FX-8120 tempature sensors????

[CrustyButt]

OK guys, so I'm on my phone here but when I'm idle I'm getting 12 C on all cores. Obviously this can't happen since I don't live in Antarctica with ambient temps of 12 C...It is reported as 30-35 C load temps. I have the CM 212 EVO. Case is the HAD 932 advanced..if you need pics, will be glad to supply them. Core temp and HWmonitor present the same temps. Thx in advanced.

 

[Psykoikonov]

That is the case with most if not all of them, keep that "temp" under ~55C full load (and stable of course) and your golden.

 

[CrustyButt]

But if i do not have an accurate temp, how can i know what to do..if you know what I mean.

Edit: I heard somewhere that after ~40 C; things start to even out and become more trustworthy though...

 

[Psykoikonov]

Pretend you live in Antarctica is what ya do lol. Seriously though many have found that beyond ~55C core temp things get tricky with stability and beyond 65C may well result in a dead CPU or one that degrades.

edit: using HWmonitor you'll likely find that one of the other temp sensors is always a certain value (usually ~12C) above the "core temp" (idle, load, and everywhere in between). My thought is AMD offsets the real core temp to keep us from knowing what the real core temp is. Regardless the "rule of thumb" still applies.

 

[Vic Velcro]

The core temperatures are not a physical 'heat' temperature. The core temperature is actually Tctl which is a molecular movement temperature at the transistor junctions within the cores.

This is why your temps seem low at idle, as the molecules are not moving very much. This is also why you ignore the physical socket temperature measured by thermometers - for AMD chips. Always CORE TCTL temp, and trust it because AMD embeded it inside the chip at time of manufacture - as opposed to a sensor that was bought from some other company and slapped on a motherboard which is dependent on BIOS to report in (a bad BIOS means a bad socket temp).

 

[Vic Velcro]

Pretend you live in Antarctica is what ya do lol.

Good plan. I like it.

Seriously though many have found that beyond ~55C core temp things get tricky with stability and beyond 65C may well result in a dead CPU or one that degrades.

Rarely at 55C and only when the chip is already weakened. 62C is the recommended max, AMD processors with throttle back at around 80C - 90C.
The materials within the chip will be fine up to a minimum of 97C. Some chips BEGIN to degrade at 97C...

 

[trents]

Vic, your confident assertions run contrary to the communities' experience with AMD CPUs. Why would so many experienced overclockers say that they commonly experience instability when core temps begin to exceed about 55c? When we say instability we don't mean that the chip degrades or is in danger of frying at 55c, we mean the system crashes or blue screens.

 

[Psykoikonov]

From what I've read on forums from respected individuals in the community "core temp" readings above 65C (~80C real) is a no no, and getting stability above 55C requires patience. Granted Bulldozer doesn't seem to be as sensitive to temp as Phenom II was.

eidt: This thread and this thread are good reads.

 

[Vic Velcro]

My assertion comes from direct contact with AMD engineers.

Many overclockers are making logical assumptions, but are simply untrue.

The temps at 55C to the CPU aren't the cause. The cause is some other temp that is coincidentally rising, like RAM or NB (just 2 examples).

Search for the info I have briefly provided. Confirm for yourself that AMD engineering has PUBLISHED PUBLICLY and CLEARLY which temps to watch, at what point the temps will cause which types of behavior, and the manner in which AMD defines and explains and measures temps. When you see it for yourself, and understand it (he keeps it realistic and plain), you won't be confused or misled any longer and you will see that while I am the odd man out I am also correct.

Problems may appear when the chip reaches 55C, but not BECAUSE the chip hit 55C. Analogy: I may hurl my cookies after 11 beers, but not because of the beers. It's because of the leprechauns that keep picking up the floor and spinning it around.

 

[CrustyButt]

Thank you all. Here's my plan now that I think about it. Don't go for 5ghz or crazy high and I'm fine...? I was thinking a mild 4 or 4.3ghz.

Edit: sorry if i sound like I'm ignoring all you guys..but I do indeed like the fact that AND offset the temps to keep us from knowing the real temp. Now I will let you all know this..when I was messing around in the BIOS this morning, BIOS reported I was at 35C while at the monitoring screen. So i would assume that is correct for idle temps? Keep in mind I've got a hyper 212 evo push/pull.

