Thuban safe load temps?

[Polizei]

A friend says 50C is pretty much melting for AMD, but I don't think so. What are max load temps for my 1055T on air? I'm currently running Small FFTs in prime95 and it's loading at 53C.

Side note, is 1.325v too much for 3500mhz?

[richierich]

Yeah, I can run 1.25v for 3.5GHz/2.5GHz NB, prime blend stable.

[PaganII]

Melting? I started running Bionic Friday, Temp was 51C and dropped to 48C after TIM burn in. Still running 100%.

1090T is 62C http://products.amd.com/en-us/DesktopCPUDetail.aspx?id=640&f1=&f2=&f3=&f4=&f5=&f6=&f7=&f8=&f9=&f10=&f11=&f12=

720 is 73C http://products.amd.com/en-us/DesktopCPUDetail.aspx?id=522&f1=&f2=&f3=&f4=&f5=&f6=&f7=&f8=&f9=&f10=&f11=&f12=

[zir_blazer]

Those values are TCase Max, that had a totally different meaning (Though never understanded good enough what it is precisely used for). My old A64 3000+ Venice DH-E6 had a TCase Max of 53°C yet due to heavy dust and issues with my silicon grease (After the first time that I cleaned it and removed the Thermal Pad) it run many times at temperatures as high as 80°C Full Load, measured with Core Temp. The maximum temperature is that when isn't stable anymore or Thermal Throttling kicks in, that should kick in before the Processors get damaged.

I also recall 8 or 9 years ago that my K6-II 500 MHz on a PC Chips Motherboard was totally unstable, that besides being overclocked to 550 MHz in the very own computer shop that I purchased it at (I purchased a 550 MHz, not 500) it has a joke of a Heatsink and never remember that thermal grease was been applied to it (Wasn't a common practice at that time), also happened to my cousin on another shop when he purchased his. Results? Everytime that it crashed, I rebooted it, enter the BIOS, then saw the Processor temperature sensor at 100°C. "Must be a joke". Touched the Heatsink, and it hurted. Then I remember from a even previous, similar experience, with an unstable Pentium 233 MHz MMX that sometimes rebooted itself until I noticed that when the BIOS started in safe mode factory settings, it booted at 120 MHz and was fully stable at that, so I decided to underclock it to 166 MHz from BIOS and never had an issue anymore (Until I upgraded it to my K6-II). So I underclocked/undervolted it to 400 MHz @ 2.0V against 500 MHz @ 2.2V. Temperatures didn't rise above 80°C or so with that joke of an Heatsink, but the machine was finally FULLY stable. A year or so later I replaced the Heatsink with one that my first Athlon XP Palomino had (That fitted both Socket 7 and A), and was ridiculous cold (Less than 40°C).
What I learned from these two computers is that temperature is severely overrated when most people worries when it runs at 60°C and issues usually appears at much higher than that. Exceptions include overclocking though, because at higher temperatures the silicon is less conductive so you have less potential to scale on Frequencies.

[wez]

50-55C is totally fine :)

[[XC] Oj101]

My 965 has spent its life at 58'c running WCG, throughout the year I adjust clockspeeds to sit at as close to 58'c as possible.

zir_blazer, are you sure about your K8 temperatures? I have enough stability to MAAAAYBE get into Windows for a few seconds above 70'c.

[Polizei]

Yeah, I can run 1.25v for 3.5GHz/2.5GHz NB, prime blend stable.

Hmm. I just got 3500mhz stable (4 hours prime blend) last night at 1.325 Core VDD, and like 1.367 vCore. I'll have to bump it down.

Still trying to figure out what voltages are safe for CPU-NB VDD, CPU-NB, NB and SB...

[zir_blazer]

My 965 has spent its life at 58'c running WCG, throughout the year I adjust clockspeeds to sit at as close to 58'c as possible.

zir_blazer, are you sure about your K8 temperatures? I have enough stability to MAAAAYBE get into Windows for a few seconds above 70'c.
It depends on Frequency too, due to silice properties. Take for example extreme overclockers that attempt to cool the Processor as much as they can, this is because silice conducts better at lower temperatures and in our world that is translated to higher Frequencies.
I recall having seeing a graph of some sorts that tested Frequencies archivable at a determinated Voltage and temperature, and the lowest the temperature, the higher Frequency you was able to reach at the same Voltage. This holds true the other way around too, the higher the temperature, the lower the Frequency that it can work at. In my case, the Venice DH-E6 was at 1.8 GHz, so is entirely possible that if you are running at higher Frequencies (A Processor of a higher model) the maximum stable temperature starts to gets lower.

[richierich]

Hmm. I just got 3500mhz stable (4 hours prime blend) last night at 1.325 Core VDD, and like 1.367 vCore. I'll have to bump it down.

Still trying to figure out what voltages are safe for CPU-NB VDD, CPU-NB, NB and SB...

Dunno but I've found that I can run lower cpu-NB and NB voltage than Deneb. Currently at 3.6GHz / 2.6GHz NB folding 1.3v, NB 1.1v, CPU-NB 1.2v.

[informal]

62C should be Tcase max.

[FlanK3r]

yes 62 C. Is the same as C3 Deneb 955/965

[[XC] Oj101]

That doesn't mean that you can keep it there, however. At 60'c I get errors on WCG whereas at 58'c I'm stable. AMD's are very sensitive to heat and have been post Clawhammer (that thing was hot!).

[FlanK3r]

its very strange problematic, some CPUs (example 965 Denebs) are diferent, someone erors at 55 C + and others can 65 C without problems. My friend has x4 940 and Primed it at 69 C OK.

[Manicdan]

i installed windows with the pump off and noticed it when bubbles appeared on the cpu block. so i know that 100C is not going to kill a chip (i dont know how, i just know my 940 is still alive and kicking, but NOT RECOMMENDED)

for safe OCs i try to keep it under 50C when possible, above that i get instability alot easier

[BeepBeep2]

I error at anything over 50c...it depends on the CPU. I know a few guys that can do 60c + and good clocks on their cpu's.

[[XC] Oj101]

i installed windows with the pump off and noticed it when bubbles appeared on the cpu block. so i know that 100C is not going to kill a chip (i dont know how, i just know my 940 is still alive and kicking, but NOT RECOMMENDED)

for safe OCs i try to keep it under 50C when possible, above that i get instability alot easier

I can vouch for this, I've switched my pump off my accident and tried booting several times before noticing. I could get as far as the BIOS and see temperatures in the high 90'c range.

[ReKcOlNu]

Those values are TCase Max, that had a totally different meaning (Though never understanded good enough what it is precisely used for).

Tcase is the temp at the center of the metal lid on the CPU. One of its main uses is for heatsink makers, in one of amd tech docs they show how its done, they drill a hole near the bottom of the heatsink to the center and insert a temp probe, then use a mathematical formula to figure out what the Tcase of the CPU actually is. Under testing with an ambient temp of 41C the heatsink should be able to keep the CPU at or below the Tcase max rating.

As far as reading the CPU temp on current AMD chips goes... I emailed AMD about that and was told to monitor the single CPU temp reading and not the core temps.

[PaganII]

Using the stock cooler that came with the 1090T the CPU temp as reported by Asus Probe never budges from 48C @ 100% load 3.2Ghz even at today's 35 C temp. Does not matter if 20 C or 35 C only fan speed changes. Today's fan speed was 4700 rpm.