Floating point performance

I’m getting really terrible floating point performance used binaries compiled with GCC.

I’ve had a look through the fpa.c source file and it looks like it should run much quicker than it does. The x86 chips have pretty fast FP after all…

I’m using the SCL, so presumably it’s using FPA. But is it emulating it within RISC OS (slow…) or passing the instructions through to RPCEmu to run at basically native speed (albeit interpreted)?

Thanks,
Daryl.

I believe it’ll be emulating it within RISC OS.

Arculator recently gained support for FPA emulation, so maybe the RPCEmu developers will consider adding that code to RPCEmu.

Thanks Jeffrey, that explains that then. I’m juggling lots of balls otherwise I would try and do it myself, or for extra points try and add VFP!! Yep, not happening right now..