I wanted to add, I do believe it is an Intel i486 DX2 or DX2ODPR but don't they have CPUID support? The back looks similar to a lot of Intel processors of the era to me.

Even without the CPUID instruction there is still "value in DX at reset" which is equivalent, and HWINFO (author on this forum) has a few tricks to try and grab that value without a custom BIOS

jakethompson1 wrote on 2025-06-23, 02:36:

Even without the CPUID instruction there is still "value in DX at reset" which is equivalent, and HWINFO (author on this forum) has a few tricks to try and grab that value without a custom BIOS

Thank you! That is one utility I did not think to try, I had no idea that a DOS version existed. It seems to have solved the mystery:

1  º Main Processor:           IntelDX2, 66.7 MHz
2  º Math Co-Processor:        Internal
3
4  º CPU ID:                   00000433  (Method 1)
5  º CPU Stepping:             B1
6
7  º Number of CPU Cores:      1
8  º Number of Logical CPUs:   1

Try running SIV (preferably under NT) https://rh-software.com/ it could most likely give som information.

I highly doubt it being an OverDrive, as they usually had a flat plate on top with its designation. (don't know if that's removable, though)

H3nrik V! wrote on 2025-06-23, 08:31:

Try running SIV (preferably under NT) https://rh-software.com/ it could most likely give som information.

I highly doubt it being an OverDrive, as they usually had a flat plate on top with its designation. (don't know if that's removable, though)

SIV refers to it as a "Generic 486 DX2 66MHz".

If I look at processor features it says "486 Processor Type Intel 486", I think it's referring to architecture.

Per the data book (https://datasheets.chipdb.org/Intel/x86/486/a … ts/29043606.PDF , p. 56)

0433
0 = not an OverDrive (although, it looks like 1=OverDrive wasn't in place yet for the DX2 OverDrive, so it could be)
4 = 486-class processor
3 = DX2
3 = stepping number 3 (which HWINFO says is a B1)

The size of the die (gold cap on the underside) also tells you things about it,
SL support or not is also a factor.
cpu-world.com is really into this level of differences if you are still interested.

jakethompson1 wrote on 2025-06-23, 18:45:

Per the data book (https://datasheets.chipdb.org/Intel/x86/486/a … ts/29043606.PDF , p. 56) […]
Show full quote

Per the data book (https://datasheets.chipdb.org/Intel/x86/486/a … ts/29043606.PDF , p. 56)

0433
0 = not an OverDrive (although, it looks like 1=OverDrive wasn't in place yet for the DX2 OverDrive, so it could be)
4 = 486-class processor
3 = DX2
3 = stepping number 3 (which HWINFO says is a B1)

The size of the die (gold cap on the underside) also tells you things about it,
SL support or not is also a factor.
cpu-world.com is really into this level of differences if you are still interested.

Thank you for the detailed info.

The die looks like most other Intel CPUs of this era to me, including the numbering, of course that doesn't prove it is Intel 100%. It just looks like one. I know the size of the die also changed throughout the production of the DX2.

I may take a look at what I can find at cpu-world.com.

I notice that the manual for my motherboard lists these types of 486 processors.

[*] 80486 variations
[*] Cyrix
[*] AMD

I'm assuming it's an Intel 80486, and that's simply how the BIOS IDs it?

bryancandi wrote on 2025-06-23, 21:30:

I'm assuming it's an Intel 80486, and that's simply how the BIOS IDs it?

AFAIK it's impossible programmatically tell if it's Intel or AMD, looking at https://www.ardent-tool.com/CPU/486_Step.html it says UMOV can be used, but my testing shows this statement is incorrect, see Re: Encoding for umov eax, 1

Below is what my AMD DX2 66 @ 80 MHz reports on Menu->Hardware->CPUID->CPU-0 and I suspect your CPU will have the same UMOV 00000001.

