They said the cause was a bug in the microcode making the CPU request unsafe voltages:
Our analysis of returned processors confirms that the elevated operating voltage is stemming from a microcode algorithm resulting in incorrect voltage requests to the processor.
If the buggy behaviour of the voltage contributed to higher boosts, then the fix will cost some performance. But if the clocks were steered separately from the voltage, and the boost clock is still achieved without the overly high voltage, then it might be performance neutral.
I think we will know for sure soon, multiple reviewers announced they were planning to test the impact.
This keeps getting slightly misrepresented.
There is no fix for CPUs that are already damaged.
There is a fix now to prevent it from happening to a good CPU.
But isn’t the fix basically under clocking those CPU?
Meaning the “solution” (not even out yet) is crippling those units before the flaw cripples them?
That was the first “Intel Baseline Profile” they rolled out to mobo manufacturers earlier in the year. They’ve roll out a new fix now.
They said the cause was a bug in the microcode making the CPU request unsafe voltages:
If the buggy behaviour of the voltage contributed to higher boosts, then the fix will cost some performance. But if the clocks were steered separately from the voltage, and the boost clock is still achieved without the overly high voltage, then it might be performance neutral.
I think we will know for sure soon, multiple reviewers announced they were planning to test the impact.
Remember Spectre? When they recommended disabling hyperthreading?