ObscureParadox Posted February 5, 2014 Posted February 5, 2014 Right I have been wondering this for a while and I never really knew the answer to it. Can anyone tell my exactly what causes a CBB or CB to happen on one chip and not on the other and what the cause of said cold bug may be. I don't mean the whole "it turns it into a super conductor thing" I mean why is it that you can have a 22nm Ivy CPU that might cold bug at -230c with LHe and then you get a 22nm Haswell CPU made from the same sort of tech that might cold bug at 100c lower than that (-130c). If anyone knows why or has a rough idea then please feel free to post Lewis. Quote
Moose83 Posted February 5, 2014 Posted February 5, 2014 Haswell has the power converters in the Die of the chip, thats why they dont like cold On Ivy the power converters are on the Mainboard. Quote
ObscureParadox Posted February 6, 2014 Author Posted February 6, 2014 I thought that might have been one of the reasons for the issue with Haswell, but then again if we were to go a gen back and compare Sandy to say Clarkdale or Gulftown and the difference again is huge! Quote
Calathea Posted February 6, 2014 Posted February 6, 2014 It's even more interesting when we compare bad Phenom 2s, I hade one x2 555 BE that CB around -120 while most worked good even on LHe. I'd also like some more technical explanation to this Quote
der8auer Posted February 6, 2014 Posted February 6, 2014 I think there is no quick answer on this question and as long as somebody has worked at Intel and investigated this issue you can never be sure if the explanation is correct or not. I can only guess but I think the reason of a cold bug is related to the basic materials used and process parameters. Every generation has a different basic wafer in terms of silicon doping, different alloys deployed on top of it and so on. With a decreasing temperature the resistance of silicon is increasing. It could be that at a certain point some parts are just not working anymore. Quote
FireKillerGR Posted February 6, 2014 Posted February 6, 2014 With a decreasing temperature the resistance of silicon is increasing. I thought it was the opposite, the lower the temp is the lower the resistance will be. Thats why @the absolute 0 (0 Kelvin) the is no resistance Quote
der8auer Posted February 6, 2014 Posted February 6, 2014 Silicon is a semiconductor and therefore the resistance increases with a lower temperature. That's the reason why CPUs die at high volts and temperatures. At a certain temperature silicon turns from a semiconductor to a conductor with almost 100% conductivity -> short -> CPU is dead. I can't remember the exact temperature but it's somewhere between 150-200°C if I remember correctly. Don't forget that a CPU is not "made" of silicon. Silicon is just the base material which you use to create layers on top. A big part are traces which are made of copper and copper behaves the way you said. Lower temperature -> lower resistance. Transistors are made of silicon tho. Quote
Crew Vivi Posted February 6, 2014 Crew Posted February 6, 2014 Nick shih told me haswell's situation is different, he said that the reason all the cpu's have such close CB's is because its not CB's, its some temperature protection trip that intel built into the CPU. Ofcourse they could shut it down with cool software or tools but intel made a mistake and made it impossible to access or change the protection. Leaving us stuck at -130. The next release will have this "issue" fixed so we "should" see no cold bug on broadwell Quote
der8auer Posted February 6, 2014 Posted February 6, 2014 There is a lot of blah blah going on in this industry. As long as I don't get a logic explanation from one of the Intel engineers I won't belive anything Quote
ObscureParadox Posted February 6, 2014 Author Posted February 6, 2014 Nick shih told me haswell's situation is different, he said that the reason all the cpu's have such close CB's is because its not CB's, its some temperature protection trip that intel built into the CPU. Ofcourse they could shut it down with cool software or tools but intel made a mistake and made it impossible to access or change the protection. Leaving us stuck at -130. The next release will have this "issue" fixed so we "should" see no cold bug on broadwell Is it not fixable in a similar fashion to the way we used to do CB mods on P55 and H55 motherboards or is it something built into the die itself hence why some of the ES models saw absolutely no CB? Quote
Calathea Posted February 7, 2014 Posted February 7, 2014 ...or is it something built into the die itself hence why some of the ES models saw absolutely no CB? Don't think this is correct, nobody have posted results with CB-free Haswell. Quote
FireKillerGR Posted February 7, 2014 Posted February 7, 2014 Don't think this is correct, nobody have posted results with CB-free Haswell. exactly, had a review sample 4 months ago which had cb @-105 C Quote
ObscureParadox Posted February 7, 2014 Author Posted February 7, 2014 My mistake, for some reaspn I was under the assumption that the 7GHz chips we saw the likes of Fredyama use had no cold bug for some reason. Quote
d0minat0r Posted February 7, 2014 Posted February 7, 2014 Nick shih told me haswell's situation is different, he said that the reason all the cpu's have such close CB's is because its not CB's, its some temperature protection trip that intel built into the CPU. Ofcourse they could shut it down with cool software or tools but intel made a mistake and made it impossible to access or change the protection. Leaving us stuck at -130. The next release will have this "issue" fixed so we "should" see no cold bug on broadwell thanks for sharing Quote
Splave Posted February 7, 2014 Posted February 7, 2014 I have heard rumors that it is possible haswell will not have CB on z97 motherboards, dont take my word for it though Quote
ObscureParadox Posted February 7, 2014 Author Posted February 7, 2014 I have heard rumors that it is possible haswell will not have CB on z97 motherboards, dont take my word for it though Would that not contradict what Nick said though? I am genuinely interested in this and glad I started the thread now. Quote
Moose83 Posted March 3, 2014 Posted March 3, 2014 Don't think this is correct, nobody have posted results with CB-free Haswell. I saw Bullshooter benching 4670k with -170 Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.