It seems like there's a bug in idamax, I'm wondering if anyone can confirm this for me.
I'm trying to call the idamax BLAS function to determine the max absolute value of the elements in an array, but it doesn't seem to give the correct answer. I'm on Windows, using the IFORT32 build, ACML version 4.1 (although I've had the same problem on 3.6).
Has this occurred for anyone else? Can someone duplicate this problem? I just tried a simple 4 element double array, and for various inputs it frequently gives the wrong answer.