• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD cpus and their low heat threshold ?

Soldato
Joined
19 Feb 2007
Posts
15,311
Location
London
Ive owned Intel and AMD cpus and im quite baffled as to why Intel chips can quite happily peform under high temps and AMD chips should be kept under 64'c or somewhere around there even when under load.
Shouldnt they have a higher resistance ?
 
Intel temps are the actual core temps, while AMD temps is read from the socket. Hence the actual AMD core temp will usually be higher. If both CPUs had the temp sensors in the same place, I would imagine the heat threshold would be the same.
 
Stupid that AMD don't read the temp from the core itself IMO makes much more sense than giving the 'socket' temperature...

..........not really, as long as you know the limit and are reading the correct temperature sensor.

Does not limit the overclock for a particular cpu.
 
But the temperature difference between the socket and cpu die will vary according to heat output and cooling used, on die sensors make much more sense. IIRC amd does have on die sensors but they don't work properly on the phenom IIs.
 
My 1055T runs at 16-17 degrees idle, 28 load with an A70.

My Q6600 ran at 32 degrees idle, and 50 load with a CoolIT ALC.

It's all relative really.
 
I would expect the actual idle temp to be at least 10 degrees above ambient, you lose about 5 degrees just through the TIM. So you live in a freezing cold house or the reported temp is way off.
 
I would expect the actual idle temp to be at least 10 degrees above ambient, you lose about 5 degrees just through the TIM. So you live in a freezing cold house or the reported temp is way off.

He's monitoring the wrong temperature.
Which is why AMD need to sort this on BD.
 
The intel core sensors are not accurate through the entire temperature range, they dont scale linearly and aren't calibrated. At least that was the case with the core2 cpu's. They dont market their cpu's as having accurate core temperature reporting. There is no real motivation to improve, why waste money on perfecting something that is there to simply monitor the aproach to a temp threshold, and those that are interested in reliable/accurate temp reporting (us) are a very small segment of the market.

At least this was the case when I spent time reading up on this 2/3 years back. For all I know they may have introduced better sensors in the i7 onwards. I'd still take the readings with a hefty pinch of salt.
 
The intel core sensors are not accurate through the entire temperature range, they dont scale linearly and aren't calibrated. At least that was the case with the core2 cpu's. They dont market their cpu's as having accurate core temperature reporting. There is no real motivation to improve, why waste money on perfecting something that is there to simply monitor the aproach to a temp threshold, and those that are interested in reliable/accurate temp reporting (us) are a very small segment of the market.

At least this was the case when I spent time reading up on this 2/3 years back. For all I know they may have introduced better sensors in the i7 onwards. I'd still take the readings with a hefty pinch of salt.
Because the AMD Thuban sensor will report a safe temperature when it could be at a dangerous level.
62c on my old 1055T would have been about 80c in reality, which is 18c over the limit.
 
Because the AMD Thuban sensor will report a safe temperature when it could be at a dangerous level.
62c on my old 1055T would have been about 80c in reality, which is 18c over the limit.

Huh, didn't know that. Definitly something for AMD to work on.



Realise my post seems very intel targeted, but it was more of an open FYI, and advice to be generally distrusful of temp readings.
 
Because the AMD Thuban sensor will report a safe temperature when it could be at a dangerous level.
62c on my old 1055T would have been about 80c in reality, which is 18c over the limit.
the temprature of an individual core doesnt really matter do you think amd are stupid or something? do you realise how large the RND budgets are? do you realise the QA is not done by some bloke in his bedroom?
 
the temprature of an individual core doesnt really matter do you think amd are stupid or something? do you realise how large the RND budgets are? do you realise the QA is not done by some bloke in his bedroom?

I'm sorry what?
AMD's "core temps" are all the same according to core temp, that's wrong from the get go. The actual reading is wrong too.. Thuban's can report under ambient, explain that? ;)
To get the true reading you look at "CPU temperature".
What I think they've done is stupid. On Deneb/Callisto the issue wasn't as apparent, thuban was horrendously off.
 
Because the AMD Thuban sensor will report a safe temperature when it could be at a dangerous level.
62c on my old 1055T would have been about 80c in reality, which is 18c over the limit.

This is misinformation. The first few batches had bad sensors ... the one I own and all the ones I've seen in acquaintences rigs, who've bought them recently, report in line with expected temperatures.
 
This is misinformation. The first few batches had bad sensors ... the one I own and all the ones I've seen in acquaintences rigs, who've bought them recently, report in line with expected temperatures.

It depends on the CPU. I've seen some Thubans be fine and then I've seen some Denebs be slightly whack.
It's not misinformation, it's still the wrong sensor, I'm giving an example why it's a problem.
 
even amd themselves have said to ignore the core temp and to monitor the cpu temp which amd bases all they max safety temps on....

hence why most oc/monitoring utilitys such as AI Suite II show the cpu temp only, not core temp.
 
Last edited:
Back
Top Bottom