FX-4350 Core temps?

[t1nm4n]

I got a new SSD for my OS and instead of transferring I decided to install everything fresh, got most everything installed and wanted to check temps and what not my Core temps seemed to be off a bit, what do you all think? This is just from surfing the web downloading updated software and installing it.



I thought it might of been the factory TIM had moved around cause of uneven mounting pressure (not sure the layout of the die). So I took it apart and it looks pretty good I would have to say.





So the only thing I can think of is that the TIM under the cover has started to shift and isn't in that good of contact cause here is what I have after applying AS5 ( I know it has a ridiculous burn in time).



I know it's not definitive proof that the TIM on the die has gone bad, but I shouldn't be getting these kinds of temps just reading forum posts.

Other than delidding the cpu which (as cheap as the cpu are, it might be an option) what can I do, just run it like I stole or it or start to worry it will burn the house down.



Edit: OK, so after some more searching I ran across a few posts on here that have me kind of confused, maybe. If I have this right, there is not actual diode for each core on an amd cpu, they can be placed elsewhere on the die and some mathematical calculation is how they get core temp, HWInfo is just able to put a value to these diodes and assign them as core temps, meaning I don't have anything to worry about and this is just a wild goose chase for me to find something wrong that isn't?

 

[Blaylock]

1. I'm not real familiar with HWinfo and the temps from this software maybe a little off. If you could use HWMonitor (free version) I believe the temps are fairly accurate depending on mobo/CPU combo and what you're looking for is Package(Core) temp and Socket (TPIN1 or 2 generally).

2. Idle temps really don't mean squat as AMD has confirmed that the sensors give bad readings at idle. You'll want to make sure to keep your Package temp under 62c and your Socket temp under 71c.


EDIT: I may be wrong here, but I also believe the FX chips have the IHS soldered on. e.g. no TIM

 

[t1nm4n]

They seem pretty close, those didn't have me concerned, just the T2 temp of 61c from HWInfo, but that might just be a specific Diode inside the cpu, never noticed it before. Just wasn't sure if it was something I should be concerned with or not.

 

[Blaylock]

Are these load temps? If they are you're golden. If not, you should run Prime 95 (blend mode) for about 20 mins and post the HWMonitor. For full stability you'll want to run it for a few hours, but for checking temps 20 mins is plenty.

 

[t1nm4n]

Those are idle temps, but the T2 is read by the MB as the NB temp, as I had overtemp for it to shut down at 70 and I hit 70c do some 3d mark runs. So those t0-3 have nothing to do with individual cpu core temps. I would imagine the T1 might be the actual core temp and T0 + T3 / 2 equals my package temp. might need to do some more investigating for that, but I do know that the T2 temp as read by HWInfo is my NB temp according to how when it hit 70c my MB shut the pc down for the overtemp I had set up. So what is a safe temp for the NB?

 

[Blaylock]

Those T0-4 are just different sensors on the board. Just for clairity, I'm assuming this is on the Formula-Z? Your package temp is your core temps. They don't always show as individual cores, though I wish they did.

If I were to guess I'd say:

Motherboard sensor = ambient case temp
CPU = CPU Socket/VRM
Package = Cores (obvious)
T2 = VRM/Board NB

Something to keep in mind is there are actually 2 NB's. One is located in the CPU, the other is heatsinked on your motherboard. Also sensors are generally placed where they can fit them on the board. Which is why they aren't as accurate as we would like. I've seen NB temps in the 100c range but that's not ideal. 70c is definitely on the "safe" side of things. It really only gives you a 7-10c margin.

If I were in your position I would turn off the thermal throttling/shut down and run P95(blend) for 20 mins at stock settings and make sure the temps don't exceed 62c/71c. Occationally touch the VRM and NB heatsinks. I know this isn't as scientific as we like, but if the sinks get to hot to hold your finger to than it's to hot(roughly 120c). If you REALLY want to identify which sensor is where, get a can of air. As your system heats up spray the air onto your VRM sink for a few seconds and see which T? drops. Do the same to the NB sink. Just be careful not to allow moisture from the air can to form on your system. That would not be desirable.

Here's a pick of my FX-4130 for comparisons.

 

[t1nm4n]

Yes the Formula-Z would be the board in question. Thanks for reminding me about the dual NB with AMD set-ups. I'll bet that sensor is on the Board, might explain why it isn't affecting the cpu like I'd think it would. I mean that's what had me going there, if there was a part of the cpu pumpin' out 60+ temps how is it not affecting the rest of the processor? Thank you Blaylock.

Kind of surprised more haven't chimed in to help.

 

[RGone]

Kind of surprised more haven't chimed in to help.

Things have been slow in the forum and when really cold up north many go out to play. Blaylock is spot on anyway.
1. We just about use 'nothing' but HWMonitor free, so we are all on the same page.

