• Sign in to Neowin Faster!

    Create an account on Neowin to contribute and support the site.

Sign in to follow this  

Worth reapplying CPU thermal paste, or not?

Recommended Posts

ThaCrip    727

I recently went from a i3-2120 to a used i5-3550 CPU (for only $20 which was a great upgrade for a minimal price) and I am using the stock i3 heatsink/fan I was using and used my Arctic Silver 5 thermal paste (which I had since March 2006) and noticed, at least shortly after installation, while idle temps where normal (call it roughly 35-40c), that on Prime95 (SmallestFFT) that my CPU temp reached about 95-96c then after some time passed I ran the Prime95 test again and temps came down to about 90-94c but mainly 91-92c and then on June 24th I was getting 84-86c most of the time but it did not 87-88c briefly. but I read that disabling AVX on Prime95 helps it so Prime95 represents a more 'real-world' 100% CPU test and doing that earlier I got about 76-81c.

 

I used these for some general info... https://forums.tomshardware.com/threads/i5-3570k-safe-temperatures.2093736/#post-14205789 ; https://forums.tomshardware.com/threads/intel-temperature-guide.1488337/#post-12703929 ; which in short, seems to suggest mid-70's is safe but try not to exceed 80c at the most. because the TCase is '67.x C' but it says to add about 5c for more of a true measurement which would be 72c etc.

 

but I noticed after looking around in my BIOS (as someone I was talking to online suggested it if possible) that I can alter my CPU's voltage and I noticed that dropping the CPU voltage by 0.050v gives a 5c(maybe a bit more) decline in CPU temps (I can adjust by 0.010 increments up or down in voltage up to 0.630v in total up or down). then I tried 0.100v (which gave another 5c+ temp drop) and finally 0.150v and on the 0.150v drop I noticed after running Prime95 that once I exited it, the system completely hard locked (like screen looked fine but nothing responded) which I assumed was due to lack of CPU voltage, so currently I got it to 0.130v lower than the default/automatic setting and so far things seem good and even on Prime95 SmallFFT/SmallestFFT tests in their default state (so AVX is enabled which puts more stress on CPU) I don't seem to get more than 76c MAX although 74-75c is more typical after things peak (room temp is about 75f or so at the time).

 

also, my CPU runs at 3.5Ghz (technically slightly less but basically is very close to 3.5Ghz) when Prime95 is loaded even though it's only officially rated at 3.3Ghz. I checked this on Linux Mint using 'watch lscpu' command (I also used the 'sudo i7z' which gives a bit more details). but I do notice in terms of general use that once in a while it will briefly shoot up to about 3.6Ghz but I assume has to do with the CPU's Turbo function since it claims it can be as high as 3.7Ghz by the official Intel specs page on my i5-3550 CPU... https://ark.intel.com/content/www/us/en/ark/products/65516/intel-core-i5-3550-processor-6m-cache-up-to-3-70-ghz.html?wapkw=i5 3550 ; but there is nothing out of the ordinary here as it's reacted this way before I touched anything as I initially installed the CPU, the BIOS detected a new CPU as it gave me a 'new CPU installed' message, and that was that and I started using it.

 

so with all of that said... do you think I am good as I am or would re-applying thermal paste be worth considering, or what other thoughts suggestions do you have? ; because at the moment my CPU temps on Prime95 peak at pretty much the mid-70's with the CPU voltage drop tweak of 0.130v. which means my CPU will likely be 60 something C tops in temps in general use. NOTE: I am running newest BIOS for my motherboard (ASUS).

 

off the top of my head... after removing the i3-2120 heatsink, which was in place since May 2012, which seems to peak at mid-60c on Prime95 (all at stock BIOS settings etc), I took a picture of the CPU surface and heatsink and I must have done a fairly good job considering the spread looked good. but, from memory, I want to say I used a little more thermal paste when applying the i5-3550 CPU recently as here is a couple of pictures of my previous i3-2120 CPU's thermal paste on heatsink/CPU surface after removal and that CPU's temps were as expected...

DSCN2988_Small.JPG

DSCN2989_Small.JPG

Share this post


Link to post
Share on other sites
Peresvet    363
Posted (edited)

You should apply new thermal paste every time you re-seat, re-install your CPU/FAN.

