This app displays ALL CPUID info in HEX and ANSI that can be obtained using cpuid command.
It recursively retrieves CPUID data for processor.
For ditailed info on HEX values in registers refer to manuals (intel, amd etc.).
Anyone can use and modify it for more informative view with mention to my nickname.
Testing on different: very old/modern CPUs is appreciated.
___________
*Added procedure to substitute control symbols (CR/LF) that passed as strings to wsprintf with spaces.
*Updated again with right procedure of parsing ctrl symbols - CPUID instr returns only ASCII (ANSI?) strings and NO unicode.
*Replaced "modern" instructions: CMOVcc, MOVZX with their old equivalents for compatibility.
*Now more informative.
*both ANSI & UNICODE versions attached.
*updated messages when cpuid instruction is n/a.
*several additions according to cpuid specification updates.

If this would run in DOS, I could test my old P1 (no MMX, no Windows, only DOS, heh). However, it's a Win32 GUI proggie, and apparently 11 others have downloaded it. I can only assume their silence means it works okay.

New version and changes in 1st message.
______________________________________
10x rugxulo for cheering me up )
Yes, it is win32 exe, but it only uses 3 API funcs - ExitProcess, MessageBox and wsprintf. just replace then with dos ints and appropriate handlers (i mean processing strings - strcat etc.) and you'll get the equivalent cuz cpuid instruction works with whole 32bit (64 in Win64) register even in 16bit mode.

Actually, you use CMOVZ in there, so it probably won't work on lower than a PII (or PPro or whatever).

i just got it the morning after i wrote the code this night )
just wrote branchless code as optimization says... not so difficult at all to fix it.

rugxulo done. Updated 1st post.

Works ok, but what does "No"-button do I think OK and Cancel should be the buttons or only one button - OK - and then "X" in the corner to close it...

just a note.

Thanks! OK here means step ahead on cpuid detection and CANCEL step to extended features flag or out if we are already in there.
[added]
New version here.

asmfan, I tested it with Win98 with P166. It only show empty window with 2 buttons (yes/no). No window caption or any text inside window.
I replace 'win32wxp.inc' with 'win32a.inc' and then it works ok.

ATV
thanks for testing. Attached ANSI version for legasy OSes.
/just changed the header )) to *axp.inc - thanks to macroses TCHAR/

asmfan wrote:

> Also could somebody with p3 and less test it?

YES (see shots).

> Updated again with right procedure of parsing ctrl symbols - CPUID instr returns only ASCII (ANSI?) strings and NO unicode.

GOOD.

> Replaced "modern" instructions: CMOVcc, MOVZX with their old equivalents for compatibility.

GOOD. Fully 80386 compatible now ?

> both ANSI & UNICODE versions attached

Seems you just fixed bug I was about to report.

Any benefit of UNICODE here at all ?

Any way to provide just 1 version ?

Issues:

- If CPUID inavailable, report " CPUID instruction NOT available, CPU is 80386 or 80486-old ", rather than silent exit.

- CANCEL button and "extended" CPUID - odd design ?

rugxulo wrote:

> If this would run in DOS

It does (see shots)

Not sure - cannot find such an ancient CPU )

Since UNICODE support is native to modern Win OS - i support it. No benefit at all - only double memory consumption.
Oh, found one bft - native languages of nonenglish speaking people )
(just joking)

Done - but alas cannot really test (not using Olly) )) the same reason.

Nope, "cancel" it's just shortcut to ext feat, since just pushing "ok" can be boring ) (e.g. i have to push "ok" 13 decimal times before i get ext. features, and only 1 "cancel").
/another reason - don't want to mess with nicer GUI developement /

Post DOS version here if you don't mind. And let people with 386/486 try it )
btw intel manual says that only late 486 cpu support cpuid instruction.

Theres a cpuid program for DexOS, done 2-3 years a go

> Theres a cpuid program for DexOS

YES. The CPUID program is very good ... the top 28 lines of the shot a bit less

> > Fully 80386 compatible now ?

> Not sure - cannot find such an ancient CPU

Checked with BIEW - seems to be, except the CPUID instruction ...

> Post DOS version here if you don't mind.

I have none ... ran yours on HX

But here there is a DOS CPUID: http://rayer.ic.cz/programm/programe.htm

asmfan wrote:

> intel manual says that only late 486 cpu support cpuid instruction.

I wrote:

> CPUID instruction NOT available, CPU is 80386 or 80486-old

I mentioned somewhere on these forums that DOSBox 0.70 supported CPUID (emulates a 486 DX2) yet I found out (through Rayer's program) that it doesn't let the ID flag be set, so it doesn't correctly let you know that it supports it (even though it does). I should probably report that to them.

As far as testing on old computers, I've got a P166 (no MMX), which does support CPUID, and a 486 Sx/25, which doesn't. I'll go test and report what happens. (But DOSBox 0.70 + HXRT.ZIP + HXGUI.ZIP + "CPUID ANSI.RAR", at least, correctly whines and lets you exit.)

In other words, HXGUI doesn't work without VESA 2.0. (And my attempts to install UNIVBE 6.7 on the 486 didn't work, probably because of no FPU.)

However, the P166 (no MMX) worked okay (although I did have to use UNIVBE there, normally is only VESA 1.2 ATI Mach64).

dpmild32 af-cpuid.exe (had to rename it b/c spaces in filenames are weird)

BTW, Ctrl+Break didn't work for me in HXGUI, had to use APPS button (aka, right-click menu button?) on Win95 keyboard to bring up menu to allow screenshots (because Snarf didn't work). Oh, and HXGUIHLP.INI needed "save=0" to not mess up the palette upon exit (weird).

*added new features detection support for modern processors

Testing on Core 2 8000 & 9000 series & i7 for input eax=0Bh and eax=0Dh parameters appreciated.

several additions according to cpuid specification updates apllied to prev. post (iterative 0xD leaf).