• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

P4 Northwood (1.6A?) Retail Thermal Tape vs. Arctic Silver 3 (AS3)

S

striker21

Member
I have been wondering how much of a difference exists in practical terms between AS3 and the thermal tape that comes with retail P4 Northwoods like the 1.6A. It seems like a lot of people have been getting really good overclocks with the thermal tape.

If you read the interview with Nevin from Arctic Silver here, he raises several interesting points that apply to the P4.

First, the P4 uses an on-die thermistor, which is crucial to getting accurate temperature measurements. I agree with Nevin and Mikewarrior2 that socket thermistors are not good enough for making accurate benchmarking comparisons--we shouldn't have that problem here, since I think all P4/motherboard combos use the on-die thermistor.

Second, it seems that you would have to think about the P4's integrated heat spreader. To some extent, this effectively increases the contact area of the heatsink with the CPU core. Nevin points out that as the contact area with the heatsink increases, for a given wattage dissipation, the thermal interface material becomes less of a bottleneck. Since the P4's integrated heat spreader is *much* larger than the exposed die on an Athlon, it seems like the thermal interface material would make less of a temperature difference on P4 systems.

With these things in mind, it seems possible to conduct a fairly scientific comparison of temperature differences when using AS3 vs. the retail thermal tape on a P4 Northwood. Lots of people have used one or the other in these forums, but very few have tried to compare both.

Of course, a proper comparison would include cleaning the CPU appropriately and lapping the heatsink between tests to remove any remaining thermal interface material. Also, the TIM should be given at least two days to "set" in order to acheive maximum thermal transfer--Nevin mentions this in the instructions for AS3, and I imagine the same opportunity should be given to the thermal tape for a fair comparison.

A lot of people are using the P4B266 with their Northwoods, and it sounds like an awesome board. However, I have seen some reports of questionable Asus Probe temp readings, so maybe it would make sense to use MBM5 for temp readings.

I think Flatbroke and I, among others, would really like to see a tech web site do this comparison the right way. Although I have a hypothesis that TIM won't make much difference for the P4, I would still rather be proven right or wrong! Maybe it could make a difference with a really extreme O/C, but I'd rather see the results of a quality test than to guess about it all day...
 
^bump

I want to know too, before I put mine together. Right now, I'm leaning towards just using the stock thermal pad for a modest 1.6a to 2.4Ghz OC, even though I already ordered the AS3.

Isn't AT a tech site?
 
From what I have read, most people notice around a 5 c difference under load with idle temps staying around the same.
 
I, like Manco, am curious about this. I have the AS3 and am not sure if I will use it or not. I did have a similar thread about it here that asks if I should use it. The responses seems to lean toward yes even though I will not be OCing that much. I will keep an eye on this to help me make my choice also. Good luck. 🙂
 
Umm there really is no debate here, Arctic Silver 3 murders the Thermal Pad. Northwoods may be new, but the Socket 478 P4's are not, why don't you ask someone with a willamete 478 if AS 3 is better or not they'll say HELL YEA.

To get an Acurate Reading of a Northwood CPU you MUST use MbM 5 and the Asus 2 (culs2) sensor.

I use Arctic Silver 3 on my Sunflower and my CPU is like an icebox here, real cool. I would never use a thermal pad (Not to mention it melts all over the board and everything else, making a mess when you change HSF's or wanna sell the board)

 
P4 Internal Diode As3 versus Thermal Pad tests.

A couple of key points:

a) They Failed to clean off the pad on the CPU... i suspect they didnt' clean off the pad on the heatsink, either.
b) Despite failure to properly clean the heatsink and CPU, they got an 8C temp drop between the two. This isn't a particularly hot P4 (1.7), nor is it a really fair to compare p4 thermal grease versus amd thermal grease(on an AMD chip, or similiar small contact surface chip, grease is much, much more critical) , but you can definately see that AS3 can and does make a huge temp difference.

c)note, the 8C drop may be higher, since the mb temp is 2C higher with the AS3 test. Which is why in good, accurate testing Ambient temp must be adjusted for and monitored at all times.

d)despite the massive contact surface, grease still plays a huge role on a p4... the role likely only increases on an AMD chip (and yes, i've seen the few diode tests with amd chips, and the overclockers.com one is highly erratic in results).

Mike
 
I just read Mikewarrior2's link to the AS3 test mentioned above. I agree wholeheartedly that the test was not carried out in a controlled manner. It does seem very promising that the tester saw an 8C drop in CPU temps without even properly cleaning/removing the old thermal pad! However, there are several things which make me question the authenticity of these results.

1) His temperatures are VERY high. In the AT threads here, most people don't seem to get out of the 40's with the stock thermal pad, much less to the 60's--even with significant overclocking! This guy reported load temps of 63C (with pad) and 55C (with AS3) in his tests. The only possible explanation I can think of is that he used a Williamette processor (which all 1.7's are). Even so, seems like a lot of heat! Several people mention this in the reader feedback that appears at the bottom of the article's first page. One guy even comments that his P4 1.7 is o/c'ed to 2.3, and he never sees temps over 100F (~38C). Of course, we don't know how the reader measured his temps, but I think you get the idea here...

