Intel Found A Velocity Boost Bug While Investigating CPU Instability, BIOS Fix Coming
Igor's Lab reported, based on what was said to be insider info, that Intel actually had found the root cause of the problem and that the chipmaker was working on a fix. In response, Intel posted up a blog post on its community site that explains the situation as it is understood. What the company found, it turns out, is simply a "confirmed contributing cause," not the actual root cause.
This contributing cause is in fact a bug in the Enhanced Thermal Velocity Boost (eTVB) feature of the top-end Core i9 parts. Only the Core i9s have this feature, which is designed to allow the CPU to hit even higher clock rates than with the normal boost algorithm as long as thermal headroom allows. Intel doesn't explain exactly what the bug is, but we can assume it has to do with improper voltage/frequency scaling.
We can assume that because, as we noted before, the enthusiast community has already determined the practical cause of the crashes: insufficient CPU voltage. By simply setting a voltage offset on your Core i9 CPU, you can completely avoid the crashes—albeit at the risk of damaging your processor over a long period. What Intel's trying to investigate is why these CPUs aren't getting enough voltage in the first place.
Improving cooling is one way to help work around this, but the ultimate fool-proof solution is really to reduce clock rates. Of course, this impacts performance, and Intel could be in hot water if it can't get these chips fully stable at their rated clocks. Still, the new eTVB patch (shipping out to motherboard vendors and being delivered in the form of BIOS updates as we write this) should help stability on Core i9 CPUs, at least.
