the battery life "decrease" might have been software related.
Doubt it, since whenever I used the iPod indoors during the same time, I had no problem with battery life. Plus, isn't cold weather notoriously bad for batteries?
Read that sentence again. I'm telling you, leaving your iPod outside in the cold will drain your battery life faster than you can say "music." I have to make a very conscious effort right now to not leave it in the car after I've driven somewhere because I know that it will significantly drain my battery if I leave it in there.
I imagine it's not much better for laptop batteries. Those suckers aren't cheap, so my suggestion is to never leave your iPod or laptop in the cold for any period of time or else you might have problems.
I think we need the expert opinion of someone knowledgeable in the area. Does anybody feel up to it ?
The comment about conductivity getting better at lower temperatures seems on track, but last thing I heard about superconductivity we were talking about temperatures of at most -183°C. But I could be wrong of course...
conductivity does get better as you lower temperature (in all ranges -- superconductivity is a special property in which there is zero resistance and this has only been observed at fairly low temps) but that isnt solely why computing gets faster. I think for the most part computing speeds up because there is less thermal noise in the processor (ie less changed bits that it has to deal with). If you understand how the n-p or p-n transistors work -- at lower temperatures you expect less electrons or holes populating the conduction band when the transistor is off than you would at higher temperatures. In other words, at high temps you expect some zero bits to become one bits etc etc.
I don't believe colder temperatures would increase speed at the system level (processor, bus and devices, all together).
Sure conductivity goes up or whatever, but computers are clocked. The processor is clocked, the RAM is clocked - everything is clocked. The clocks determine the speed, not the temperature.
Just because the cold temperature would allow faster processing doesn't mean it will just magically happen.
The only way I can see this happening would be if the iBook monitored the temperature and clocked-up when the temperature was low enough. Somehow I doubt that's occurring.
What fraction of the total clock cycles are wasted by an error?
I dont have enough info to provide an answer but it has to do with the number of transistors, their likelihood for an error, and the number of clock cycles before that error is identified and corrected.
This is why people build ridiculous refrigeration and water-cooling systems for their PCs.
not only that, just a mod as simple as 2 88mm fans in the front of the case can push enough air to boot cpu preformance expanintaly (to a point) thats why i wonder why apple didnt put a fan rpm swich on the g5, for compositing, et in fcp, and so on, turn all fans at max rpm for more speed if user chooses
ofcourse more fan rpm = more noise so the normal setting is great for every-day iapp-email-word stuff
Not enough errors within usable operating temperatures to make a difference.
If steves computer came in from the cold at lets say -10 C (263 K) and was operating at that temperature when it was noted it was "zippier" as compared to the normal operating temperature of around 60 C (333 K). The relative population (probability) of electrons to be found in the in the conduction band of a transitor with a band gap of 1.5 eV (about the middle of the road as far as these things are concerned) is very small on the order of 10^-20. However, the relative populations between these two temperatures is a much more significant number indicating the relative likelihood that an error would occur. At 333 K each transitor is 1 million times more likely to have an electron in its conduction band than a transitor at 263 K. That means roughly that at normal operating temperatures there is a million times greater chance that a transitor will be on when it should be off than at an operating temperature of 263 K. That I would say could be noticeable.
Lets assume for the sake of argument it is 40 C that still gives a 40000 fold greater chance of finding an electron in the conduction band at that temp.
not only that, just a mod as simple as 2 88mm fans in the front of the case can push enough air to boot cpu preformance expanintaly (to a point) thats why i wonder why apple didnt put a fan rpm swich on the g5, for compositing, et in fcp, and so on, turn all fans at max rpm for more speed if user chooses
ofcourse more fan rpm = more noise so the normal setting is great for every-day iapp-email-word stuff
PC Overclockers like me don't control "CPU performance" by adding fans. You increase performance by boosting the clockrate (speeding up the FSB, or changing the clock multipliers). Many overclockers push their PCs well past their limits and do see errors. But then again, are you going argue that an overclocked CPU seems slower than one operating within spec?
I have no way of knowing what temperature the processor was at, especially since i dont know what conditions (humidity, whether it was asleep or not etc etc) the computer was under. So I stated this assumption when I initially calculated the values. Fine. But if my arguments thus far havent been convincing, try this: at 10 C difference lets say the processor at 303 K when it came in versus 313 K normal operating temperature that would still correspond to nearly 10 fold difference in conduction band occupancy. Again, 10 fold is significant.
