I think that the 486SLC is a 486sx compatible, as opposed to the 486DLC which is more like an enhanced 386DX.

^Agrees. Seems your board can be upgraded with an FPU. It's kinda too bad your CPU doesn't come socketed though. I'd say, give it a wirl 😉

I think that the 486SLC is a 486sx compatible, as opposed to the 486DLC which is more like an enhanced 386DX.

AFAIK, the SLC and DLC both have exactly the same instruction sets and pipelining, the main difference between them being the bus width. They can be considered just as much a 'real' 486 as... well... a real 486.

You may be thinking about the difference between Cyrix and IBM versions of the SLC/DLC, which, despite their similar names, are quite different from each other... the IBM ones I believe had a larger cache, and were based around an Intel core.

My CPU looks like this - http://upload.wikimedia.org/wikipedia/commons … ts_TX486SLC.jpg - only it's PJF instead of PAF

From what I've read so far, Cyrix and Texas Instruments shared some manufacturing facilities for a while, and wikipedia writes this about Cyrix's 486SLC/DLC:

"Its early CPU products included the 486SLC and 486DLC, released in 1992, which, despite their names, were pin-compatible with the 386SX and DX, respectively. While they added an on-chip L1 cache and the 486 instruction set, performance-wise they were somewhere between the 386 and the 486."

So I guess mine should perform somewhere between a 386SX and a 486SX - I'll only compare it to a 386DX-40 (the closest one I have), and I'll have to find a way to "quantify" the lack of L2 cache on this board...

[edit] Oh, here's more: http://en.wikipedia.org/wiki/Cyrix_Cx486SLC

I'll have to find a way to "quantify" the lack of L2 cache on this board...

That actually isn't nearly as big a deal as you may think... though it lacks L2 cache, it does have a much faster L1 cache which a 386 doesn't have. IIRC, a 486SLC without L2 cache would come pretty close in performance to a similarly-clocked 386DX with L2 cache.

My second PC was a 486 SLC 33 and it was absolutely abysmal. It cost almost as much as the 486 DX 33 but the man in the shop assured me that the SLC was almost as powerful. How he lied...

My friends had a 486 DX 50 and a DX2 66 respectively. With my SLC I could only play Doom in low detail and by reducing the screen size a few times. In other words, it performed more like a 386 than a 486. Doom 2 was out of the question, being a slideshow of about 5 fps in high detail and barely playable in low detail (about 10-12fps).

Other games suffered too - UFO Terror From The Deep was so slow the mouse cursor was hard to track. It seemed I got totally ripped off paying £1000 for a PC that was less than half as fast as the DX-2 66 my friend bought for just £1100.

I was stuck with that sodding PC for 2 years, but at least it taught me how to use DOS and edit my AUTOEXEC and CONFIG to squeeze every possible bit of performance out of it.

As it happens I got buggered again when I bought my 3rd PC. They guy sold me a Pentium 133 for £1400 but when I got it home and booted it up the BIOS read "AMD K5 100MHZ"... I called him and he assured me the K5 chip was actually FASTER than a real Pentium 133 (despite costing half the price, no doubt why he sold them to his sucker customers). So I loaded Quake and did a timerefresh, only to score 11fps compared to my friends P133's 19.6.

The happy ending was I stole a Pentium 200 CPU from the exact same guy's shop over a year later and my timerefresh score went up to 32fps!

So yeah, the SLC was bad... performed similar to the laptop 486 33 chips of the time.

I think in the 486 era that by far the best thing to do was buy a 50 or 66 MHz chip. They were usable for the longest. They were a sort of sweet spot. You could get the Cyrix and AMD 486s pretty cheap too.

486DLC was actually decent compared to a real 486, but the SLC was pretty constrained with the narrow memory bus, and the fact that most of the boards didn't have any external cache. Hell, I would be willing to bet in a lot of cases people didn't even have the tiny 1k L1 turned on either. It usually just caused problems anyway. So what you were using was closer to a 386SX (similar to a 286 in many ways) than a 486.

Great job on stealing that P200 from the asshole that screwed you on the K5.

Iris030380 wrote:

My second PC was a 486 SLC 33 and it was absolutely abysmal. It cost almost as much as the 486 DX 33 but the man in the shop ass […]
Show full quote

My second PC was a 486 SLC 33 and it was absolutely abysmal. It cost almost as much as the 486 DX 33 but the man in the shop assured me that the SLC was almost as powerful. How he lied...

My friends had a 486 DX 50 and a DX2 66 respectively. With my SLC I could only play Doom in low detail and by reducing the screen size a few times. In other words, it performed more like a 386 than a 486. Doom 2 was out of the question, being a slideshow of about 5 fps in high detail and barely playable in low detail (about 10-12fps).

Other games suffered too - UFO Terror From The Deep was so slow the mouse cursor was hard to track. It seemed I got totally ripped off paying £1000 for a PC that was less than half as fast as the DX-2 66 my friend bought for just £1100.

I was stuck with that sodding PC for 2 years, but at least it taught me how to use DOS and edit my AUTOEXEC and CONFIG to squeeze every possible bit of performance out of it.