2. We want on the same page because no telling what T0 thru T3 is on the CHV-Z board and even less knowledge when they plug in a Giga board or MSI or whatever since they all assign values differently too too often when not talking about Cpu/Socket temp or Core/Package temp.

3. AMD has verified that any idle temp is wonky. They are not in the business of reporting idle temps since they are only concerned about when a temp is too high and cpu needs to shutdown. So at idle temps for CPU/Socket and Core/Package are not valid except under load and above approximately 40c or greater.

4. And like Blaylock said; if a person wanted to try and See what the T0 thru T3 were actually giving a reading for, then canned air sprayed on the various heat sinks and checking for a lowered temp reading might point to what a piece of software we don't use...is giving reading for or about.

RGone...

 

[Mandrake4565]

Chime Chime,

A lot of us running these Fx chips, put fans directly on the NB and Vrm heatsinks, especially when running the Fx 8xxx chips overclocked. Additionally running a AIO or Custom H2O unit exacerbates the issue of that section getting hot because you lose the airflow you would normally get with an Air heatsink. Going back to the idle temps Blaylock hit the nail on the head, these things just aren't accurate when below 40c and not under load. Mine idles at 7c in a 22c room Oced to 4.7 on ambient cooling.

 

[RGone]

Chime Chime,

A lot of us running these Fx chips, put fans directly on the NB and Vrm heatsinks, especially when running the Fx 8xxx chips overclocked. Additionally running a AIO or Custom H2O unit exacerbates the issue of that section getting hot because you lose the airflow you would normally get with an Air heatsink. Going back to the idle temps Blaylock hit the nail on the head, these things just aren't accurate when below 40c and not under load. Mine idles at 7c in a 22c room Oced to 4.7 on ambient cooling.

Chimes are in.

RGone...

 

[t1nm4n]

Not concerned with the low idle temps, read that in a couple of areas, it was that T2 temp that HWInfo retrieved from some sensor, at first I thought those were individual proc temps for each core, then I read AMD doesn't do that it's some proprietary algorithm, but wouldn't that kind of temp affect the rest of the cpu, Blaylock reminded me there are two NB's with AMD. Now I got one of those funny pics, I didn't do anything to this and this is all on the up and up.





Now I know this lil AIO cooler works pretty good

Now I would think something should be smoking if I had those temps, I guess during the Prime95 I got bored cause I couldn't get the temps up so I ran Aida 64 with everything going and some data got lost in the bombardment of processes.

I did take my heat gun and the heatsink closest to the rear I/O shield was temping from 50-52.5, got hotter as it go closer to the top GPU. So I will guess (wild shot in the dark more likely) T2 from HWInfo is the Board NB temp and something I shouldn't be concerned with, as it doesn't seem to warm up much unless I am doing graphic tests or watching videos online ( I also went through the bios and found the NB voltage was at 1.4 for some ungodly reason), I must of fat fingered it or something, I was darned sure I changed everything back to stock.

 

[t1nm4n]

Things have been slow in the forum and when really cold up north many go out to play. Blaylock is spot on anyway.
1. We just about use 'nothing' but HWMonitor free, so we are all on the same page.

2. We want on the same page because no telling what T0 thru T3 is on the CHV-Z board and even less knowledge when they plug in a Giga board or MSI or whatever since they all assign values differently too too often when not talking about Cpu/Socket temp or Core/Package temp.

3. AMD has verified that any idle temp is wonky. They are not in the business of reporting idle temps since they are only concerned about when a temp is too high and cpu needs to shutdown. So at idle temps for CPU/Socket and Core/Package are not valid except under load and above approximately 40c or greater.

4. And like Blaylock said; if a person wanted to try and See what the T0 thru T3 were actually giving a reading for, then canned air sprayed on the various heat sinks and checking for a lowered temp reading might point to what a piece of software we don't use...is giving reading for or about.

RGone...

I had gathered that much about HWInfo being different on different boards, just thought someone on this forum might have the same board and have already figured out what the T0-3 temps are for on this software. I wish I could be out there playing in the snow, wait no I don't I like to stay warm.

So P95 been running for 45min, other than when I tried to add more to the plate with Aida64 temps are nice and low, surprisingly T2 keeps getting lower also

 

[t1nm4n]

Chime Chime,

A lot of us running these Fx chips, put fans directly on the NB and Vrm heatsinks, especially when running the Fx 8xxx chips overclocked. Additionally running a AIO or Custom H2O unit exacerbates the issue of that section getting hot because you lose the airflow you would normally get with an Air heatsink. Going back to the idle temps Blaylock hit the nail on the head, these things just aren't accurate when below 40c and not under load. Mine idles at 7c in a 22c room Oced to 4.7 on ambient cooling.