[CPUID CPU-0] 🡐 SIV32N - System Information Viewer V5.82 Beta-05 RI4::ray

CPU-0 Generic 486 DX2-WT 80MHz STK 79.995MHz CPU 79.995MHz FSB 39.997MHz L2 39.997MHz

[_]CPUID EAX EBX ECX EDX [=]CR0 80010031 [=]Protected Mode [_]FPU Emulated [=]Cache Enable [=]Cache Write-Through [=]UMOV 00000001
00000000 00000001 656E6547 36383469 20636972 [=]Generic [_]"Generic i486" [=]Detect [_]Architecture 0 Type 486 Level 4 Revision FFD0
00000001 00000430 00010000 00000000 00000001 [_]09 APIC 00 [=]16 Count 1 [_]53 x2APIC [_]15 CMOV [_]08 CX8 [_]05 MSR [_]04 TSC [_]31 HVM

[_]MSR EDX EAX Thermal 00000000 [_]20 TSCIV [_]21 SLAT [_]58 Boost Flags 0000000000000002

[ OK ] [ Copy] [Windows] [Processors][Features] [Groups] [ NUMA] [ CPU-0] [Cache-0]

The Architecture 0 Type 486 Level 4 Revision FFD0 is the information NT returned in the SYSTEM_INFO block.

IIRC the UMOV is supported by all Intel 386 and 486 (thus also AMD because these are clones), and dropped in Pentium. Cyrix does not support it.

Now what does "supported" mean exactly - the CPU should not trip invalid opcode exception. What the instruction actually does is another thing, and I have a theory that it will work as MOV on CPUs that do not support energy saving modes. That's because originally UMOV was meant for ICE and hardware debugging on 386 but it can be used by the built-in system management ROM to transfer data to/from userland while SM is running. So CPUs SM-capable should at least ignore UMOV in userland code (peferably also trip an exception but clearly that's not the case) but CPU before SM will just execute it as-is.

At some point AMD lost a court case to Intel and was forced to remove ICE code from their 486 clones. This is noted in the product code. I'm not sure what will happen if UMOV is executed on AMD CPUs that came later with their own SM ROM, might be worth investigating.

red-ray wrote on 2025-06-24, 09:17:

AFAIK it's impossible programmatically tell if it's Intel or AMD, looking at https://www.ardent-tool.com/CPU/486_Step.html it sa […]
Show full quote

bryancandi wrote on 2025-06-23, 21:30:

I'm assuming it's an Intel 80486, and that's simply how the BIOS IDs it?

AFAIK it's impossible programmatically tell if it's Intel or AMD, looking at https://www.ardent-tool.com/CPU/486_Step.html it says UMOV can be used, but my testing shows this statement is incorrect, see Re: Encoding for umov eax, 1

Below is what my AMD DX2 66 @ 80 MHz reports on Menu->Hardware->CPUID->CPU-0 and I suspect your CPU will have the same UMOV 00000001.

[CPUID CPU-0] 🡐 SIV32N - System Information Viewer V5.82 Beta-05 RI4::ray

CPU-0 Generic 486 DX2-WT 80MHz STK 79.995MHz CPU 79.995MHz FSB 39.997MHz L2 39.997MHz

[_]CPUID EAX EBX ECX EDX [=]CR0 80010031 [=]Protected Mode [_]FPU Emulated [=]Cache Enable [=]Cache Write-Through [=]UMOV 00000001
00000000 00000001 656E6547 36383469 20636972 [=]Generic [_]"Generic i486" [=]Detect [_]Architecture 0 Type 486 Level 4 Revision FFD0
00000001 00000430 00010000 00000000 00000001 [_]09 APIC 00 [=]16 Count 1 [_]53 x2APIC [_]15 CMOV [_]08 CX8 [_]05 MSR [_]04 TSC [_]31 HVM

[_]MSR EDX EAX Thermal 00000000 [_]20 TSCIV [_]21 SLAT [_]58 Boost Flags 0000000000000002

[ OK ] [ Copy] [Windows] [Processors][Features] [Groups] [ NUMA] [ CPU-0] [Cache-0]

The Architecture 0 Type 486 Level 4 Revision FFD0 is the information NT returned in the SYSTEM_INFO block.

