is too thick. I will test this also, with ketchup at least.

So stay tuned.

[naokaji]

i have to say i'm surprised about the relatively small difference...
but it would be interesting though how it would turn out if you would use it over a longer time... like try it on a oc'd quadcore and then have it crunch for a week...

[Calmatory]

Well, it is quite obvious that water-based compounds will dry out in minutes. Water has relatively bad thermal conductivity, thus the conductivity might possibly even increase after the "burn-in". Though, I don't know what will happen when the stuff starts molding(or is it moltng?). Also as STEvil said, corrosion will be a problem. Not for me though, as I can relap the CPU and HSF when needed, and they won't be used for 24/7 anyway.

I think I will run some GPU testing aswell with my GeForce (OC'd of course. ). 3DMark06 run should take the GPU to the max.