George Gwilt once told me that there is a extension that lets one
use the Floating Point Unit of Q40 and Q60 (also QXL with replacement
processor with FPU) from BASIC. The software is written by Simon N
Goodwin and he was so kind to pass it on to me for release.
There may be some guys who already know it, but to me it was new and
I also couldn't find it at the usual places for QL software.
It is free and open source. Please get it from
http://q40.de (Programs section)
Speed increase can be huge, depending on the math functions used.
I have used Simon's floating point functions for some time now on my Q40. Having recently tried them on my Q60 I found two mistakes in FPSP_60, the floating point software package containing code for the floating point instructions not in the 68060 hardware. The first mistake will treat SIN as COS if COS has previously been requested. The second causes the routine I_TO_Q, which translates IEEE floating point to Qdos format, to crash the Q60. I have thus produced a new version of FPSP_60.
Version 1.4 of the FPSP_60 will very soon be on the SQLUG web site inside the updated FPSV package. Also included is a new version (1.20) of FPSAVE. This will not accept versions of FPSP_60 earlier than 1.4. Otherwise version 1.20 of FPSAVE
is the same as version 1.19 which can continue to be used by 68040s (eg Q40).
George
