Was the k6-3+ with no cache cpu id added?

I wonder if CPU-Z can tell if CPU is capable for executing CPUID instruction from its main window?

The newest version 2.10 can no longer be installed on 32-bit Windows XP (installer icon is different as well, along with a 59% larger size compared to 2.09); an error message "[path]\cpu-z_2.10-en.exe is not a valid Win32 application" pops out. It can be installed onto 32-bit Windows 7 and 10, however.

dormcat wrote on 2024-07-14, 07:27:

The newest version 2.10 can no longer be installed on 32-bit Windows XP (installer icon is different as well, along with a 59% larger size compared to 2.09); an error message "[path]\cpu-z_2.10-en.exe is not a valid Win32 application" pops out. It can be installed onto 32-bit Windows 7 and 10, however.

Crap.

I think great efforts should be taken to keep cpu-z working on all platforms physically capable of running such a program, and even fringe systems. Basically, 386+

Wasn't here for over two years...time flies... When I saw the last two posts from @dormcat and @Sphere478 I had to login again to write this post!

Back in February (20th) this year I noticed that the latest version of CPUID HWMonitor, version 1.53 (I always use and recommend the zipped versions), doesn't start anymore on WinXP SP3 and just give the weird error message "Die Anwendung wurde mit STRG+C unterbrochen" (German for "The application was interrupted with STRG+C"). So I decided to contact @CuPid directly, but he had no idea/solution.

When the latest CPU-Z version (2.10) was released on July 12th, I got the same error message. I contacted @CuPid again and he still had no idea. After some exchange of information this time he finally found the reason! It was an API call which works on WinXP x64 and later, but not on WinXP x86 - I guess this is a Kernel 5.1 <-> 5.2 thing.

Long story short:
HWMonitor recently (July 17th) has been released in version 1.54 which works again on WinXP, surely @CuPid will also release a working CPU-Z version soon.

kind regards
soggi

Fingers crossed!

Good job sorting it out!

CPU-Z 2.11 has been released by @CuPid and it works flawlessly on WinXP SP3 -> https://www.cpuid.com/softwares/cpu-z.html.

Hopefully the new microcode revision report in "Mainboard" tab / BIOS also will find its way into the Vintage Edition of CPU-Z - I'll contact @CuPid, i guess.

kind regards
soggi

soggi wrote on 2024-09-21, 14:58:

CPU-Z 2.11 has been released by @CuPid and it works flawlessly on WinXP SP3 -> https://www.cpuid.com/softwares/cpu-z.html.

Unfortunately not in my cases: I've tried two WinXP builds but still receive "not a valid Win32 application" warnings.

OK...above you're talking about an installer, maybe it's this? I don't give a damn on installers and always use the zipped version and the cpuz_x32.exe from the .zip works flawlessly on WinXP SP3 - as said above. The system is POSReady "hacked" and has all updates until April 2019 installed, but I don't think this makes any difference.

On the screenshot below you can see it works on our IBM/Lenovo ThinkPad T60.

kind regards
soggi

soggi wrote on 2024-09-21, 22:48:

OK...above you're talking about an installer, maybe it's this? I don't give a damn on installers and always use the zipped version and the cpuz_x32.exe from the .zip works flawlessly on WinXP SP3 - as said above.

Ah, yes, so it was indeed the new installer of 2.10 and 2.11 causing problems. Problem solved; thanks a lot!

You're welcome!

dormcat wrote on 2024-09-21, 23:45:

...so it was indeed the new installer of 2.10 and 2.11 causing problems

That's not right...version 2.10 actually didn't run on WinXP as described in my post above, independently of the problem with the installer! CPU-Z 2.10 used a function (RegGetValue) which is available on WinXP x64 and up but not on WinXP 32 bit, this has been resolved.

@CuPid will keep up the support of older Windows versions as long as possible.

But such installers are programmed by others, f.e. WinSCP uses InnoSetup which doesn't support WinXP anymore, but the program itself still supports it and you also have to use the zipped (portable) version of it - BTW there also was an issue with WinXP, but Martin Přikryl (the main developer) fixed it after my request.

In conclusion it's always a good idea to contact the developer(s) via e-mail (or some other channel), if you have problems with their program and of course it's also a good idea to donate some bucks to the developers of such free / open source tools (I did so to both).

kind regards
soggi

I wonder if dev is still watching this thread? CPU-Z vintage edition doesn't work well in NT 3.51:
https://x.com/kurist2010/status/1873894295775961478

The main release of CPU-Z seems to have had a regression at some point: (screenshots from AMD Kaveri CPU on A88X chipset)

I have an issue with CPU-Z Vintage that I am unable to the validate the results and it returns this error message:

System is NexGen Nx586-P90 on Win95 OSR2, CPU-Z Vintage 1.04.

elianda wrote on 2025-02-19, 13:19:

I have an issue with CPU-Z Vintage that I am unable to the validate the results and it returns this error message: […]
Show full quote

I have an issue with CPU-Z Vintage that I am unable to the validate the results and it returns this error message:

System is NexGen Nx586-P90 on Win95 OSR2, CPU-Z Vintage 1.04.

I am getting the same issue with multiple processors, including a genuine Intel DX-2/66 that's old enough to not be properly displayed.

My suspicion is that this happens when there is a CPU that does not have a CPU-ID instruction.

elianda wrote on 2025-04-25, 02:00:

My suspicion is that this happens when there is a CPU that does not have a CPU-ID instruction.

Actually, it might be something deeper than that, as I was able to generate and get accepted validations from 1.03. It did reject the 486, but that's as expected.
https://valid.x86.fr/x3r479
https://valid.x86.fr/weyi9b

AFAIK, there isn't much information you can read from 486 and older CPUs. The parameters can only be detected from the system configuration. That means that you don't really get information about what CPU you have, but as what CPU you have your motherboard configured.

You could take Am486-100 CPU and run it with 2x33MHz speed and set the L1 cache to Write-Through and the software would be incapable to detect that this is not 486DX2-66 and is really a Am486-100. (In case of Am5x86, there's double L1 cache size, so software could start assuming some parts based on that)

I might be wrong. This is based on behavior of multiple software that i've used. Mostly SpeedSys - this is the most straight forward - i change the jumpers on motherboard and it shows a different CPU model.

GigAHerZ wrote on 2025-04-28, 07:26:

AFAIK, there isn't much information you can read from 486 and older CPUs. The parameters can only be detected from the system configuration. That means that you don't really get information about what CPU you have, but as what CPU you have your motherboard configured.

You could take Am486-100 CPU and run it with 2x33MHz speed and set the L1 cache to Write-Through and the software would be incapable to detect that this is not 486DX2-66 and is really a Am486-100. (In case of Am5x86, there's double L1 cache size, so software could start assuming some parts based on that)

I might be wrong. This is based on behavior of multiple software that i've used. Mostly SpeedSys - this is the most straight forward - i change the jumpers on motherboard and it shows a different CPU model.

You are partly right. Without CPUID it is much harder to detect exactly what CPU you have in protected mode. In real mode DOS however, there are tests you can perform to detect most early CPUs. A 486 DX2 and a DX4 at 2c clock speed will appear the same, but even with an 8086 or earlier, there are tests you can perform to tell the difference between various (not all) makes and types.

The stepping is 0xB1 and even if it wasn't an Overdrive it should be mxB1/xB1. The reported speed is also wrong both HWiNFO and SIV report around 199.6 MHz.