 

[Vic Velcro]

I got 4.4GHz from a very recent overclocking on my 8150. I hit the wall, but not because of temps. My cores were at 68C and my socket was 91C at 4.5GHz and running fine until the NB topped 162C. It was the NB for me. If I can cool that, I can go higher. I confirmed this by spraying a non-conductive coolant directly on the NB while clocking up to 4.8GHz (CORE @ 69C and socket at only 96C probably because the sensor isn't accurate when it gets that high) and all was fine until I stopped spraying. Keep in mind that CORE temps get more accurate as they get higher and socket becomes less accurate as it gets higher. My socket sensor is at equilibrium when it reads 76C, confirmed with my IR thermometer and a calibrated mercury thermometer. Below 76C it reads warmer than actual and above 76C it reads less than actual.

I know, that ran on a bit. Anyhow, my suggestion above 4.2GHz is to watch the other temps moreso than CORE and socket.

 

[trents]

My assertion comes from direct contact with AMD engineers.

Many overclockers are making logical assumptions, but are simply untrue.

The temps at 55C to the CPU aren't the cause. The cause is some other temp that is coincidentally rising, like RAM or NB (just 2 examples).

Search for the info I have briefly provided. Confirm for yourself that AMD engineering has PUBLISHED PUBLICLY and CLEARLY which temps to watch, at what point the temps will cause which types of behavior, and the manner in which AMD defines and explains and measures temps. When you see it for yourself, and understand it (he keeps it realistic and plain), you won't be confused or misled any longer and you will see that while I am the odd man out I am also correct.

Problems may appear when the chip reaches 55C, but not BECAUSE the chip hit 55C. Analogy: I may hurl my cookies after 11 beers, but not because of the beers. It's because of the leprechauns that keep picking up the floor and spinning it around.

Well, when you start to get instability at 57c core temp and get a better cooler to take those core temps back down to say 52c and it becomes stable again, without cooling the NB or anything on the board and you don't make other bios setting changes, it might be fair to conclude it was the core temp causing instability. And this has been the experience of some I have followed on the forum. I suppose you could make the argument that it helped cool other components in the area indirectly (or even the IMC) but most of us aren't into splitting hairs.

 

[Vic Velcro]

I'm with you on the splitting hairs statement. If 55C eliminates a warm fuzzy feeling, stay below it and keep the warm fuzzy. Cooler is better, regardless of engineers, anyhow.

I did pop in just to contribute what I know regarding which temp to bother measuring and when a temp is dangerously high vs when a temp isn't dangerously high (by AMD official standard).

Core is the important temp, which is internal molecular movement and not heat and not directly measurable from outside the chip.

62C is recommended MAX Core. 80C-90C is an automatic throttle Core temp. 97C might cause damage but might not (probably will). The CPU is designed to cut out before 97C.

Mine has surpassed decent temperatures a few times, but never for very long. It's still alive and well.

35C is the max CORE that I am comfortable with at idle. 55C Core under load makes me a bit squirmy but not terrified. 65C for brief periods is barely acceptible. 75C Core and you all can call me an undertaker and donate my organs, as I won't be needing them anymore.

 

[trents]

Sounds good. My point is that sometimes what something is engineered for and how it actually performs in real life are two different things. It's also true that specs and ratings arrived at under test lab conditions aren't always confirmed in consumer experience. Many of us have experienced that truth in, for instance, the fuel economy ratings of vehicles.

 

[Psykoikonov]

Thank you all. Here's my plan now that I think about it. Don't go for 5ghz or crazy high and I'm fine...? I was thinking a mild 4 or 4.3ghz.

Edit: sorry if i sound like I'm ignoring all you guys..but I do indeed like the fact that AND offset the temps to keep us from knowing the real temp. Now I will let you all know this..when I was messing around in the BIOS this morning, BIOS reported I was at 35C while at the monitoring screen. So i would assume that is correct for idle temps? Keep in mind I've got a hyper 212 evo push/pull.

What I do is to go up 1/2 multi at a time check for temp and stability, raise voltage if needed, rinse and repeat. 4.0Ghz should be easy, past 4.4Ghz on mine gets into "too hot for my liking".

 

[RGone]

It's also true that specs and ratings arrived at under test lab conditions aren't always confirmed in consumer experience. Many of us have experienced that truth

Man you did not choke on that did you? That is a freeken mouthful of truth for sure.

