BASIC VI using VFP instructions for floating point arithmetic
|Theo Markettos (89) 650 posts||
Someone over on the Raspberry Pi forum was enquiring about floating point support. There was a suggestion about modifying BASIC VI to use VFP instructions rather than old FPU instructions, which would give a considerable speedup (on float code).
That doesn’t look too hard… the BASIC V/BASIC VI difference appears to be a compile-time flag. The relevant source files are fp.s and fp2.s . Those just have routines that either call an FPU instruction, or emulate it. So rewriting those to use VFP instructions probably wouldn’t take too long and would have decent speed benefits.
(Use of VFP in the BASIC assembler being something unrelated, of course)