Here is my CPU-0 output. It does include UMOV 00000001. It's not detecting L1 cache info, I suspect maybe because my system is running Windows 95 rather than NT, detection may not be as reliable/support as many detection features. DOS utilities like HWINFO do detect it as 8KB W-T and the main output of SIV: L1 Unified Cache 4-way 16-byte 8KB

Thanks for the links, great resources.

[CPUID CPU-0] <- SIV32L - System Information Viewer V5.81 BRYANCAN::bryanc

CPU-0 Generic 486 DX2 STK 66.278MHz CPU 66.278MHz FSB 33.139MHz L2 0.000MHz

[_]CPUID EAX EBX ECX EDX [_]CR0 00000013 [=]Protected Mode [_]FPU Emulated [=]UMOV 00000001
00000000 00000001 656E6547 36383469 20636972 [=]Generic [_]"Generic i486" [=]Detect [=]Architecture 0 (0) Type 486 (486) Level 4 (0) Revision 0000 (0000)
00000001 00000400 00010000 00000000 00000001 [_]09 APIC 00 [=]16 Count 1 [_]53 x2APIC [_]15 CMOV [_]08 CX8 [_]05 MSR [_]04 TSC [_]31 HVM
00000001 Check CMOVxx + CMPXCHG8B + RDMSR + RDTSC [_]15 CMOV [_]08 CX8 [_]05 MSR I [_]04 TSC

[_]MSR EDX EAX Thermal 00000000 [_]20 TSCIV [_]21 SLAT [_]58 Boost Flags 0000000000000000

[ OK ] [ Copy] [Windows] [Processors][Features][CPU-0] [Cache-0]

It was actually CHKCPU that displayed cache information. Here is the output from that:

CPU Identification utility v1.27.1 (c) 1997-2022 Jan Steunebrink
------------------------------------------------------------------------------
CPU Vendor and Model: 'Classic' 486DX/DX2/DX4
Internal CPU speed : 67.0 MHz
CPU-ID Vendor string: - No CPUID -
Current CPU mode : Real
Internal (L1) cache : Enabled in Write-Through mode
Size of L1 cache : 8 KB

CPU Identification utility v1.27.1 (c) 1997-2022 Jan Steunebrink
------------------------------------------------------------------------------
Debug Mode!
Running DOS version : 7.10
Current CPU mode : Real
- No CPUID support detected -
AC Flag test result : 00h -> FF=386; 00=486+ (Alignment Check support test)
NexGen test result : FFh -> 00=NexGen CPU (or Intel P6)
Cyrix test result : 95h -> 00=Cyrix CPU (DIV 5/2 test)

'Classic' 486DX/DX2/DX4 67MHz

bryancandi wrote on 2025-06-24, 14:46:

It's not detecting L1 cache info, I suspect maybe because my system is running Windows 95 rather than NT
L1 Unified Cache 4-way 16-byte 8KB

It looks live SIV correctly reported/measured all the all but the WB/WT information, this is down to it being 9x rather than NT.

The L1 cache size is measured using much the code as Menu->Hardware->Cache Latency->CPU-0 uses.

red-ray wrote on 2025-06-24, 15:59:

bryancandi wrote on 2025-06-24, 14:46:

It's not detecting L1 cache info, I suspect maybe because my system is running Windows 95 rather than NT
L1 Unified Cache 4-way 16-byte 8KB

It looks live SIV correctly reported/measured all the all but the WB/WT information, this is down to it being 9x rather than NT.

The L1 cache size is measured using much the code as Menu->Hardware->Cache Latency->CPU-0 uses.

Your utility is very helpful, I added it to my collection.

After looking at a ton of photos, including on eBay, I am pretty confident that it is an Intel. The backside markings on most of the known Intel CPUs seem to have 10 characters, followed by 2 chars, followed by 2 sets of 2 characters. I haven't seen this specific pattern on any AMD or Cyrix CPUs in my searching.

1XXXXXXXXXX
2    XX
3 XX    XX