• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Zambezi Fx-4100 overheating

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

toxicsock

Registered
Joined
Dec 6, 2011
Hello everyone,

I just recently started to have temperature problems with my CPU about a week ago. The first time my computer shut off by itself, i rebooted it and check the temp on my BIOS and the CPU was @ 60 degrees celsius idle. Is this too hot?? The heat sink is stock and i am looking into buying this one: http://www.newegg.com/Product/Product.aspx?Item=N82E16835103099

Also, I have seen it has high as 70 degrees celsius on my BIOS... what could be causing such high temperature readings? All my fans are working and i have not overclocked any hardware. My computer is located at a relatively open area.

Here is a screenshot of my hardware temps: http://desmond.imageshack.us/Himg856/scaled.php?server=856&filename=hardwaretemps.png&res=landing

The max value most likely occurred while i was playing Diablo 3. The current values were after exiting the game and taking the screenshot.

Also, any recommendations for an after market heat sink would be greatly appreciated.
 
Last edited:
Those temps are definitely borderline. Has the heat sink or the case fan filters become clogged with dust? Are you ambient temps higher now because it's summer (in the northern hemisphere, anyway)? Are all the fans turning as they should? Has the cooler been moved off the CPU so as to break the contact of the thermal paste? Have you tried redoing the thermal paste and remounting the heat sink? The Cooler you reference is a good match for that CPU.
 
Please use CORETEMP to measure the temps, the coretemps matter, not anything else.

Bios temps do little to help, they are mosty inaccurate.

We have MANY MANY posts here about better coolers. Time to get a drink and read for an hour or so.
 
That Hyper 212 EVO you linked is a decent cooler for it's price and will do loads better than the stock crap heatsink that came with your processor. And like Conum said, check your temps with the Core Temp program and see the temps reported by it jive with Hardware Monitor. As for other cooler options, there are better out there at various price points than the EVO. But the EVO is a massive upgrade over the stock cooler and most probably will fix you up pretty well.
 
toxicsock, I still think those temps are higher than they should be, even with the stock cooler unless you have overclocked the CPU or unless you're ambient temps are quiet high. You did not mention that the CPU was overclocked so I assume that is not the case.
 
I picked up a Cooler Master Hyper 612 PWM. It comes with one 1 fan (up to 85 cfm @ 2.1 pressure) but i replaced it with 2 similar fans with 50% more pressure in a push-pull deal. This thing is awesome. Between it and replacing my dinky case with a HAF XM, I dropped my idle temp from 53 to 34. i'm o/c'd to 4050 ( don't have the knowhow to keep it stable past that, yet). At load it tops 44-45. All air cooled baby! Not only that you have directional options for it. I placed mine blowing toward upper exhaust fans so it helps with the GPU, too.

it is a big one though so you need the case for it.
 
Please use CORETEMP to measure the temps, the coretemps matter, not anything else.

Bios temps do little to help, they are mosty inaccurate.

We have MANY MANY posts here about better coolers. Time to get a drink and read for an hour or so.

That's actually wrong. Even TheCoolest (the member here who wrote coretemp) will tell you that if you go by the "core temps" you need to add around 15C to get an approximation of your CPU Temp.

TCase = The sensors on the die package that are averaged to give a temperature value

TCore = The individual core temps that are "approximated" by taking the nearest TCase sensor and mathmatically "guestimating" that particular cores temperature

TJunction = The sensor beneath the socket where the CPU's pins connect with the board.

TCase Max = The maximum operating temperature listed in all of the AMD Thermal information guides (i.e. 61C for the 8120/8150 etc.)


And Who is Brown679 and why does he keep posting a pic that nobody can see in all the threads?
 
Did you remove the label on the bottom of the CM Hyper 212+ that says "Remove before instillation"?

Dead serious, I've seen this mistake made all to often.
 
That's actually wrong. Even TheCoolest (the member here who wrote coretemp) will tell you that if you go by the "core temps" you need to add around 15C to get an approximation of your CPU Temp.

TCase = The sensors on the die package that are averaged to give a temperature value

TCore = The individual core temps that are "approximated" by taking the nearest TCase sensor and mathmatically "guestimating" that particular cores temperature

TJunction = The sensor beneath the socket where the CPU's pins connect with the board.

TCase Max = The maximum operating temperature listed in all of the AMD Thermal information guides (i.e. 61C for the 8120/8150 etc.)


