Hi, On Thu, Aug 22, 2024 at 6:14 PM tom ehlert via Freedos-user <freedos-user@lists.sourceforge.net> wrote: > > > > But DIV is slow enough (or irrelevant?) that many RISC cpus don't > > include a hardware DIV instruction. So it is an actual slowdown that > > is, in many cases, not strictly needed. Right? > > Beside the fact that this discussion abou DIV is completely irrelevant
Not because it's "optimal", but because it's "good enough", right? > to the " Minor performance anomaly ", Because the main problem is disk-bound, not compute-bound?? > newer ARMS *DO* implement DIV > > https://developer.arm.com/documentation/100235/0004/the-cortex-m33-instruction-set/multiply-and-divide-instructions/sdiv-and-udiv There are dozens of ARM variants. I'm vaguely aware of Cortex-M33, but I know many things are missing (or extra) in many variants. Wikipedia says Cortex-M33 is 32-bit and from 2016. Cortex-M also has no MMU (needed for "full-fledged" operating systems). Most modern work seems focused on ARM64. _______________________________________________ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