So keep on laughing, at least I can read and stated my assumptions.
Comments
Originally posted by billybobsky
the battery life "decrease" might have been software related.
Doubt it, since whenever I used the iPod indoors during the same time, I had no problem with battery life. Plus, isn't cold weather notoriously bad for batteries?
Read that sentence again. I'm telling you, leaving your iPod outside in the cold will drain your battery life faster than you can say "music." I have to make a very conscious effort right now to not leave it in the car after I've driven somewhere because I know that it will significantly drain my battery if I leave it in there.
I imagine it's not much better for laptop batteries. Those suckers aren't cheap, so my suggestion is to never leave your iPod or laptop in the cold for any period of time or else you might have problems.
The comment about conductivity getting better at lower temperatures seems on track, but last thing I heard about superconductivity we were talking about temperatures of at most -183°C. But I could be wrong of course...
Sure conductivity goes up or whatever, but computers are clocked. The processor is clocked, the RAM is clocked - everything is clocked. The clocks determine the speed, not the temperature.
Just because the cold temperature would allow faster processing doesn't mean it will just magically happen.
The only way I can see this happening would be if the iBook monitored the temperature and clocked-up when the temperature was low enough. Somehow I doubt that's occurring.
Originally posted by billybobsky
its a difference of sucessful cycles versus non sucessful cycles. processors make errors.
Not enough errors within usable operating temperatures to make a difference.
I dont have enough info to provide an answer but it has to do with the number of transistors, their likelihood for an error, and the number of clock cycles before that error is identified and corrected.
Originally posted by Brad
This is why people build ridiculous refrigeration and water-cooling systems for their PCs.
not only that, just a mod as simple as 2 88mm fans in the front of the case can push enough air to boot cpu preformance expanintaly (to a point) thats why i wonder why apple didnt put a fan rpm swich on the g5, for compositing, et in fcp, and so on, turn all fans at max rpm for more speed if user chooses
ofcourse more fan rpm = more noise so the normal setting is great for every-day iapp-email-word stuff
Originally posted by Eugene
Not enough errors within usable operating temperatures to make a difference.
If steves computer came in from the cold at lets say -10 C (263 K) and was operating at that temperature when it was noted it was "zippier" as compared to the normal operating temperature of around 60 C (333 K). The relative population (probability) of electrons to be found in the in the conduction band of a transitor with a band gap of 1.5 eV (about the middle of the road as far as these things are concerned) is very small on the order of 10^-20. However, the relative populations between these two temperatures is a much more significant number indicating the relative likelihood that an error would occur. At 333 K each transitor is 1 million times more likely to have an electron in its conduction band than a transitor at 263 K. That means roughly that at normal operating temperatures there is a million times greater chance that a transitor will be on when it should be off than at an operating temperature of 263 K. That I would say could be noticeable.
Whether it is or not I cant say.
Originally posted by billybobsky
normal operating temperature of around 60 C (333 K).
That seems a bit too high. I'd say it's more like 40
Billybobski, your comments seem to make lots of sense. Where did you get the numbers ? Just so I could do some reading, not to negate your comment.
Actually all of the numbers/equation I used can be found here: Nice website describing band gap etc
Originally posted by Eugene
-10C? +60C? Why are you mixing environmental temperatures with die temps?
Read the first post.
Originally posted by a_greer
not only that, just a mod as simple as 2 88mm fans in the front of the case can push enough air to boot cpu preformance expanintaly (to a point) thats why i wonder why apple didnt put a fan rpm swich on the g5, for compositing, et in fcp, and so on, turn all fans at max rpm for more speed if user chooses
ofcourse more fan rpm = more noise so the normal setting is great for every-day iapp-email-word stuff
PC Overclockers like me don't control "CPU performance" by adding fans. You increase performance by boosting the clockrate (speeding up the FSB, or changing the clock multipliers). Many overclockers push their PCs well past their limits and do see errors. But then again, are you going argue that an overclocked CPU seems slower than one operating within spec?
Originally posted by billybobsky
Read the first post.
So you think the junction temperature was anywhere near -10C while his Mac was on?
So keep on laughing, at least I can read and stated my assumptions.