And Who is Brown679 and why does he keep posting a pic that nobody can see in all the threads?

So, when peeps use coretemp for AMD and show the cores at 55C under load and we say that's okay. That's wrong? Meaning the hottest core is 15C above that at 70C but the hottest core temp allowed by AMD is 60Cish for stability? I see that as a standard for at least a year.:shrug:

I must be way wrong I guess. Please write a sticky or something so we get it right.

I been wrong for a while I guess.:confused: Fix me please. So I'm right.
 
So, when peeps use coretemp for AMD and show the cores at 55C under load and we say that's okay. That's wrong? Meaning the hottest core is 15C above that at 70C but the hottest core temp allowed by AMD is 60Cish for stability? I see that as a standard for at least a year.:shrug:

I must be way wrong I guess. Please write a sticky or something so we get it right.

I been wrong for a while I guess.:confused: Fix me please. So I'm right.

No, there are no actual "core" temperature sensors. There are only sensors on the die package itself (not actually on the die itself). Meaning the individual "core" temperatures you see listed are "guestimates" of each individual core temperature. Meanwhile, TCase is an "averaged" result taken from all the sensors spread out across the die package (somewhere above the pins but below the die. Exactly "where" is proprietary and they won't say). It gives a "whole" CPU temperature. That's why you see "TCase Max" listed in the thermal white papers not "TCore Max". So when you see "core temps" they are usually somewhere around 15C or so (varies of course from chip to chip) below the TCase temp which is the "overall" temperature of the CPU. That started with the K8's.




From:[email protected]
Sent:Thu 5/17/12 8:14 AM
To:[email protected]

Dear Stephen,

Your service request : SR #{ticketno:[8200488157]} has been reviewed and updated.

Response and Service Request History:

You pretty much nailed it. I was able to get a little more info from the embedded team into borderline-proprietary information, so I'll try to elaborate on what you understood. TCase for AMD processors comes from a few thermistors (not one, apparently, just found that out) inside the processor case (at the bottom, where the pins are), connecting down to the CPU via the Junction. There are always more than 1 (at least 2, up to 6-8 potentially, but no elaboration given on how many per model), but the TCase temperature is determined by averaging those values out, done by the processor. TJunction is the temperature where the pins hit the board, and is usually a couple degrees cooler as all 940/941 pins aren't all firing at the same exact time, and not always evenly distributed when only 400 are on at one time.
TCore is actually mathematically guessed based on the varying TCase values, as there is no way to get a diode on top of the cores inside the processor, and putting it underneath the cores (between the bottom of the case and the bottom of the cores, which hover on a little silicon platform) would yield an inaccurate reading. As such, optimizing the core space on the wafers by keeping thermistors off, they just mathematically extrapolate the core temperature from the TCase values, based on core location on the processor and the values retrieved in that general area, plus some mathematical calculations.
TJunction is still a diode on the board, under the processor, which most boards still have, just in case the TCase values (or TJunction value given by Intel processors) are wrong for whatever reason. Though in some cases, TJunction can be off by as much as 20F, so it's obviously not an ideal value. Still, there are a lot of board manufacturers who will still include it, regardless of how necessary, because it's how they've always done things, and if there are problems with new processors or broken thermistors, they can still report a temperature, even if it's not the most accurate.

Sorry for the misinformation about the cores, I really had to get the embedded guys to give a little to get some information confirmation, including the number of thermistors in the case and where this coretemp comes from. A coworker summarized it well by saying that it's so tough trying to get confirmed information, because you get different reports from 3rd parties, and the actual designers/manufacturers want to keep as much information secret as possible. Sorry that this still isn't 100% concrete, but they finally gave in a bit and gave me a bit more information to work with this time, so now you (and I) have a clearer definition at least of what's going on temperature-wise.

In order to update this service request, please respond, leaving the service request reference intact.

Best regards,

AMD Global Customer Care

_____________________________________________________________________________________________

This email is a direct result of your contact with AMD Global Customer Care and not part of a campaign. There is no need to unsubscribe to this email as you will only be contacted again if you directly request another service from AMD Global Customer Care.

The contents of this message are provided for informational purposes only. AMD makes no representation or warranties with respect to the accuracy of the contents of the information provided, and reserves the right to change such information at any time, with or without notice.