2) It's hard to say whether he was overclocking or not, but I'm guessing he didn't. Otherwise, I figure he would have mentioned it. When not overclocking, the informal temps I've seen reported with the stock thermal pad on P4's seem to change very little when AS2/3 is applied. Certainly not 8C...

3) AT forum member DRS switched from the P4's stock thermal pad to AS3, and saw temps go up by 1C! DRS definitely tried to clean the old thermal pad off before applying AS3, which is more than I can say for the procedure in this review.

4) The reviewer says he also tried this on his Athlon CPU, and saw core temps drop 4-6C. Based on comments from Nevin and Mikewarrior2, using AS3 on the smaller core of the Athlon should make MORE of a difference in temps, not less. Of course, I realize that he was probably relying on an off-die socket thermistor reading, so this is probably unreliable info to start with...

DRS also said that core temps go up more quickly under load with the AS3. This seems counterintuitive to me, since I would think that core temps should go up more slowly (if anything) with the AS3 if it was doing a better job of moving the heat from the CPU to the heatsink.

I'm not sure how applicable the reviewer's results would be to the Northwood, since Northwood is built on a different process. I think a Northwood would probably run cooler at a given speed/voltage--does that mean that AS3 is less important for Northwoods, or that it's more important for Northwoods because keeping it cool will allow a much larger O/C? Oops, starting to speculate... hand-smack :-

Don't get me wrong, I would like to believe that AS3 makes a worthwhile difference with the Northwoods--however, I haven't really seen any reliable, conclusive numbers yet.

I guess that's why I'd like to see a controlled experiment with some AS3, thermal pads, Northwoods, MBM5, and overclocking!
 
I hope this doesn't sound to ignorant, but I'll ask anyway🙂 Do you actually put grease UNDER the heat spreader, with a needle or something? I just thought I had read that somewhere, but can't remember exactly.
 


<< I hope this doesn't sound to ignorant, but I'll ask anyway🙂 Do you actually put grease UNDER the heat spreader, with a needle or something? I just thought I had read that somewhere, but can't remember exactly. >>



If you are talking about a heat spreader similar to the ones in the new Celeron Tualatin the answer is no. The CPU's already come assemled with a thermal compound between the heat spreader and the core. Taking the heat spreader off and putting a new thermal compound will not only void any warranty and threaten the life of the CPU, but will also interfere with the heat sink / cpu core specified distance.
 
Striker21,

further regarding heatsink grease testing, there's a reason that ambient case temp monitoring must be maintained... even in a well ventilated case with a well ventilated room, the ambient air temp can change several C over the course of the day, and that is beyond control of the user/tester.... so a 1C "increase" in temp with an un-monitored 1C increase in ambient means a net "increase" of 0. Probably the hardest of cooling tests to perform, grease testing is time consuming and very, very tedious...

I'd test the pad versus as3/2, but I've already removed the pad.

Coudl you describe DRS' method for removing the pad? Maybe he used a cleaner that he forgot to clean off (for example, something like WD40 would need to be removed with some isopropyl alcohol, etc).



Mike
 
Well, I just removed the stock pad and used AS2....and temps under load dropped by 3C.

While removing the heatsink, I can't believed I ripped off the cpu from the socket while it was locked in and bent a pin because the cpu was stuck to the heatsink. At least I didn't cause permanent damage. 🙂
 


<< Well, I just removed the stock pad and used AS2....and temps under load dropped by 3C.

While removing the heatsink, I can't believed I ripped off the cpu from the socket while it was locked in and bent a pin because the cpu was stuck to the heatsink. At least I didn't cause permanent damage. 🙂 >>



OMG!! Does the thermal pad really stick that badly?? That makes me weary to take mine off and put AS on :Q
 
<<OMG!! Does the thermal pad really stick that badly?? That makes me weary to take mine off and put AS on>>

yeah....it was very tight. I think I should of taken it off right after I powered the machine off instead of waiting 2 hours later while it cooled and hardened. maybe the veterans around here has good tips on how to take the heatsink off safely. 🙂
 


<< further regarding heatsink grease testing, there's a reason that ambient case temp monitoring must be maintained... even in a well ventilated case with a well ventilated room, the ambient air temp can change several C over the course of the day, and that is beyond control of the user/tester.... so a 1C "increase" in temp with an un-monitored 1C increase in ambient means a net "increase" of 0. Probably the hardest of cooling tests to perform, grease testing is time consuming and very, very tedious... >>


Agreed--ambient temps are definitely a factor that should be accounted for when performing tests like these.


<< Coudl you describe DRS' method for removing the pad? Maybe he used a cleaner that he forgot to clean off (for example, something like WD40 would need to be removed with some isopropyl alcohol, etc). >>


I think he used carb cleaner and followed it up with alcohol. You can read about it in this thread.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top