Okay, thanks A.C. !

I also found a package named "cyrix100.zip" here:

ftp://ftp.chatnfiles.com/Simtel-CDroms/Simtel … 2/disc2/sysutl/

Old thread, i know.
I just wrote 2 small programms to (de)activate Clock-Doubling on the 486sxlc2-50

Just use this debug-script, it'll give you "double.com" and "single.com"
Works perfect on my 386sx modded laptop with a TI486SXLC2-50 soldered in.

1a
2mov ax, 00c0
3out 22, ax
4in ax, 23
5or ax, 0040
6mov cx, ax
7mov ax, 00c0
8out 22, ax
9mov ax, cx
10out 23, ax
11mov ah, 0
12int 21
13
14rcx
1519
16ndouble.com
17w
18a107
19and ax, ffbf
20
21nsingle.com
22w
23q

Cool. Thanks.

It's worth mentioning that the Cyrix-Tools seem to reset the doubling because they don't know about it.

Does anyone know the correct / useful settings for the normal SLC/ DLC CPUs related to caching stuff?

Maybe i can integrate all together in the tools which are available as source code of the CX486DLC-Processor Configuration Utility.

Does anyone have benchmarks?

Checkit gives me 4250 / 4800 (without / with doubling) but also weird MHz-readings (160-220MHz)
Speedsys only displays "Cyrix" and freezes in Real-Mode.

Edit: cyrix -i1 -f and double clock gives me 14105 and norton sysinfo8 is correctly detecting the clock.

Addendum:

On SXLC2, the Clock-Double-Bit sits in the "Internal Cache Mode"-Bit from the SLC/DLC.
When using cyrix.exe, you can switch with -c.
Direct-Mapped = Clock doubling
2way associative = Normal clock

Does anyone know if Evergreen ever made a 66MHz version of the SXLC2 upgrade for 386SX systems? So far all examples I have seen are based on the 50MHz part. 66MHz SXLC2 CPUs do exist, however they seem to be quite rare.

Also, I wanted to put this out there for anyone thinking of trying a CPU swap. There are at least three versions of the SXLC as far as I can tell.

There's:

TI486SXLC2-050 : This should be purely 5V part. The 40MHz version is readily available. I haven't seen too many 50MHz parts though.

TI486SXLC2-G50 : This seems to be the most common version available. In the manual it is listed as a hybrid 3.3V/5V part. It is described as being 5V tolerant (I take that to mean 5V I/O with 3.3V core). However, the manual doesn't make it clear either way.
I saw photos from a guy online who sliced off the 386SX chip in his Sega PC, and soldered a G50 in place. I believe it worked, but who knows how long it will last if the core is really 3.3V.

TI486SXLC2-V50: This should be a purely 3.3V part. I think I've only really seen these on paper. I believe they use 3.3V I/O, which makes them pretty weird. Maybe they were used in laptops. (*edit*, online you can find pictures of a 25MHz version)

At least for now, the chip I recommend for a 5V swap is the SXLC-040. These are common, and will work in clock doubled mode too

Anonymous Coward wrote:

TI486SXLC2-050 : This should be purely 5V part. The 40MHz version is readily available. I haven't seen too many 50MHz parts though.

At least for now, the chip I recommend for a 5V swap is the SXLC-040. These are common, and will work in clock doubled mode too

I only found one seller who carries a desirable model and it's the Tx486SXLC2-40

Where can you get a TI486SXLC2-040? You say they are readily available...

Thanks

The only new part I can find now is this one but the naming convention doesn't follow the 1994 datasheets.

Searching for 9408FOB1 doesn't turn up anything either.

Anyone knows if it's a 5v, 3.3v or 3.3v core / 5v IO?

These chips are pretty funny. My guess is they come from an early batch before product lineup was finalised. As far as I know the SXLC-40 and the SXLC2-40 are completely identical except for the top markings. All the TI SXL and SXLC chips support clock doubling whether they have a "2" in the name or not. The chip you have is almost certainly the 5V only version. -040 and -40 markings are the same.

The 5V SXLC-40 that I mentioned as being readily available can be found on taobao for about $10USD.

I also verified that an Evergreen Revto486 with a 66MHz SXLC2 chip was infact released. I think there is a Chinese collector that has one.

Not to be rude but "readily available on taobao" is true only in China. As it happens I'm also looking for a SXLC-040 chip - is there any way to get those in EU? Through AliExpress.com maybe?

I've put a G50 part in place of 386SX before and it works fine. It's a 3V3 chip with 5V tolerant I/O but it's absolute max rating for VCC is 5.5V so it should survive. But I'm not sure if it can cope with 40MHz bus, whereas the 040 chip is described as compatible (and true 5V). If I can't source a 040 I guess I'll solder in another G50 instead.

There are websites that can help you get things off Taobao, just as there are to get things from yahoo auctions Japan.

I'm pretty sure the SXLC2-50 can cope with a 50MHz bus. I have the 386DX version, the SXL2-50, and I run it at 1x50 with no problems.

Shipping costs seem rather costly for many of them it seems though. Maybe buying them in bulk would help ?

Edit : nevermind, there are many of them available on ebay for 15$ with free shipping !