 

The paste should cover the whole surface of the CPU.

 

IF, after installing the heatsink, some of the thermal paste leaks out due to pressure, just carefully remove it.

 

It's better to apply extra paste and then remove it than not having enough of it (as your pic shows).

 

 

Share this post


Link to post
Share on other sites
ThaCrip    727
Posted (edited)

I should have been clearer. I completely cleaned the old stuff off (with 91% isopropyl alcohol) of heatsink/CPU and reapplied from scratch. what you see in those two pictures is shortly after I removed the i3-2120 CPU before I installed the i5-3550 and I completely cleaned off that before applying Arctic Silver 5 with the usual small-ish dot in the center of CPU.

 

so I was just wondering if it's worth re-doing or do you think I am good since my current temps are safe?

  • Like 1

Share this post


Link to post
Share on other sites
Mindovermaster    3,436

Any time you take the heatsink off, you need to clean it and reapply,.

 

I'm suprised that i3 still works. Those are burn marks on the CPU...

  • Like 1

Share this post


Link to post
Share on other sites
ThaCrip    727

@Mindovermaster

 

4 minutes ago, Mindovermaster said:

Any time you take the heatsink off, you need to clean it and reapply,.

 

Yeah, I know that. I always do. that's not the issue... I was just saying whether you think I am currently okay or not? (like I mentioned in my above post, while I am not 100% sure I am thinking this time I applied a bit more thermal paste than when I did my i3.

 

4 minutes ago, Mindovermaster said:

I'm suprised that i3 still works. Those are burn marks on the CPU...

Ahh, I think I see what you mean... on the outer edge with the small lines/marks that match up to where the heatsink is? ; but even so, the CPU generally ran cool as the hottest it got was probably mid-60c tops and that was only a little on Prime95 tests.

 

so given that info, are you saying I did not apply enough thermal paste on that i3-2120? ; because if so I am pretty sure I applied a bit more on the i5-3550 this time around.

 

because looking around online... there are some articles that show different methods of thermal paste application and in terms of temps they are all pretty much the same which is why I use the small dot type in center of CPU since it wastes the least amount of thermal paste.

 

so do you think I am okay with my current thermal paste application, or do you think it would be worth re-doing what is currently in place on the i5?

Share this post


Link to post
Share on other sites
Mindovermaster    3,436
1 minute ago, ThaCrip said:

@Mindovermaster

 

 

Yeah, I know that. I always do. that's not the issue... I was just saying whether you think I am currently okay or not? (like I mentioned in my above post, while I am not 100% sure I am thinking this time I applied a bit more thermal paste than when I did my i3.

 

Ahh, I think I see what you mean... on the outer edge with the small lines/marks that match up to where the heatsink is? ; but even so, the CPU generally ran cool as the hottest it got was probably mid-60c tops and that was only a little on Prime95 tests.

 

so given that info, are you saying I did not apply enough thermal paste on that i3-2120? ; because if so I am pretty sure I applied a bit more on the i5-3550 this time around.

 

because looking around online... there are some articles that show different methods of thermal paste application and in terms of temps they are all pretty much the same which is why I use the small dot type in center of CPU since it wastes the least amount of thermal paste.

 

so do you think I am okay with my current thermal paste application, or do you think it would be worth re-doing what is currently in place on the i5?

I'm talking about the center bit...

Share this post


Link to post
Share on other sites
ThaCrip    727
Posted (edited)
6 minutes ago, Mindovermaster said:

I'm talking about the center bit...

 

The i3-2120 CPU looks fine right now after I cleaned off the thermal paste as nothing looks out of whack as the surface of it looks like the used i5-3550 CPU I bought.

 

also, if this matters, I am pretty sure when I first installed that i3-2120 I think I may have swashed things around a bit (like moved the heatsink a bit) before fully seating the heatsink to the CPU, if that matters?

 

so I doubt it's damaged from the looks of things and it's been rock solid stable since I bought it and still is.