Thank you so much, ( I know I'm kind of slow) but after a couple of minutes to think about whats changed int he case and your comment about losing Airflow from the AIO ( not something I have a problem with), the top rad is an intake and blows directly down towards the VRM and NB heatsinks and the exhaust fan keepis moving in the proper direction. I did earlier this week change setting in the Link software to lower the speeds of the fans until te cpu heats up more and changed the speed of the exhaust fan to a lower speed to quiet it up some, thus creating the problem you mentioned and I just said I don't have.

OK so I created the problem making my pc quieter and didn't realize it, but I learned something and those who read this realized I'm probably the most ignorant person on the forums here, or lost a few IQ points themselves

Thanks to those who Chimed in, I guess this problem is solved, just another user created problem.

 

[RGone]

Honestly t1nm4n...

...we all do the two steps forward and one back more often than we would like to admit. Actually that is where the learning comes in. I would never know what n0t to do unless I did something that did not work so very well.

I will be very honest when saying that quiet and overclocking can often be at opposite ends of endeavors. You can get 'big' water cooling but not at anything like a cheap price. Big rads and where to mount them and what kind of case and what fans for the FPI and it just goes on and then on. And when you get all that cool in hand...you can really never have too much air across the VRM sinks and the NB onboard sinks.

Just glad you got it sorted out and good luck man.

RGone...

 

[Blaylock]

... or lost a few IQ points themselves...

Glad you got everything worked out t1nm4n! I learned the old cold air can from the best Mr. Bobert. I'm just trying to give back.

Now for a slight thread jacking. I almost toasted my FX-4130 last night. I have newer (and louder) fans on my rads and was doing a full run of 3D Mark Firestrike just to get a baseline score and didn't realize I couldn't hear that my pump wasn't running for about a half hour (through the whole bench). Curiously I touched my CPU block and almost burned my finger it was so hot. Instantly hit the power button. I got pretty lucky as it seems everything is alright. Looks like I have a loose molex or something but I definitely need to keep a close watch now.

 

[Mandrake4565]

Chimes are in.

RGone...

That was great Rbobster thanks for the laugh.

Thank you so much, ( I know I'm kind of slow) but after a couple of minutes to think about whats changed int he case and your comment about losing Airflow from the AIO ( not something I have a problem with), the top rad is an intake and blows directly down towards the VRM and NB heatsinks and the exhaust fan keepis moving in the proper direction. I did earlier this week change setting in the Link software to lower the speeds of the fans until te cpu heats up more and changed the speed of the exhaust fan to a lower speed to quiet it up some, thus creating the problem you mentioned and I just said I don't have.

t1nm4n, from what I've seen having the case airflow go from front to rear/top usually works best. That said even with good case airflow, having fans on the NB/Vrm section will be helpful. If you look at my signature my 8350 is in a Mountain Mods case, I have 15 120mm fans blowing from front to back in it. Even with that, by adding a 80mm fan to the Vrm/Nb heatsink and putting a 120mm fan on the backside of my motherboard, it dropped my Nb/Socket temps 10c. Now take that with a grain of salt because the 8350 will definitely get that section hotter then the 4350 will but it will help.

Glad you got everything worked out t1nm4n! I learned the old cold air can from the best Mr. Bobert. I'm just trying to give back.

Now for a slight thread jacking. I almost toasted my FX-4130 last night. I have newer (and louder) fans on my rads and was doing a full run of 3D Mark Firestrike just to get a baseline score and didn't realize I couldn't hear that my pump wasn't running for about a half hour (through the whole bench). Curiously I touched my CPU block and almost burned my finger it was so hot. Instantly hit the power button. I got pretty lucky as it seems everything is alright. Looks like I have a loose molex or something but I definitely need to keep a close watch now.

And you think that you're the only one to do this? I had my 6300 shut off completely 3 times while benching it and couldn't figure out for the life of me why, until I realized the pump wasn't on. DOH

 

[Blaylock]

LOL, yep. When I saw my resevoir sitting there and nothing moving I'll bet I looked like a 3rd grader staring at a Quantum Physics story problem. Blank stare. Then my hand went instinctivly to the water block. Only when I jerked my hand back did the story problem make sense. LMAO. I'm just glad these AMD's are build hardcore.

 

[Johan45]

And you think that you're the only one to do this? I had my 6300 shut off completely 3 times while benching it and couldn't figure out for the life of me why, until I realized the pump wasn't on. DOH

Ha ha, me too, me too. You really do feel like a tool once you figure out why it won't boot.

 

[RGone]

Ha ha, me too, me too. You really do feel like a tool once you figure out why it won't boot.

And not the sharpest tool in the tool box.

RGone...

 

[Johan45]

And not the sharpest tool in the tool box.

RGone...

Nope not at all GoneBobster quite dull indeed. Have to be so I don't hurt myself. Ha ha