Although there is a widely-known theory that processor will never burn if it runs cool, it seems that this time this idean does not work. First of all, microprocessors are very complex and contain loads of separate blocks nowadays; for example, the temperature of FPU can be lower than a temperature of ALU when working in office applications. Secondly, manufacturing technologies are becoming thinner and thinner these days, therefore their working conditions should be determined very precisely and any deviations can lead to incorrect work or malfunction. Finally, computer hardware cannot be considered as “toys for big boys” – you should work carefully with it, just as you did with your first i486-based PC. With the so-called Sudden Northwood Death Syndrome (S.N.D.S), all those who have forgotten the following statements definitely revised them.
Now let us take a look from another point of view. There are a lot of explanations of the reason why Intel Pentium 4 “Northwood” processors malfunction even though they run at lower than 45 degrees Celsius average temperature, according to thermal diodes located inside the CPU. Honestly speaking, only Intel engineers may suspect the real source of the problem and maybe even localise it if it is possible. We can only make intentions and maybe some of them appear to be real. It seems that the roots of the issue are in the manufacturing process itself as well as the hot-spots inside the CPU that make their appearance when the processor functions in unusual conditions. With thinner fabrication processes, the possibility of the so-called electromigration effect increases drastically. Electromigration is generally considered to be the result of momentum transfer from the electrons, which move in the applied electric field, to the ions which make up the lattice of the interconnect material. When electrons are conducted through a metal, they interact with imperfections in the lattice and scatter. Scattering occurs whenever an atom is out of place for any reason. Thermal energy produces scattering by causing atoms to vibrate. This is the source of resistance of metals. The higher the temperature, the more out of place the atom is, the greater the scattering and the greater the resistivity. Moreover, according to people with knowledge of the matter, electrical voltage influences the atomic vibration even more than the temperature, as a result if you start to increase the voltage, there will be a lot more chances for your Pentium 4 CPUs to start working incorrectly and die after all. So, even if the average temperature of the core is relatively low, there may be some hot spots inside the CPU that can lead to the electromigration effect. The problem was not so widely spread before because semiconductor manufacturing processes were less thin and electromigration effect was not a common thing at all even with very high voltages.
You now can call it as Sudden Overclocked Northwood Death Syndrome, but you should understand that from this point you will start to hear about such effects more and more often. When Intel and AMD start to utilise the 90 nanometer process in 2003 and 2004 respectively, a big challenge for them will be the electromigration effect because it is exponential and depends on manufacturing process, voltage, heat, the quality of material and some other factors. All the mentioned factors influence the core-speed and the latter influence the performance. So, it all form the cycle and CPU developers have to find a consensus between the speed and reliability. In fact, it seems to be found for the current Northwood processors, but when you go above the recommendations, it may cause your CPU to burn down.
I wonder if the same things may happen to AMD’s Athlon XP processors that are manufactured using 0.13 micron process and that are sometimes overclocked with the core voltage raised to 2.0V and above from the nominal 1.5 to 1.65V.
- - Sudden Northwood Death Syndrome Discussion at X-bit Labs Discussion Board.
- Introduction in Electromigration