I would put the slowest celeron on. Most games from the 80s may have a speed issue the faster you go.

Here is a link to the CPU compatibility list on MSIs site.

https://www.msi.com/product/motherboard/suppo … tml#support-cpu

It will support all socket 370 CPUs from Coppermine Celeron and up. The Mendocino Celeron 466 isnt officially supported but you can always try it anyway.

Honestly, I'm not sure any of those can be slowed down to 386/486/early Pentium levels. The hit from disabling the caches is too big (I've seen people claim 286 levels of speed) and you don't have the necessary granularity. You can always try software slowdown utilities though.

IMHO you should consider a VIA CPU (their multiplier can be lowered in software) or something like a low-multi Coppermine rated for 133MHz FSB (you can slow it down further by downclocking the FSB to 100MHz and 66MHz). Celerons usually have higher multipliers than their Pentium III counterparts due to being rated for lower FSB clocks.

That is true. It's no so easy, to emulate a 386/486 with a "modern" retro cpu (core coppermine or tualatin). I have also made some test with these three CPU's:
Pentium (P54CS), AMD K6-2 (Chomper Extended) and a PIII (coppermine).

My experience was, with a Pentium you can get 486DX-25 and 386DX-40 speed. But with the Pentium you cannot play late 90's games anymore.
The Coppermine was little strange. I've tested the three FSB clocks (66, 100 and 133 MHz), but only the 66MHz FSB also with disabled caches. I've tested only the 2 first benchmarks from Phils package. And the result was: 3DBench like a 486DX-50/66 and PCPBench like a 286 PC.
The best option for me is the AMD K6-2. You can reach low 486 and 386 with the lowest clock settings, and you are able to clock stepwise higher. With a K6-2 the multiplier is free from 2,5 to 6.
You can also take the K6-2+ or K6-3(+) for these 'time-machines'.

You can try and test the celerons... install Phil's benchmark package, and test the different speeds with 'cache on' and 'all caches off'. According to what a guy here told me, with the 'modern' cpu's and the integrated L2 cache you cannot deactivate separately.

I'm not sure the 694T will like the higher voltage the early Celerons require (would be a shame if the board got damaged because of this).
I remember I did once try a Mendocino Celeron in a Coppermine board once years ago ( don't remember which board it was, except it was probably an OEM board) and it didn't work the way I had hoped for.

Tetrium wrote:

I'm not sure the 694T will like the higher voltage the early Celerons require (would be a shame if the board got damaged because of this).
I remember I did once try a Mendocino Celeron in a Coppermine board once years ago and it didn't work the way I had hoped for.

All boards from the larger manufacturers should just fail to post if the VID isnt recognised. There is of course no guarantees but I have never managed to break a board that way and I have tried some crazy stuff. 😁

Skyscraper wrote:

Tetrium wrote:

I'm not sure the 694T will like the higher voltage the early Celerons require (would be a shame if the board got damaged because of this).
I remember I did once try a Mendocino Celeron in a Coppermine board once years ago and it didn't work the way I had hoped for.

All boards from the larger manufacturers should just fail to post if the VID isnt recognised. There is of corse no gurantees but I have never managed to break a board that way and I have tried some crazy stuff. 😁

Ahh, I just noticed your reply after I edited my previous post to include the board I was testing some CPUs. It was in my early days, so my knowledge wasn't so great and my collection was very small, so I had few options and the board behaved strangely I think, I ended up putting that board away.

Tetrium wrote:

Skyscraper wrote:

Tetrium wrote:

I'm not sure the 694T will like the higher voltage the early Celerons require (would be a shame if the board got damaged because of this).
I remember I did once try a Mendocino Celeron in a Coppermine board once years ago and it didn't work the way I had hoped for.

All boards from the larger manufacturers should just fail to post if the VID isnt recognised. There is of corse no gurantees but I have never managed to break a board that way and I have tried some crazy stuff. 😁

Ahh, I just noticed your reply after I edited my previous post to include the board I was testing some CPUs. It was in my early days, so my knowledge wasn't so great and my collection was very small, so I had few options and the board behaved strangely I think, I ended up putting that board away.

With motherboards one cares about it's always better to be safe than sorry I guess. Trusting MSI is perhaps not always good either when I come to think about it, they fell for the pressure and allowed overclocking Prescotts on their i875P boards and we know how that ended up.

The 694T Pro is one of the few Tualatin compatible motherboards with an ISA slot. Using it with a slow celeron CPU would be a waste. My advice would be to build a win98 box with this board and get a slower machine for early dos games (486, pentium 1, etc).

kanecvr wrote:

The 694T Pro is one of the few Tualatin compatible motherboards with an ISA slot. Using it with a slow celeron CPU would be a waste. My advice would be to build a win98 box with this board and get a slower machine for early dos games (486, pentium 1, etc).

Seconded 😀

With for instance a ss7 one would have more flexibility (a time-machine) for DOS and the Tualatin board could be the used to build a nice (and faster) Tualatin rig around it. With the Mendocino Celeron you'd kinda get a hybrid which won't do either task as well as 2 dedicated systems would.