I just got me a TI486SXLC2-G50-PQ off ebay for a song. can it actually be verified if these will work just soldered right onto a board?
The board it's going onto is in an NEC PowerMate SX/20vi. more specifically, one custom-ordered for FedEx for the PowerShip 2 line of PCs during the Federal Express => FedEx rebranding.
someone was naughty and put a 16MHz chip in at the factory, probably to save money, but I've confirmed it's running at 20MHz. so all in all, I'd expect the stock clock to be 20MHz and the cock doubled speed to be 40.

...or would I absolutely need to go and change the oscillator with this upgrade as well? it wasn't in the budget to go and grab one at purchase time, but if I need to I'll end up shelving the upgrade until I grab one quick. Kinda don't want to do that since there *is* a 20MHz Intel 387 in there. I want to be able to keep that if possible and not have to buy another of those.

You can actually solder a 3.3V SLC in place of a 5V SX if you attach a heat sink. It's good up to 50MHz. Gets hot, but works fine.

ph4nt0m wrote:

You can actually solder a 3.3V SLC in place of a 5V SX if you attach a heat sink. It's good up to 50MHz. Gets hot, but works fine.

mine's 5v. just making sure I don't have to add in some weird circuit off a pin or something.

According to this:

Anonymous Coward wrote:

I've used TI486SXL as a direct replacement for 386DX in several boards. Some work okay and others don't. I suppose it's a bit more of a pain in the ass if you solder in an SXLC chip to your 386SX board and it doesn't work. At the very least all of my boards booted up just fine, some just didn't like certain features turned on. It was years ago and I don't remember specifics.

The Tx486SLC shouldn't have worked any differently than a Cx486SLC. It is an exact copy. Cyrix used TI's fabs to make the chip. The only difference between the two should be the label. TI 486SXL(c) chips are basically the same as the 486DLC except that they have a larger cache and remap one of the registers to turn clock doubling on or off.

I should be good. however with how old that post is (6 years in this thread!) I'm just making absoutely sure I'm gonna be okay to go through with this before doing an upgrade that has a heavy percentage chance of not working. the BIOS is from like 1993 or 1994 so I'm expecting no real issues, but there's always that chance.

Sudos wrote:

ph4nt0m wrote:

You can actually solder a 3.3V SLC in place of a 5V SX if you attach a heat sink. It's good up to 50MHz. Gets hot, but works fine.

mine's 5v. just making sure I don't have to add in some weird circuit off a pin or something.

just wanted to update this, mine turns out is 3.3v and not 5v. the TI486SXLC2-G50 is designated as a 3.3v, but 5v tolerant chip in the datasheet. the 5v tolerance is for the I/O only. not for the core voltage which it still expects to be 3.3v. the box doesn't like it and isn't POSTing, and I've probably fried the CPU anyway. will try dialing the 5v rail down a bit and try again tomorrow, but for now, I think it's toast and may as well just get a proper 5v chip to drop in.

basically: I have a lot of soldering to do with replacing the newly-added-in SXLC2 with an AM386SX-40 since I've given up. more cache is just not worth the trouble.

Are you sure it says in the datasheet that the 5V is for the I/O? I've looked at the datasheets, and while you can make that conclusion I don't recall it specifically stating that anywhere.

As far as I know, the 3.3V and 5V SXL(C) chips were manufactured using the same die process, so in this case its unlikely you'd kill the CPU running it at 5V. It seems the 3.3V version is actually just a handpicked 5V version that still works at lower voltages.

I'd double check the soldering job to make sure nothing is shorted.

Anonymous Coward wrote:

Are you sure it says in the datasheet that the 5V is for the I/O? I've looked at the datasheets, and while you can make that conclusion I don't recall it specifically stating that anywhere.

As far as I know, the 3.3V and 5V SXL(C) chips were manufactured using the same die process, so in this case its unlikely you'd kill the CPU running it at 5V. It seems the 3.3V version is actually just a handpicked 5V version that still works at lower voltages.

I'd double check the soldering job to make sure nothing is shorted.

Nothing is shorted, chip is on correctly. The 1994 datasheet doesn't seem to properly list this specific chip, but from what I can gather, I may need to bridge one of the NC pins to 5V for it to get working properly, since there's supposed to be voltage on a vcc5 pin and it's supposed to be NC for everything else not an SXL2-G chip. May try that out first.

wanted to do another quick update. Looking around at a datasheet for an SBC, I think the *proper* pinout has been found.
on Pin 45, which is NC on all other literature, there's an IOV + pin. on the TI486SXLC chips for a QFP144? package, there's a VCC5 pin for +5 to the I/O. I wonder if this is needing to be lifted and manually tapped to +5 to get I/O power. since i have two boards and one is in worse shape than the other, I can try and see if this is the case. Pics below are the pinouts for the QFP100 package here and the pinout from the schematic of the 486 SBC.

The TI486SXL pinout for QFP144:

The pinout from the SBC schematic for an SXLC2-G50:

and finally, the pinout of a normal TI486SXLC, NOT SXLC2, since there is no literature in the 1994 datasheet for the SXLC2 in a QFP100 package:

Am I crazy (probably) or is this actually a valid course of action?

edit: it was a crazy idea that didn't work. I tried. back to getting the am386sx-40, oh well.

Have you cleaned the board well after soldering? What's displayed on your POST card?