Yeah, at the present moment the pitch slides are emulated, but emulated incorrectly. I don't know of a fast/good sounding way to slide between two notes.

That aside, I've been meaning to post a new update but again, my attention has been detained elsewhere. Don't worry, I haven't forgotten about the emulator!

As for compiling it, I use Visual Studio .NET for most of my development and haven't used VC6. Exactly what errors are you getting? Send me some compiler output and I'll see what I can do to help you.

Yeah, the .dsp is way out of step, files are missing from it etc. I attached mine, it may work better (I know for sure it compiles, just change the paths to SDL and zlib in the proj. properties.)

Actually the vc6 workspace will be removed from the cvs.
as vc6 creates broken code in the cpu core.

I failed to mention, get the latest SP for VC (SP5 IIRC) and the CPU pack (for the MMX optimisations in the MT-32 code.) And some hacks will be necessary in the scaling (2x, 2xsai etc.) cdoe because VC6's parser has serious problems with pointers to "templatized" functions.

I have problems compiling emulator with VS.NET 2003 (7.1). Compiler gives:

1\VS\dbcvs\mt32cvs\CSynthMT32.cpp(2531) : error C2666: 'pow' : 7 overloads have similar conversions
2        C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\math.h(620): could be 'long double pow(long double,int)'
3        C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\math.h(618): or 'long double pow(long double,long double)'
4        C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\math.h(572): or 'float pow(float,int)'
5        C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\math.h(570): or 'float pow(float,float)'
6        C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\math.h(534): or 'double pow(int,int)'
7        C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\math.h(532): or 'double pow(double,int)'
8        C:\Program Files\Microsoft Visual Studio .NET 2003\Vc7\include\math.h(195): or 'double pow(double,double)'
9        while trying to match the argument list '(int, double)'

About 10-15 occurrences.
It also complains about cases with (double,float) and (float, double) combinations.
This doesn't happen with VS.NET 2002. It seems that declarations in "math.h" are changed.

vladr:

Yep, this was the main thing VC was choking on was the templates (also, a couple defineguid statements in one of the CD-ROM header files. (Commented out, and the complaints dissapeared.... )

I wound up hacking together some pretty ugly macros that "look like" templates, It appears now I have a problem linking curses, and it appears the MT-32 files weren't included in the dsp, so I will have to add those.

Also, IIRC I have the latest SP for Visual Studio, but can't remember if I installed the processor pack or not... I know I can at least build programs with MMX ASM, but browsing the MT-32 source file, I don't know about SSE & 3DNow... If all else fails, I suppose I could just #define those out, and use the non-sse filter functions.

Does anyone know exactly what the broken code is in CPU, when building under VC6?

For VC6, the latest service pack release contains support for 3DNow and SSE opcodes.

I got it to build, however, I get an access violation at this line:

1		tpoly->pStatus[x].pulsewidth = pcache[x].pulsewidth + pwveltable[pcache[x].pwsens][vel];

Here's the contents of pcache[x] (x = 3 in this case, it seems that there's not enough room allocated for the variable:

1-	&pcache[x]	0x03e6aa78
2	rawPCM	0x00000000
3+	convPCM	{...}
4	playPartial	0x00 ''
5	usePartial	0x00 ''
6	PCMPartial	0x00 ''
7	waveform	0xcd 'Í'
8	pulsewidth	0xcdcdcdcd
9	pwsens	0xcdcdcdcd
10	pitchshift	0xcdcdcdcd
11	fineshift	0xcdcdcdcd
12	sustain	0xcd 'Í'
13	lfodepth	0xcdcdcdcd
14	lforate	0xcdcdcdcd
15	lfoperiod	0xcdcdcdcd
16	modsense	0xcdcdcdcd
17	keydir	0xcdcdcdcd
18	pitchkeyfollow	0xcdcdcdcd
19	pitchkeydir	0xcdcdcdcd
20	filtkeyfollow	0xcdcdcdcd
21	tvfbias	0xcdcdcdcd
22	tvfblevel	0xcdcdcdcd
23	tvfdir	0xcdcdcdcd
24+	ampbias	0x03e6aad8
25+	ampblevel	0x03e6aae0
26+	ampdir	0x03e6aae8
27	ampdepth	0xcdcdcdcd
28	ampenvdir	0xcdcdcdcd
29	amplevel	0xcdcdcdcd
30	tvfdepth	0xcdcdcdcd
31	prevsample	0xcdcdcdcd
32	useBender	0x00 ''
33+	pitchEnv	{...}
34+	ampEnv	{...}
35+	filtEnv	{...}
36	ampsustain	0xcdcdcdcd
37	pitchsustain	0xcdcdcdcd
38	filtsustain	0xcdcdcdcd
39	partCount	0xcdcdcdcd
40+	padding	0x03e6ab44 "ÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍÍ"

Nope, you still need the CPU pack even with SP5. There is a "CPU pack for SP5". Anyhow, for the GUID you need the latest platform SDK (and change VC6's "directories" to point ot it - I have whatever was "latest" one year ago, when I needed it for XP themes in LaunchPad.) This may also solve the math.h "out of the box". Other than that, disable debug and screenshots etc. if you don't want to link with libcurses and libpng. Unless you want debug.

Oh, and I also built my own SDL with DX6 support enabled for notifications, much cleaner sound output under load, it's worth the extra effort. 😀

For the access violation: see my "TODO" post in this forum.

V.