"ARM Architecture Reference Manual" yields quite a few hits for various ARM systems, including ARMv8; the Raspberry Pi's ARMv6 instruction set is also available.
One big change is that modern ARM processors tend to run mostly in Thumb (16-bit instruction) mode. It's faster, but you lose all the fun condition codes. High-performance code should take advantage of NEON.
If the above does not answer your question, please ask a more specific one. ;-)
One big change is that modern ARM processors tend to run mostly in Thumb (16-bit instruction) mode. It's faster, but you lose all the fun condition codes. High-performance code should take advantage of NEON.
If the above does not answer your question, please ask a more specific one. ;-)