Hardware Intel Skylake bug causes PCs to freeze during complex workloads

[J65nko]

From http://arstechnica.com/gadgets/2016/...lex-workloads/

Bug discovered while using Prime95 to find Mersenne primes.

Intel has confirmed that its Skylake processors suffer from a bug that can cause a system to freeze when performing complex workloads. Discovered by mathematicians at the Great Internet Mersenne Prime Search (GIMPS), the bug occurs when using the GIMPS Prime95 application to find Mersenne primes.

Update: We've been informed that the bug was reportedly discovered and tested by the the community at hardwareluxx.de before being passed onto GIMPS, which conducted further testing. Both groups passed their findings onto Intel.

[e1-531g]

It would be interesting to see whether Skylake Refresh will be free from that bug.

[Carpetsmoker]

Lifted from the comments section:

Originally Posted by fivemack

There is a hardware design concept of 'chicken bits' - if as a hardware designer you have added a feature which you're pretty certain works, but are not absolutely confident you have tested perfectly, you add a not-publicly-documented bit in a control register that turns it off. The BIOS update would arrange to set that bit at boot time. It is entirely conceivable that this will reduce performance on some workloads - a typical chicken bit might be 'always pick the first two entries in the queue of instructions with their operands ready, rather than using a clever method to pick the most efficient pair'.

The Pentium FDIV bug was due to a wrong entry in a look-up table in ROM; in 1993 it would have been absurd to have 8192 chicken bits so that every entry in the look-up table can be corrected, and so there had to be a recall.

The problem is apparently in the "HyperThreading plus the AVX1 pre-FMA3 code-path of the benchmark"; so if your CPU doesn't have HT you should be fine (the i5-6400, which is probably the most popular desktop Skylake CPU and what I have, doesn't have HT).