_____________________________________________________________________________________________


Original Text


From: [email protected]
To: [email protected]
CC:
Sent: 05/16/12 12:44:16
Subject: RE: AMD Service Notice:{ticketno:[8200488157]}



I'll sum up what I've learned and deduced and tell me if it's makes sense and/or is correct. As you've stated the Tcase is a singular temperature taken at the top of the CPU (I'm assuming from a diode at the top of the die where it makes contact with the IHS?). What seems to me as a dead giveaway that Tcase is not being used by programs like HWmonitor is that they list seperate values for each of the cores that can be the same but usually differ from each other (usually by 1-3C). Are individual "core" temperatures taken from a diode within each core? Or is it mathmatically based off of something like Tjunction to give a "guesstimate" for each core? But clearly the standard assumption I see that says quote: "CPU Temp = Tjunction or true Junction Temperature (This reading is taken from the sensor fixed in CPU socket on Motherboard.)" is clearly wrong. I was also under the assumption that motherboards didn't use a "socket" diode anymore as well. Does all of that make sense?
 
Last edited:
Yep. Kinda, not a mobo engineer.

So if we ask a AMD user to use Coretemp and the temps of the cores and try to keep them under 60C is that wrong? I NEVER recommend an AMD user to use realtemp, ONLY coretemp.

Is Coretemp right? I know unlocking cores on older AMDs screw up temps, but we are past that with the new ones.

That is the crux of an answer I need. Are we telling our users the right info?
 
Yes and no. If your "TCase Max" is 61C AMD is telling you to keep the CPU Temp no higher than 61C. The better the cooling the more thermal energy it's able to remove from the die package, so it really doesn't matter what the "cores" are as they are going to be derived by the same sensors just with a formula instead of a direct reading. HOWEVER, even they admitted that TCase Max is a "conservative" limit.
 
CoreTemp is reporting TCore. I'll have to dig up the thread where thecoolest weighed in. His advice was if you're using CoreTemp add 15C to the "core temp" reading to give you your CPU Temp.

Here's a better understanding of how the temperature monitoring goes. The CPU outputs TCase and the TCore readings to the BIOS/UEFI. The BIOS/UEFI then assigns "labels" to them (which differ between manufacturers) such as CPUTIN, TMPIN1 etc. The monitoring programs like CoreTemp and HWMonitor just report those outputs from the BIOS/UEFI. For CoreTemp it's taking the individual core temp readings from the BIOS/UEFI output (which comes from the TCore output from the CPU) and using them. I'm 99.999% positive that for Asrock and Asus "CPUTIN" is the TCase reading however I've been trying to get them to respond to requests for info.
 
Last edited:
Im not exactly sure how it all works either. But Conum, the Tcase max isnt 61C for all fx chips. My 6100 is 71C max. And Bubba is right, the Tcase max operating temp is not your core temp. I've heard people say add 10C to your core temps and thats the temp you need to watch for, but I prefer 15C to be even more safe. I can dig up an email from AMD Tech Support stating that the Tcase Max Temp is the CPU temp as a whole and not for the individual cores.

However, I have had my core temps hit 70C for 4-6 hours of prime tying to reach 5Ghz and my cpu is still alive. Like bubba also said I think that 71C (for my chip) is a conservative limit for 24/7 use but it is not recommended to run at those temps.
 
It's 70C for our chips. When I pushed mine to 5GHz I was hitting 74C during prime95 with ambient around 19C or so. But even if I kept it at 5GHz I would rarely see 70C during normal use. Just while stressing. I ran it for 4 hours and it would fluctuate between 69C and 72C with some spikes up to 74C and of course my chip didn't explode :p I think it's somewhere around 90 that the thermtrip on the CPU actually kicks in (not that I'm recommending you take it up to 70's or 80's).
 
Conumdrum. I've just realized that you have an Intel CPU. Remember that it's completely differen than AMD. Intel goes by Tjunction Max for their "max recommended temps" if I remember correctly and I don't believe there is a "TCase" for Intel CPU's but you would have to verify that. Not completely sure how the "core temps" work on Intel CPU's either.
 
If i am not mistaken, I think core temps are higher than cpu temp on Intel chips, and its been so long since I've owned one I dont remember what temp they go by.

Bubba you mean 74C for core temps?
 
No, the 4100 and 6100 (the 95W) have a TCase Max of 70C. It's 61C for the 125W FX CPU's. I would love to know why that is since if you overclock the 95W 8120 (if you can find them) you basically turn it into a 125W 8150 I would think.
 
Back