 

[Vic Velcro]

Sounds good. My point is that sometimes what something is engineered for and how it actually performs in real life are two different things. It's also true that specs and ratings arrived at under test lab conditions aren't always confirmed in consumer experience. Many of us have experienced that truth in, for instance, the fuel economy ratings of vehicles.

Good point. I suppose that may also be why some get BSOD at 55C and it could be the CPU in some cases but not in others (but by intended design shouldn't be).

I did find out (in the past hour) that quite a few of the 8150s have been confirmed to have higher 'leakage' than others. When I did the steps to check if mine was high leakage or not, I discovered that mine is very low leakage for the 8150s. So, possibly some here have had a bad 'trip' because they may have higher leakage devices than the one I have. Seems my NB is weaker than my CPU and my milage isn't the same as others.

So, to anyone who hasn't heard of this 'leakage', I suggest chasing it around the search engine and find out if you got a leaky unit (sounds icky) and what you can do to accomodate it.

 

[CrustyButt]

Good point. I suppose that may also be why some get BSOD at 55C and it could be the CPU in some cases but not in others (but by intended design shouldn't be).

I did find out (in the past hour) that quite a few of the 8150s have been confirmed to have higher 'leakage' than others. When I did the steps to check if mine was high leakage or not, I discovered that mine is very low leakage for the 8150s. So, possibly some here have had a bad 'trip' because they may have higher leakage devices than the one I have. Seems my NB is weaker than my CPU and my milage isn't the same as others.

So, to anyone who hasn't heard of this 'leakage', I suggest chasing it around the search engine and find out if you got a leaky unit (sounds icky) and what you can do to accomodate it.

Thanks for the info, will definitely research that. Sitting at 4ghz @ 1.35, going for 4.2ghz @ 1.35..Once again, thanks for all your help guys. And that "Pretend you in Antartica quote is now my desktop background..to remind me to keep OCin.. As you guys were saying, or one of you did, one of the core temps in HWMonitor are almost always accurate..I cant wrap my head around this, all my core temps are at 12C...none of them are remotely close to accurate. I'm guessing no hope in ever getting correct temps?

 

[Vic Velcro]

At low values, Core reads below ambient because it is not reading physical heat. Core temp is based on the number of electrons jumping the gaps in the P-N junctions of the transistors. It's a mathematical formula that AMD uses to calculate an equivalent heat value based on how much juice is running around inside the chip. However, by coincidence, as Core electron flow increases, that value will become more and more similar to an actual temperature that is physically inside the silicon and ceramics of the chip, if you could stick a thermometer in it at around 47C.

On a side note:
I have a Phenom II X4 920, and I've never gotten more than 3.4GHz out of it. Basically, it's the non-black-edition 940. I have a phase-change cooler on it with a copper heatsink and 80mm 6000RPM fan. It's always cool, even at full load. Before putting the phase-change in, I was using a Thermaltake 120mm fan and aluminum cooler. It ran pretty hot under load. Different heatsink material, different thermal transfer process, and SMALLER fan. Keep in mind I was not able to change the multiplier and was using the HTT and bus speeds to increase the overclock.

I don't know which cooler you have, but you might consider a phase-change unit before going much further.

 

[CrustyButt]

At low values, Core reads below ambient because it is not reading physical heat. Core temp is based on the number of electrons jumping the gaps in the P-N junctions of the transistors. It's a mathematical formula that AMD uses to calculate an equivalent heat value based on how much juice is running around inside the chip. However, by coincidence, as Core electron flow increases, that value will become more and more similar to an actual temperature that is physically inside the silicon and ceramics of the chip, if you could stick a thermometer in it at around 47C.

On a side note:
I have a Phenom II X4 920, and I've never gotten more than 3.4GHz out of it. Basically, it's the non-black-edition 940. I have a phase-change cooler on it with a copper heatsink and 80mm 6000RPM fan. It's always cool, even at full load. Before putting the phase-change in, I was using a Thermaltake 120mm fan and aluminum cooler. It ran pretty hot under load. Different heatsink material, different thermal transfer process, and SMALLER fan. Keep in mind I was not able to change the multiplier and was using the HTT and bus speeds to increase the overclock.

I don't know which cooler you have, but you might consider a phase-change unit before going much further.

Soo, at this point, what shall I do, I'm kind of clueless at the moment. I don't want to burn the thing, but I also want to OC, just for shiz and giggles.