defter wrote:
hmm, in that case, would you say temperatures alone can degrade CPUs? if so, how?
whether it took 2 years or 2 days, i've always thought current would be the main reason behind most dead chips that were overclocked (unless temperatures were extremely high 85C+ [for ivy bridge] for very long periods of time).
You are correct. What kills or degrades a processor rapidly is heavy loads at elevated voltages, where the current becomes excessive. This is exacerbated at higher operating frequencies as the current drawn is proportional to operating frequency - higher operating frequency needs elevated voltage which is the potential for current to flow.
People are mistaken when they state only one or two of these contributors as the main factor. Voltage, current and temps can lead to degradation. However voltage and heat are
not as significant as current.
i) Consider a CPU running at stock settings on a stock cooler at full load. If heat alone were an issue, the CPU would degrade rapidly.
ii) Consider an elevated voltage, and a light load. As the current being drawn is not high, we cannot say that voltage alone leads to or is the primary contributor to degradation. Even at high operating frequencies. Case in point for those that benchmark - we know that the "heavy load" benchmarks degrade a CPU faster than a light load benchmark. Why? Because more current is drawn.
iii) Consider the fact that the stock VID on these CPUs is variable - yet they all meet Intel's TDP. The variance in current draw between a high VID CPU and a low VID CPU is within a couple of watts most of the time. Hence the warranty offered on a high VID CPU versus a low VID one is the same.
The gap does widen as we increase the operating frequency as does the disparity in the level of voltage required between two such CPUs. Still, that change is not completely significant as quite often the CPUs will draw similar levels of current at a given frequency, regardless of how much voltage each of them needs to be stable (maybe 20W between the two samples at the kind of OCs we run on air and water).
The rest of this relationship is down to variables that fall outside my scope of education though I know enough to understand the elementary. Either way, not something I want to try and fudge my way through wrongly on a forum like a muppet.
We can say for certain that there is a relationship between required voltage and frequency - bearing in mind that the change is the amount of current that flows. Depending upon the temperature co-efficient of the substrate, heat can play a part in elevating the required current. I guess we need to understand that voltage is potential, while current is the name we give to flow of that potential. Also worth understanding that P=VA...
Of course, you could brick a CPU on purpose if you wanted; running it out of spec by grossly abusing any three of these contributors. However, as we're talking about the way we use these CPUs while overclocking (applying
required voltages) - it's current we need to fear most.
-Raja
.jpg "Raja"){kind=avatar}
