I'm going to (try to) buy the DX40 I had my eye on as a gateway into 486. I can always upgrade motherboard and CPU to a DX2-66 if I'd want to later, but at least this will give me an awesome case + psu + 16 MB RAM (type might not be compatible with some later 486 mobo's as some only have the wide type while this PC comes with the narrow ones...), a Cirrus VLB card and a SB16. And another card which I keep secret until I have the PC. So at the very least a good basis, possibly my final 486 destination.

Sounds like a good plan. Buying a complete working system is always a good starting point, since it gives a great base to make upgrades or changes to.

Well, there goes my money! Unless I'm royally scammed, you can now call me a 486 club member. And a Varta club member, so one of the first things will be removing the dead can and looking into a CR2032 solution.

Am I right that the current motherboard, a 1992 OPTi 486DX aimed at a DX40, will likely not be compatible with a DX2-66?

Nicolas 2000 wrote on 2025-12-30, 20:38:

Am I right that the current motherboard, a 1992 OPTi 486DX aimed at a DX40, will likely not be compatible with a DX2-66?

To my understanding, a DX2-66 runs at 33 MHz externally.
So it works same as a 486DX-33 from a mainboard point of view.

Special support for a 486DX2-66 is not required, thus.
A 486 motherboard aimed at a DX40 can be configured for a DX33, I think.

5v vs 3.3v is *maybe* something to look for, but the original, normal i486DX2-66 was used in ordinary 5v sockets.

I'm just a layman, though. Some other user(s) maybe know more about it.

Nicolas 2000 wrote on 2025-12-30, 20:38:

Am I right that the current motherboard, a 1992 OPTi 486DX aimed at a DX40, will likely not be compatible with a DX2-66?

The DX2 66 was released in March 1992, and is pin compatible with earlier 486 chips.

It's possible your board's BIOS wouldn't support L1 cache write-through, but you can enable that with a utility from DOS. And even then it's a 1-2% performance increase.

Some extremely early 486 motherboards didn't have clock multiplier jumpers, or the ability to set the bus beyond 25Mhz. But if it's running a 40Mhz chips you should more than likely be able to run a DX2 66. It's possible the BIOS doesn't display the correct chip name, or there to be some instability. But people have created patched BIOSes that enabled extra chips on various boards, so if the stock BIOS has issues. You can more than likely fix it.

Nicolas 2000 wrote on 2025-12-30, 20:37:

Well, there goes my money! Unless I'm royally scammed, you can now call me a 486 club member. And a Varta club member, so one of the first things will be removing the dead can and looking into a CR2032 solution.

Good news: I am, in fact, NOT member of the Varta club. This motherboard has an Emmerich, likely lithium ER14250. So no leaks and one for one replaceable with a new 14250 from wherever.

Good to hear that there is a good possibility of being able to drop in a common DX2-66 without replacing the motherboard!

The current cpu has a heatsink and a fan on it and the (dumb) case display says 50MHz while the ad and case sticker say 40, which is why I am unsure about the actual cpu in this processor. A 40 with wrong display settings, a 50 with outdated case sticker, a 40 overclocked to 50...? Wait and see when the PC arrives and is up and running.

Just look at the CPU from below. It is possible the model name is written down there, so you can look up its real speed.

The standard DX2 66 only requires a 33MHz mobo setup. Internally it doubles the clock. Just be careful about 5V vs 3V. All my 66 MHz are Intel and use 5V. I believe AMD made 3V 66MHz CPUs as well as 5V.
If your board has a 40MHz CPU then I bet it supports 25/33/40 MHz buses. I don't like 40 and its multiples (e.g. DX2 80) because it doesn't play nice with many VLB cards. 33MHz is the max speed that I was able to reliably run my VLB cards (VGA & I/O). The DX2 66 uses a 33MHz base clock. DX4 100 uses 33MHz. DX2 50MHz uses 25MHz...etc

I think same! 😀

But if the DX40 CPU has fan/heatsink, maybe it runs at 50 MHz?
I mean, a 10 MHz overclock seems feasible on such a CPU. It merely gets hotter.

Many intel i386DX-33 owners ran their 386 CPU at 40 MHz before AMD sold the am386DX-40.

So I'd recommend to run a benchmark utility on DOS to make sure.
There are small programs such as CPULEVEL or WCPU that tell CPU type and speed.

Alternatively, there's NSSI, which has a fine benchmark feature.

The classic "SI" (System Information) is part of Central Point PC-Tools 7/8 and Norton Utilities.
It also has some benchmark feature, albeit a less accurate one.

PS: About the bus speed.. A faster bus speed helps to improve memory throughput.
It's great for a Novell file servers in a network, for example.

For computational stuff, such as CAD or games, a higher CPU clock is increasing performance.

That's why the 40 MHz 386 processor was so good back then.
It reduced bottlenecks in a pure ISA motherboard design.

On a 486 a slower bus speed might be not so noticeable, though.
Because the 486 CPU has internal caches (write-back or write-through),
so fast access to RAM isn't always needed when making calculations.
An 486 VLB mainboard has faster i/o on bus than a plain ISA mainboard, also.

So 33 MHz might be the more stable, less electrical noisy choice here.
A smooth 33 MHz operation can be better, thus.
Especially, if that implies that it allows using an VLB HDD controller instead of an ISA model.

But sure, a higher bus speed is nice to have if the hardware supports it.
That's why the DX50 was the king back then.

It had a high 50 MHz bus speed/50 MHz CPU speed,
while same time was just a little bit slower than the 486DX2-66 in CPU intensive tasks.
In practice, this made the DX-50 more balanced in throughput. It was thinking fast and had a high bandwidth.

I'll have to wait until the PC arrives, which will be a while given the time of the year.