When cputype=386 some programs are reporting DOSBox-X as having a "NexGen 5x86" about dosbox-x

[joncampbell123]

A-ha:

How to detect an Intel 386 vs a NexGen 5x86

So if you set ZF=1 from XOR DX,DX, then execute a divide instruction that will have a nonzero remainder like 0x5555 / 2, the CPU is an Intel CPU if ZF changes, else the NexGen 5x86 does not change ZF.

[joncampbell123]

OK. Most of the programs now detect cputype=386 as a 80386. The only stubborn program that still detects the CPU as a NexGen 5x86 is AGSI.

[joncampbell123]

Fixed. It took some additional work on the DIV instruction emulation. Now most programs detect cputype=386 as an Intel 386DX, except for one that sees the CPU as a Chips & Tech 38600 (Super386), and Windows 95 OSR2 which sees the CPU as a "CyrixInstead" (argh).

I have no idea why Windows 95 OSR2 would misdetect cputype=386 as a Cyrix, but I do have a good idea why the other program misdetects it as a C&T 38600: http://computer-programming-forum.com/46-asm/c4c6d67250609049.htm

The common test seems to be to execute POPAD and a memory EA reference repeatedly to purposely trigger the POPAD bug, then assume that if the bug doesn't trigger the CPU is not an Intel 386. Well, DOSBox doesn't emulate the POPAD bug, so... it thinks we're a C&T 38600 instead of an Intel 386DX.