- Joined
- Sep 7, 2003
- Location
- Warsaw, Poland
Gnerma - in my opinion such risk exists, however...
Ok, here we start. As we all have access to sites like www.amd.com or www.intel.com, we can read a lot. Personally, I'm after reading whole formfactors.org and big part of 2 I mentioned earlier. Each CPU has given working conditions - they tell what temperature and voltage is ok.
Some "overclockers" were saying "oh noes, your cpu will get fried" when I was thinking that 2.25v / >60°C is ok for Palomino. It used to run earlier at stock voltage at temperatures close to 90°C before I learnt that there should be fans in case. And it was according to datasheets. CPU still runs and still OCes well (as for Palomino).
I've seen some reports that OC possibilites decreased after long-term serious overvolting without proper cooling, however nothing got fried.
Problem is with modern CPUs. When we check their tolerance to temperature and voltage, it's just a fraction of what older CPUs can stand. As most of us don't overvolt till it starts to glow, let's focus on temperature. I'm not really into very old CPUs, but I'll try to show it on example of some AMD CPUs, so here are some max temps, data given by manufacturer:
AXP Palomino: 90°C
AXP Barton: 85°C
A64 0.13u: 70°C
A64 0.09u: 63°C
and it still decreases, so finding CPU with max temp given in datasheet <60°C shouldn't be anything unusual.
What does it mean?
Modern CPUs produce more and more heat, while they have lower max temp (theoretical, as I haven't found reports about frying them). They require better and better cooling - check intel datasheets, max °C/W that they recommend for coolers is VERY low.
Frying older CPUs is something difficult, but I think that with stuff like top AMD or intel CPUs, including dual cores, we have to deal with such possibility.
So before we say "omg, 40°C on your CPU, that's hot" or "60°C? nothing to worry about", let's check datasheets and compare long-term load temperatures.
Will running CPU > maximum temperature given in datasheet make it fry?
I don't think so, at least not when CPU was exposed to such conditions for short period and they were close to maximum theoretical values. It may reduce life time of CPU, so I'd try to keep components in ranges that manufacturer says to be safe.
The Coolest has made nice application that can read max rated temperature from A64 CPUs, so if you have such CPUs, you may try it out.
Maybe we should make something like list of maximum rated temperatures for CPUs and other stuff, like GPUs, memory chips etc. It's also worth keeping in mind that we often focus on cooling CPU, GPU etc forgetting about other components. Some of us put active cooling on HDDs, RAM on video cards, system RAM, NB, but what with rest of chips? There are lots of components on motherboards and graphics cards that haven't got heatsink, that don't draw much power and don't produce much heat but... But they still produce some heat. If we don't cool something and it constantly produces heat, temperature increases. So we have some shocking "temperature photos" of mobos showing that some components are really hot while we perhaps haven't even known that they existed. That's one of reasons why I'm against "passive" cooling.
So ending this a little too long post, I'd like to suggest checking maximum temeprature of your components that manufacturer says to be ok. Also have some rational case airflow to keep all your components in proper conditions.
Ok, here we start. As we all have access to sites like www.amd.com or www.intel.com, we can read a lot. Personally, I'm after reading whole formfactors.org and big part of 2 I mentioned earlier. Each CPU has given working conditions - they tell what temperature and voltage is ok.
Some "overclockers" were saying "oh noes, your cpu will get fried" when I was thinking that 2.25v / >60°C is ok for Palomino. It used to run earlier at stock voltage at temperatures close to 90°C before I learnt that there should be fans in case. And it was according to datasheets. CPU still runs and still OCes well (as for Palomino).
I've seen some reports that OC possibilites decreased after long-term serious overvolting without proper cooling, however nothing got fried.
Problem is with modern CPUs. When we check their tolerance to temperature and voltage, it's just a fraction of what older CPUs can stand. As most of us don't overvolt till it starts to glow, let's focus on temperature. I'm not really into very old CPUs, but I'll try to show it on example of some AMD CPUs, so here are some max temps, data given by manufacturer:
AXP Palomino: 90°C
AXP Barton: 85°C
A64 0.13u: 70°C
A64 0.09u: 63°C
and it still decreases, so finding CPU with max temp given in datasheet <60°C shouldn't be anything unusual.
What does it mean?
Modern CPUs produce more and more heat, while they have lower max temp (theoretical, as I haven't found reports about frying them). They require better and better cooling - check intel datasheets, max °C/W that they recommend for coolers is VERY low.
Frying older CPUs is something difficult, but I think that with stuff like top AMD or intel CPUs, including dual cores, we have to deal with such possibility.
So before we say "omg, 40°C on your CPU, that's hot" or "60°C? nothing to worry about", let's check datasheets and compare long-term load temperatures.
Will running CPU > maximum temperature given in datasheet make it fry?
I don't think so, at least not when CPU was exposed to such conditions for short period and they were close to maximum theoretical values. It may reduce life time of CPU, so I'd try to keep components in ranges that manufacturer says to be safe.
The Coolest has made nice application that can read max rated temperature from A64 CPUs, so if you have such CPUs, you may try it out.
Maybe we should make something like list of maximum rated temperatures for CPUs and other stuff, like GPUs, memory chips etc. It's also worth keeping in mind that we often focus on cooling CPU, GPU etc forgetting about other components. Some of us put active cooling on HDDs, RAM on video cards, system RAM, NB, but what with rest of chips? There are lots of components on motherboards and graphics cards that haven't got heatsink, that don't draw much power and don't produce much heat but... But they still produce some heat. If we don't cool something and it constantly produces heat, temperature increases. So we have some shocking "temperature photos" of mobos showing that some components are really hot while we perhaps haven't even known that they existed. That's one of reasons why I'm against "passive" cooling.
So ending this a little too long post, I'd like to suggest checking maximum temeprature of your components that manufacturer says to be ok. Also have some rational case airflow to keep all your components in proper conditions.