Validating UTF-8 In Less Than One Instruction Per Byte

Created: 2021-10-26T00:20:59-05:00

Return to the Index

This card pertains to a resource available on the internet.

Testing a range of ASCII chars by loading them in to SIMD instruction, AND'ing by 0x8080(cont'd) to see if anything in the range has a header flag.

Folding two words of ASCII by OR'ing two eight byte sections then AND'ing the remaining word by 0x8080(cont'd)

Cramming validation in to a state machine

Taking half of a byte from each byte in a two byte set then looking the pattern up in a dictionary to see if it matches a known probable error pattern.