Share this post


Link to post
Share on other sites
Mindovermaster    3,436
1 minute ago, ThaCrip said:

 

The i3-2120 CPU looks fine right now after I cleaned off the thermal paste as nothing looks out of whack as the surface of it looks like the used i5-3550 CPU I bought.

 

also, if this matters, I am pretty sure when I first installed that i3-2120 I think I may have swashed things around a bit (like moved the heatsink a bit) before fully seating the heatsink to the CPU, if that matters?

Jiggling it a little won't hurt it. When it is clamped down, it will naturally spread your paste anyways...

  • Like 1

Share this post


Link to post
Share on other sites
Jim K    15,378

Couple of things ... why are you stress testing it with the stock Intel i3 cooler?  The i5-3550 is a hotter chip and the stock Intel cooler for it had a copper core...the i3 cooler is all aluminum.  Intel coolers are *good'ish enough* but you've stepped down the cooling ladder. 

 

Your problem is, IMO, stressing it with an all aluminum stock Intel cooler which the i5 wasn't shipped with...probably less to do with the thermal paste. How is it under *normal* workloads/gaming ... that is where your concern should be.  

 

IMO  :)

  • Like 1

Share this post


Link to post
Share on other sites
ThaCrip    727
1 minute ago, Jim K said:

Couple of things ... why are you stress testing it with the stock Intel i3 cooler?  The i5-3550 is a hotter chip and the stock Intel cooler for it had a copper core...the i3 is all aluminum.  Intel coolers are *good'ish enough* but you've stepped down the cooling ladder. 

 

Yes, that sounds correct as I thought I recalled that to with the proper i5 coolers having a copper core as you said. so assuming that copper core makes a noticeable difference, then what I am experiencing is probably about expected given the lack of copper core on bottom of my i3 cooler.

 

2 minutes ago, Jim K said:

Your problem is, IMO, stressing it with an all aluminum stock Intel cooler which the i5 wasn't shipped with. How is it under *normal* workloads/gaming ... that is where your concern should be. 

 

Well like I say, with the full on Prime95 test in it's default mode (on SmallestFFT/SmallFFT) with the 0.130v drop in CPU voltage, it peaks around 75c (75f or so room temp) with 71-73c being more typical. so it's pretty safe to say gaming and general tasks won't get above mid-to-high 60's.

 

in fact, since you brought it up... I just did a 20 minute Prime95 test with AVX disabled and in SmallFFT mode (which according to that site seems to be more inline with official Intel stuff and should be closer to a real-world general use 100% CPU load) and my temps peaked at... 71c peak, but generally 67-69c. (room temp is about 81f) ; so this should be a rough worst case scenario for general usage and will likely often be a fair amount less than this. but like I say a lot of these CPU temps drops came from the CPU voltage decrease as prior to doing that my CPU temps where 10-15c or so hotter (each 0.050v drop shaved 5c+ from CPU temp). my CPU is running @ 3.5Ghz (3491.xxMHz) during this Prime95 test.

 

p.s. on Linux it shows 85c is high and 105c is critical for whatever this is worth for my i5-3550 CPU.

Share this post


Link to post
Share on other sites
ThaCrip    727

I just finished a Prime95 24hr test (SmallFFT for 24hrs straight) about  a hour ago now.

 

when I was observing it, things generally were 73-75c (mostly 73-74c) but had some fluctuations depending on my room temperature and even by Intel's 67.4c official rating for 'TCase' on the i5-3550 that seems to translate to about 72.4c, which if that one site is correct (since they claimed to add 5c to the TCase rating), I would imagine anything around that or lower is probably considered safe enough and I am pretty close to that either way and will likely be in the clear in normal real-world use settings.

 

but at least things are stable with my 0.130v decrease in CPU voltage. without the decrease in CPU voltage temps would have been around 13c hotter then the numbers I listed above which probably means likely between 86-90c or so, which is too high for my liking (since 85c is considered 'high' and 105c is 'critical' and I think CPU will throttle at that temp at the highest). even based on my previous comment in my previous post of "71c peak but generally 67-69c" that means I would probably be around 80-84c on more of a real world 100% load for any length of time which is pushing it.

 

so thanks for everyones time 😉

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

  • Recently Browsing   0 members

    No registered users viewing this page.