Opty165 has 14C diifference in core temp under load - remove IHS?

[derekoldschool]

My Opty's 2 core temps have a difference of 7-8C when idle.

 

This difference goes up to around 14-15C under load, where Im getting around 67C on CPU#1 and about 80C on CPU#2 as reported by Everest Ultimate 3.50.817 Beta. However, the overall CPU temp reported is between 49-52C (generally about 50C, which is a bit hot, but should be fine - correct me if I'm wrong)

 

I'm going to reset the HSF just to see if that makes a difference.

 

What I'd like to know is if anyone else has had this much of a difference in core temps, do you think it is affecting system stability, and if removing the IHS may help solve the problem.

[likewhoa]

on air i wouldn't recommend taking your IHS off, do that on Water. and yes you should do a remount if your load temps are not equal,. my idle temps have a 5-8C difference but that balances itself on load..

keep your load temps below 55C max.

[Mailok]

keep your load temps below 55C max.

 

Temps of both the cpu cores or the overall CPU temp?

[likewhoa]

Temps of both the cpu cores or the overall CPU temp?

 

overall

[Mailok]

HUmm I am a little confused about your answer. I mean, I read that the max temp for this kind of processor is 55c, the core temp is readed inside the cpu so it should be more realistic than the one read by the bios (which I know is something outside the cpu), the overall you wrote. But we have a dual core cpu and I never seen a dual core with the same temp for both core so an average of both temps can be more realistic or simply the worst temp, but the overall one is seems so low to take into account, what do you think?

[derekoldschool]

it would make sense that the internal core temperatures would be more than the temperature reading wherever that other CPU temp sensor is located - if its outside the cpu cores then it wont be as hot as something that is closer to or inside the core.

 

I use the same 55C temp rule, likewhoa - I dont really like to see it go over 50C under load. However, since everest got the new version that shows the "core" cpu temp awhile back, it seems that those temps are significantly higher than the average or even the lower temp of my 2 cores. For example, right now, according to Everest, Core#1 = 35C, Core#2 = 27C and overall CPU = 26C

 

I'd like to know what the limit of the core temp is before the cpu starts to get flakey

[Mailok]

I agree with you, probably the best question to ask is which temp the core should max before unstable or damage.

[likewhoa]

just watch your temps set by MBM or SmartGuardian,.. that's what i meant by overall. Stay below 50-55C on load and you should be fine for the long run.

 

I keep my procs under 45C on load if possible and a 47C max load temp.

[PhoetuS]

Around 50c load temps are fine, I always try to keep them under 50c, but that is just me. What is you Vcore?

 

You can pop the lid off that cpu since you are using the Zalman 9500, (I have one on my naked 170 right now) it should work just fine for you.

 

Definately try reseating it first, maybe a couple times to see if it makes any difference, but if you are willing to potentially kill you CPU and don't mind runnung it lidless, I say go for it. I guess the big question is, Can you afford to replace it if you kill it? If yes then give it a try. I am not suggesting that you will kill it, as long as you are careful, it is pretty easy to do.

 

I don't know if running it lidless will equalize the temps or not, but it is possible that you have uneven or poorly seated TIM under the IHS... If that is the case, it should make a difference.

 

You asked if the delta between cores is affecting stablity... So I ask you: Is it affecting your system stablity? Are you experiencing instablity?

 

 

If you choose to pop the lid and you are going to use the 9500 I have a couple of suggestions:

 

1. Before you mount the HSF, cover the IC's with electical tape. This will help prevent getting any TIM on them & if the HSF sits a bit crooked it could prevent a short.

 

2. Find a washer, aprox ~3/32"-1/8" thick, that fits over the mounting post of the 9500. Put it on the post & place the mounting bracket on top of that. There should be enough of the post above the washer that the mounting bracket still catches on it. The purpose of the washer is to make up the extra height that is now gone with the IHS. It will help keep adequate pressure on the cpu core and prevent the 9500 from gradually settling and putting more pressure on the down side of the core.

 

3. I suggest using AS Ceramique as the TIM (not AS 5). I have used both on my naked 170, & I definately like the Ceramique better.

 

4. Someone suggested in this thread a possible solution to not having a shim for the naked A64 (like the socket A shims). Place 3 layers of electrical tape in each corner of the cpu. Not big pieces, probably 1/4" maybe. That should help even out the pressure & keep the HSF sitting squarely. Whenever I reseat my 9500, or if I pop the lid on my new 170, I will be trying this technique.