"Edge cases" are compiles were the floting point type used for the core
may be different from the one used by solvers, e.g. double/float.
This does not happen in MAME, here currently this is always
double/double. Since floats are better suited for GPUs whose double
performance is limited it was time to review the code.
This commit fixes warnings about type conversions and precision loss.
In addition there is better support for INT128 and FLOAT128.
The MAT solver has seen some minor performance increases.
---------------------------------------
ibm5150: 3-D FloorPlan Designer V2.0, Ad Lib Instrument Maker V1.51, Ad Lib Juke Box V1.5, Ad Lib Juke Box V1.6, Ad Lib Juke Box V1.64, Ad Lib Juke Box Registration Bonus Disk V1.51, Ad Lib Juke Box Registration Bonus Disk V1.62, Ad Lib Visual Composer V1.51, SYSGRATION Mouse Driver 7.03 (5.25"), SYSGRATION Mouse Driver 8.00 (5.25"), SYSGRATION Mouse Driver 8.00 (3.5") [Justin Kerk]
New not working software list additions
---------------------------------------
ibm5150: Mouse Software Test/Demo [Justin Kerk]
(nw) moved some compatible mouse drivers to ibm5150.xml from ibm5170.xml, mark sicmouse as a bad dump, make 5.25" version the parent for indycrus to match the rest of the ibm5150 softlist
* Created softlist for Timex cassettes (timex_cass)
Timex computer emulation is currently using the Spectrum softlist for cassettes, but many of those entries are actually incompatible unless the "Spectrum Emulator" cartridge is being used.
So I've created their own compatible (depending on the model) softlist, with every actual Timex dump I could find in World of Spectrum and Spectrum Computing.
These should work with the Timex "equivalents" to the Spectrum 48K and their clones:
-TS2068
-UK2086
-TC2068 (not currently emulated, or is it exactly the same as the UK2086?)
-Komputer 2086 (not currently emulated)
And some of them should work with the Timex equivalents of the Spectrum 16K:
-TC2048
-TC2048 (NTSC) (not currently emulated)
-TS2048 (prototype) (not currently preserved)
I've noticed that many of the TZX dumps were fake ones made from TAP files (so the real pauses between blocks were lost), so I just added the TAPs for those while any real TZXs are created.
Also, there were a few bad dumps of which I could only fix one at the moment, but they seem to load in Fuse anyway so I've left them in the softlist, labeled as such.
Also included some initial notes on the cassette dumps at the beginning of it, based on the spectrum_cass softlist but more to the point.
* Hooked up timex_cass to the TS2064 and TC2048
Now the TS2064 and TC2048 systems use both the spectrum_cass and timex_cass softlists.
Also, small fixes.
* Fixed interface types
* new WORKING clones
-----
Guitar Super Star (red 'Gibson Flying V' style) [Sean Riddle, David Haywood, Peter Wilhelmsen, anonymous]
* typo fixes (nw)
* updating notes (nw)
* wrlshunt missing tilemap notes (nw)
* fix Operation Wolf palette format based on schematics (+ verified on PCB) [Bryan McPhail]
Added a define NL_USE_ACADEMIC_SOLVERS and disabled it in the build.
This will not compile in solvers which are either illustrative or only
perform for large sparse matrices like GMRES.
Out of whatsnew: Corrected as many ROM names as could be verified
New working clone
------------------
Touchmaster 5000 (v7.11 U.K.) [Muddymusic, Porchy, The Dumping Union]
add wofr1bl, very similar to wofpic but with usual z80/ym/oki sound h/w instead of pic
tweak wofpic bit more, more priority issues fixed
fcrash, cps1bl_pic source tidy-up
