Front page | perl.perl5.porters |
Postings from December 2013
Karl Williamson wrote: > ISTR ax and ix being hardware register names in some assembly language > on some platform On x86, ax is the 16 bit version (union) of general purpose register "A", as in AL, AH, EAX and RAX. The "A" stands for Accumulator but I'm not sure if that is an backronym or a nickname that became legitimized. There are some x86 opcodes that can be written in a shorter byte sequence when using EAX vs any GPR. There is also a SI/DI (now extended to ESI/EDI/RSI/RDI) called Source Index and Destination Index. SI/DI have a strlen() and memcpy() clone opcode that requires the pointers to be in SI/DI. Most literature says never to use the strlen/memcpy instruction because they are "slow", "deprecated", and "emulated in microcode, not silicon". A small amount of literature says to use them on AMD64 CPUs vs doing the algorithm using more modern ops and that they are faster on AMD64 than mov, test, increment, jmpcondition. Google says IX is a Z80 register that contains a pointer to memory address, on which to do an op. The op doesn't take a register from a pool, it only works on IX. I'm not sure if this influenced P5 in any way.Thread Previous | Thread Next