Intel has found a bug in these processors. <https://www.phoronix.com/scan.php?page=article&item=intel-jcc-microcode&num=1> The bug must be commonly triggered because I've noticed no reports. Their fix is to patch the microcode in such a way as to turn off one of their accelerations for any conditional jump that ends on or crosses a 32 byte boundary. They claim that this microcode patch will slow down programs somewhere in the range 0-4%. Furthermore, Intel has released changes to GAS (GCC's assembler) that try to avoid the slowed down JCC by adding padding to the program. This too will slow down the program, but less. They have not released changes for the LLVM assembler. Nor for other things that generate x86-64 code. I wondered if the program changes would slow down AMD processors. Apparently not: <https://www.phoronix.com/scan.php?page=news_item&px=AMD-With-Intel-JCC-Assembler> It surely will have increased program size.