As far as I know they are completely identical.. only have a super33 though so I cant test.

I have only yesterday watched a video of someone on youtube actually and they mentioned Cyrix 487 wont work with UMC chips. I forget who it was though, possibly High Treason..

Might help if there was an official datasheet for the U5S, rather than some scattered text in forum. I might have better luck getting the SXL2 to work with the Cx487S if I can mod the motherboard to support the SXL2.

Hopefully someone with both the super and non-super can benchmark these chips. Perhaps "Super" was just a lame marketing scheme.

A slightly late reply to this topic, but I can confirm the chips appear identical internally (stepping may very but impossible to tell).

I have a non-super chip, with "U5SX 486-33 markings" and "Windows Compatible" logo, dated week 46, 1994.
I also have a model marked U5S-SUPER33, dated week 39, 1994. This one also has the "Not for U.S. sale or import" markings.

CHKCPU identifies both chips identically as "UMC U5S (486SX) 25/33/40". Both benchmark the same in all tests and both overclock successfully to 40- and 50-MHz without breaking a sweat on my EISA/VLB system. Performance is very impressive, matching or outperforming its Intel counterparts and even the SGS-Thompson DX2/66.

I am awaiting delivery of a SUPER40 variant and an AMD DX/40 to make further comparisons.

Edit: it arrived. It's week 34 of '94, so the earliest of the three. It behaves exactly the same way and is seen by various tools as the same chip. If they only manufactured one 'SX' die, and marked them based on how they were binned, I have no idea what the criteria were because there's no measurable performance difference here. They all overclock like absolute champs (shame my motherboard doesn't do 60MHz). One thing I noticed is that all 3 chips run very cool. Must be the 0.6 micron process.

U5S never has FPU, the U5D (unobtainium) was the version with FPU.

Just noticed 40MHz UMC was available in Poland in December 1994, cost ~$84, half the price of AMD DX/40
Re: Identifying an old intel CPU
Month prior also listed 33MHz version at ~same price.

At that price UMC was a fantastic deal for gamers.

rasz_pl wrote on 2022-12-01, 10:38:

Just noticed 40MHz UMC was available in Poland in December 1994, cost ~$84, half the price of AMD DX/40 Re: Identifying an old i […]
Show full quote

Just noticed 40MHz UMC was available in Poland in December 1994, cost ~$84, half the price of AMD DX/40
Re: Identifying an old intel CPU
Month prior also listed 33MHz version at ~same price.

At that price UMC was a fantastic deal for gamers.

Yeah considering the 40MHz UMC performs roughly on par with a DX-50, it would have been adequate for even Doom II at the time. I bought a DX-33 in late 1993, and if I knew of the processor was around in 1994 I would have felt pretty bad..

That said, 14 year old me at the time was convinced that Cyrix CPUs burned themselves to death and AMD CPUs performed slower, so no idea what bullshit about UMC would have been fed to gullible young me at the time.

Has anyone tried to use a UMC U5S with Cyrix 487S FPU adaptor? This is basically just a PGA168 male to PGA168 female socket adaptor with a PCB in it that contains a Cyrix 487S FPU. It was normally used with the very early Cyrix 486 chips, which had 2 KB cache. Since pretty much all other PGA168 486 chips contained an internal FPU, I can see the UMC U5S or maybe 486SX being a good candidate for this adaptor.