fav thermal paste?

Now whats the difference between what I said and the Wiki stuff you posted? Sucking the heatsink against the cpu as tight as possible? What does that mean? Same thing just explained in a simple way without getting highly technical. Simple

 

This One http://www.aliexpress.com/store/314...PU-Heatsink-thermal-Compound-paste-HY810.html

 

I used to use AS because it was the only one that you could find at most places. As years went on, lots of new stuff came out.

Using a "good" paste verses a "generic" paste has always given me lower temps. Talking about athlon xp/64/fx era stuff, and stock hs paste vs. Artic Silver. Things have changed a lot since then though.

I have been using the paste that comes with my thermalright coolers. Works well enough for me. I have always put a very thin layer on the cpu, but recently I got a mobo RMA from intel for my 2600k. They said to put the paste (in a small tube they provided) on the middle of the cpu and then put the cooler on. I decided to try that. Strangely enough, whether from thier "generic" paste or the method/amount of paste, the temps are actually a few degrees cooler than they were using the "better ?" paste.

Go figure. I have always taken the approach that the cooler electronics are for long term, the longer they will last.

Sul.

 

by using what some call the pea size drop the cooler spreads the paste for you to fill in the gaps evenly. the only times i spread paste are if i have a exposed die like a gpu or laptop cpu or if im using a heatpipe on die type cooler where it has exposed heatpipes that touch the cpu directly. i have had better results with those types of heatsinks that way or by using a "line" method where you run a small thin line between the pipes

 

First, I didn't post wiki stuff and second, you said, "Thermal Paste is to suck the heatsink against the cpu as tight as possible". That is incorrect. The purpose of TIM is to fill the microscopic pits and valleys only. Holding the heatsink in place is the function of the clamping mechanism, not the TIM. While it is true, TIM may cause a bond that makes removal difficult, that is not its purpose (except for adhesive TIM, which is not used on CPUs).

As far as TIM drying out, we are not talking about plain ol' white silicone grease from 50+ years ago. TIM used on CPUs are based on advanced technologies - this includes the OEM pads provided by Intel, AMD and many aftermarket cooler makers. And it is the TIM that remains between the mating surfaces that matters, not what has oozed out under pressure, or was "pumped out" due to expansion/contraction cycles. And regardless, it is direct metal-to-metal contact that provides the best transfer. And for sure, HS and CPU manufacturing techniques are pretty darn good at making flat surfaces.

And of course, I am talking about a properly applied layer of TIM, not big globs, and not too little either. And I am not talking about forever, but rather the expected lifespan of the hardware.

1. If TIM wore or dried out and lost its effectiveness over time, the TIM makers would be recommending replacing it so they could sell more. Or they would be using the longevity of their products as marketing selling points. And neither is happening.
   
   
2. If TIM wore or dried out and lost its effectiveness over time, Intel and AMD would be telling us to replace it periodically. And that is not happening either.
   
   
3. Note the following from AMD (my bold added):

For those insisting TIM must be replaced periodically, PLEASE show us a white paper or MSDS or some report from a TIM maker or hardware manufacturer that says TIM degradation for TIMs used today is such that it requires periodic replacement.

So again I contend (1) if properly applied and (2) if the bond is not broken from rough handling or HSF removal, the TIM is fine - even after years of service.

No doubt there is a great deal of truth to that. But that does not mean a CPU running at 30°C will last longer than a CPU at 40°C - not if 40° is still within its normal operating range - and it is. So is 50°C or even a bit higher. The extremes either way are not good. And you cannot just look at individual components when dealing with entire systems.

When you lower temps in a conductor, you increase resistance. Ohm's Law says if you lower resistance, to maintain a specific voltage you must then increase the current. So this then places a greater demand on other circuits and other components supplying and regulating that voltage. So while your CPU may last a little longer, you risk decreasing the life of other motherboard components, for example. So while cooler is better than hotter, in the middle of "normal" is best, when talking about whole systems.

Motherboard designers, for example, cluster many heat generating/sensitive devices around the CPU socket so they too can take advantage of the air turbulence created by the OEM CPU fan. I have seen too many cases where newbie overclockers replace the OEM HSF with alternative cooling devices and neglect the surround devices/case cooling, then wonder why their motherboards failed prematurely.

 

Not to beat a dead horse... ...but in the interest of getting some supporting evidence instead of expecting you to take my (or AMD's) word or it, on Nov 27th, I sent the following email to Arctic Silver tech support:

Today, I received the following reply:

 

Hmm. I had not thought about it like that. I realize how Ohms law would come into play like that now that you bring that up. Just not something I had ever thought about before.

I wonder where the threshold really is. As an example, I flashed my 8800gtx bios with my own fan algorithm, so that the temps stayed around 70c at full load and fan @ 100%. The fan algorithm would not even move above its default 60% until temps reached 90c, which I did not like. Two of my buds bought the exact same card at the same time. Realizing every card is different things might be a little off, but mine is still going to this day while thiers were RMAd about a year 1/2 ago. Their temps (presumably within the defined acceptible parameters) were average 20c higher than mine, yet both of thiers started artifacting and were replaced. So, I must have "accidentily" found the correct threshold perhaps.

A very interesting tidbit you bring up, one I am sure to give more thought to in the future.

Thanks.

Sul.

 

Thai Kitchen Green Curry Paste


.

