It wasn't a question of not-optimizing for non-Intel compilers, the compiler detected AMD chips and deliberately slowed down the code. They had previously got

It wasn't a question of not-optimizing for non-Intel compilers, the compiler detected AMD chips and deliberately slowed down the code. They had previously got caught detecting a particular set of benchmark code and generating pre-calculated results.