As it happens I got buggered again when I bought my 3rd PC. They guy sold me a Pentium 133 for £1400 but when I got it home and booted it up the BIOS read "AMD K5 100MHZ"... I called him and he assured me the K5 chip was actually FASTER than a real Pentium 133 (despite costing half the price, no doubt why he sold them to his sucker customers). So I loaded Quake and did a timerefresh, only to score 11fps compared to my friends P133's 19.6.

The happy ending was I stole a Pentium 200 CPU from the exact same guy's shop over a year later and my timerefresh score went up to 32fps!

So yeah, the SLC was bad... performed similar to the laptop 486 33 chips of the time.

Are you sure you're talking about the same chip? The Texas Instruments 486 SLC/DLC were upgrades to extend the life of 386 motherboards and were never meant to be used as a substitute for a real 486 machine. The only 386/486 hybrid that could be said to even come close was the snap on upgrades that Evergreen made to fit over the surface mounted PLCC 386 chips using the IBM version because it had a much bigger cache than the Cyrix and Texas Instruments chips. I can see where some unscrupulous dealers might have loaded up some older 386 motherboards that they wanted to get rid of and try to convince people it was the same as a real 486, but that's something you should have investigated before buying instead of blaming the lack of performance on the chip. You also needed a software program to activate the onboard cache which boosted the performance so if you never got that from the seller, your machine would run even slower.

I just got the chance and time to perform some brief testing:

- with CPU cache disabled in BIOS, SpeedSys score is 6.87 and SysInfo score is 22.4
- with CPU cache enabled in BIOS, SpeedSys score is 16.56 and SysInfo score is 54.2

Looks like a real competitor for my 386DX-40 / 256K cache system... a few interesting things are:

- absolutely nothing (scores, info, etc) changes if the Turbo switch is jumpered or not
- SpeedSys identifies the CPU somehow correctly as Cyrix 486SLC, but of 11 MHz
- the barrel-type battery (clean, no leakeage) seems to be holding BIOS settings successfully (at least over a few days' time)

A (poor quality) photo is attached...

Apparently some of the Socket 1/2/3 Cyrix 486 clones were slower than the AMD and Intel chips. I think they may have been bugged at one point too because I remember working at a place a decade ago and they had a tray of Cyrix 486s (green heatsinks) from RMA replacements they had done.

I know Cyrix always claimed that 486SLC/DLC chips weren't meant to compete against the intel chips, but come on...they were clearly named in such a way to fool the uninformed. A lot of people got burnt by this "scam".

The first Socket 1/2/3 Cyrix 486 chips only came with 2kb of internal cache, versus the full 8kb of the real deal. That probably explains why it was a bit slower (but not by much). As far as I know those 486S chips are kind of goofy. They're 4V, and might be based on the DLC core. They're writeback cache too as far I can remember.

The 486slc chips by cyrix, even the 50mhz slc2 ones where only somewhat comparable to the slow 386DX chips IIRC. A 50mhz slc2 I use to upgrade my IBM 386sx pc helps but a 386DX still beats it.

On the other hand the 486slc2 chips that where made by IBM where ultra-optimised and speedy! They where only outperformed by the 486DX2/66 IIRC. Sadly I never owned one...

Remember, all the above are 386sx clones in reality with various optimizations

edit: found some links

http://redhill.net.au/c/c-3.html#cx486slc
http://redhill.net.au/c/c-4.html#slc2

It's really the lack of cache that bogs down the 386SX systems. I have a pretty sweet 25MHz 386SX board with 32kb cache and memory interleaving. With the Cyrix SRx2-50 chip installed, I get performance similar to a 386DX-33.

Markk wrote:

I think that the 486SLC is a 486sx compatible, as opposed to the 486DLC which is more like an enhanced 386DX.

I've been reading an old magazine last night, and I saw that I didn't remember well. Sorry guys. What I had in mind that meant to be like the 486sx, was the Cyrix 486S33 , not the 486SLC.

iulianv wrote:

I just got the chance and time to perform some brief testing: […]
Show full quote

I just got the chance and time to perform some brief testing:

- with CPU cache disabled in BIOS, SpeedSys score is 6.87 and SysInfo score is 22.4
- with CPU cache enabled in BIOS, SpeedSys score is 16.56 and SysInfo score is 54.2

Looks like a real competitor for my 386DX-40 / 256K cache system... a few interesting things are:

- absolutely nothing (scores, info, etc) changes if the Turbo switch is jumpered or not
- SpeedSys identifies the CPU somehow correctly as Cyrix 486SLC, but of 11 MHz
- the barrel-type battery (clean, no leakeage) seems to be holding BIOS settings successfully (at least over a few days' time)

A (poor quality) photo is attached...

I'm surprised this is a surface mounted chip. The only TI SLC/DLC chips I have seen were all PGA 132 like these.

I know IBM made PLCC chips for their own motherboards because that was part of the deal they made with Intel to use the 486 technology was that they would not compete directly with them for CPU sales but this is the first time I have seen a TI version in PLCC.

386SX was pretty much only available as surface mount (okay, so maybe prototype and military versions are PGA, but that doesn't count), so if you're going to make a pin compatible 386SX part, how else would you go about it than use surface mount or clip on?