 

Yeah, but the exact same motherboard? Exact same CPU (which can determine FSB speeds)? Exact same PSU? Exact same case and fan setup? There are many factors that come into play. Even ambient (room) temps.

 

You can't go wrong with anything from EVGA:

EVGA Frostbite Thermal Grease

http://www.evga.com/products/moreInfo.asp?pn=M019-00-000003&family=Accessories - Hardware&sw=4

 

I recently replaced a motherboard and needed some thermal paste. My Dad had a tube of Dow Corning 340. I was a little bit skeptical about it since all I have ever heard is Arctic Silver is the best. I tried it and so far I have been impressed with it. The temps are great with it. I even used it on my own computer last week. I had to have a new motherboard because the southbridge was starting to fail. I used the same Dow Corning 340 and got the same results. It is good stuff imo.

 

hahaha thanks for the chuckle

anyway, it really does not matter which paste you go with as there is only a couple degrees difference between the bunch.

what DOES matter is how you apply the paste and heat sink. (i build computers for a living)

 

That's exactly right! If you do see a big before and after difference, then the odds are the "before" was not properly applied. And if you really need those extra few degrees to stay out of the danger zone, then you need to be looking at case cooling, your voltages, or other factors that may be causing your temps to be too high in the first place.

What DOES matter is first, you use TIM, then second, you apply it properly. I am often surprised to see no TIM at all on some computer that come across my bench. And sadly, there are some who feel that more is better.

The 3 most common TIM mistakes we see in the shop is:

(1) No TIM,
(2) Too much TIM,
(3) Re-used TIM/poorly cleaned mating surfaces. ​

Having said all of that, if you are an extreme overclocker or an enthusiast just seeking cooling bragging rights, there is nothing wrong with trying to eke out a couple extra degrees in cooling, as long as you remember there's more to cooling than just CPU temps.

For fun see 5GHz - Liquid Nitrogen Cooling Project (note this was from 2003).

 

agreed i cant remember how many times i have seen NO THERMAL PASTE!!! lol...and i see people use the ENTIRE tube also i dont know what they are thinking with all that mess on there..

proper application is def VERY important but it does slightly vary today depending on the type of sink and the type of surface you are dealing with.. an example is a heat pipe direct contact type of sink. the dot or pea size method imo sucks for this heatsink.. i use a line method and i also spread a pre application to fill the voids between the heatpipes and the base then run 1-2 lines and mount the sink. i find this MUCH better than a dot with these. and with a exposed die i ALWAYS spread it.. very thin even coat. a normal cpu i use a dot and mount the sink... etc. there are many ways but i find these work best for me.

and i personally overclock with things like dry ice, liquid nit.... etc.. i have even used ac r122 once and have tried all kinds of things like taking apart a ac unit and making a cooling chamber among anything else you can think of lol..i used to compete in the oc'ing competitions they have but i just dont have the time anymore though i wish i did.

 

Yes it very good, what they use here at work for all heat sink devices - I usually 'pinch' a tube when needed on my out of work.

 

If he had read Electrical Resistance and Conductance on Wikipedia, he wouldn't have stated:

Electrical Resistance and Conductance - Wikipedia

I would rather it be in GangGuy than on CPU.
Have you tried Nam Pric Platoo? Very spicy.

Recently, my CPU had been having temps at 44C-45C. Looking in the case the fins of the heatsink were clogged with dust. Because the fan is mounted to the fins, I couldn't just remove the fan, but had to remove the heat sink to clean it. Replaced it with the existing paste, wiggling it to rebond it. Now my temps are 34C-35C. The existing TIM is 3 years old, was not dried out, and came with the CPU.

If your are Overclocking then keeping temps at 30C-40C is great, which is why people go with water cooling, to keep it near that range. GPU's are more sensitive to being, and more likely to be, Overclocked. GPU Overclocker's should water cool before OC'ing, not all of us are capable of rewriting the fan control scripts.

 

Geez, first I am accused of using Wikipedia, then I am blasted for not using it.

That said, forgive me. I did err! I used the word "conductor" when I should have said "semiconductor". I note when put in context of the paragraph you pulled that one sentence out of, semiconductors (IE; the CPU) was the primary subject of the whole statement! But nevertheless, my apologies for messing up one word. And thank you for bringing it to my attention.

So, without taking my sentence out of context of the whole paragraph it was in, I re-state my whole comment/paragraph again, correcting my apparently horrendous one-word error:

When you lower temps in a semiconductor, you increase resistance. Ohm's Law says if you lower resistance, to maintain a specific voltage you must then increase the current. So this then places a greater demand on other circuits and other components supplying and regulating that voltage. So while your CPU may last a little longer, you risk decreasing the life of other motherboard components, for example. So while cooler is better than hotter, in the middle of "normal" is best, when talking about whole systems.​

***

Ha! Yeah we did that once. I got a shipment from Omaha Steaks which typically come in a Styrofoam cooler with a big block of dry ice. We laid the computer on its side (like a desktop) and removed the CPU fan, leaving the heatsink in place, then placed a nice chunk of dry ice on the heatsink. We were able to push a 3.06GHz P4HT to 4.2GHz before a cap "popped" in regulator circuit and shut the whole motherboard down. Of course the problem with dry ice is it evaporates rather quickly. And it can burn your fingers with frostbite in no time. But it is is fun to play with.

 

lol....we make actual tubes for the cpu's and insulate the mobo etc...but yeah i have done almost anything you can think of before doing it the "right" way if there is such a thing =)