วีดีโอ

The AST is a 4066d. Nice all in one but it would be 100 times better if it had a freaggin feature connector on the board!

The Magnavox is interesting in other ways too: the CPU is actually located on a daughtercard that plugs into a connector on the motherboard, together with the system oscillator and a little glue logic. Probably these systems were upgradable, and in fact I know they were because I have 2 of those CPU cards, one 25MHz and one 33MHz. Been thinking of reverse engineering the daughtercard and maybe making a new one with say a 486SLC40 or SLC/2-50.

I was a bit lucky but I've been watching the CPU gold recovery auctions for a few months trying to get 3x GE7 geometry engines for my Iris Indigo that I've yet to feature on the channel and this came up during that. For the upgrade, if you have a 25MHz card something like an SLC2 25/50 would fit in there fine. Absolute worst case you end up with a 50MHz 'SX' but even with the barb register instead of the flush input there is a decent performance bump.

@@DKJones96 yeah I have a few 486SXLC2-G50; unfortunately they are of the 3.3V variety, so they would need voltage conversion to work on this system (yes the I/O pins are 5V tolerant, but the core voltage is not, 3.3V is required and 5V will fry the core)

@@paulpsomiadis5847 oh man! Yeah this cpu ran over 100c when I didn't have anything blowing on it. With a fan is a much more manageable 70c.

@@SudosFTW yeah it gave me some fits even after soldering the processor pins back. I will run it for a bit and see how I like it. Hard to complain about the performance the 8kb TI model gave.

@@Blackkspot I don't see why not. It would only run 16/32 speed but it should work. Using the barb input for the cache alleviates the lack of flush control on an older system. I bet that 16 might take fine to a 4mhz overclock if you wanted to push 20/40.

Yes, it worked on my IBM PS/2 55sx, but the gain was offset by the proprietary bottlenecks on that terribly slow system. If you have a reg. PC and can upgrade the storage and graphics, you might get better results. Just memory alone is not enough; I had 8 MB on board, +8 MB on a card, but the 55SX was bottlenecked by the 16-bit bus, 8-bit non-accelerated VGA, and slower-than-ESDI hard drive. I didn't want to throw more money on expensive PS/2 compatible parts, so I gave up. You need a later 386sx-16; the earlier ones (A step) lacked a float pin to disable the original CPU. My PS/2 was purchased in 1991. Google this phrase: 'cyrix 486srx2 float pin' I don't have that PS/2 or the upgrade anymore. I might have a copy of the Cyrix utility floppy somewhere. I remember it was difficult getting the upgrade to stay attached to the original CPU. I had plenty of room, but that thing kept detaching at the slightest bump to it.

Yeah the gasket is a bit chonky but it expands outward and gives you room for error which is probably inevitable with this kind of manual job.

I did try goo gone on the glass, it got rid of the word stuff at the bottom but the fogginess it didn't touch. Seems like an outgas reaction with the acrylic from the rubber gasket.

Here you go! They appear to be about $13-ish now. www.ebay.com/itm/176352881630

Thanks! I think leaving out the solder paste that I was tempted to use for applying it to the motherboard is what helped. Just throw on more flux!

The 386SX has 8x 32-bit internal registers. Outside of that it has no cache like the 486 series did. The Cyrix units like my Cx486SLC-40 runs 1KB of internal cache and that speeds things up considerably. This TI unit has 8KB. There are some others out there running 16MB but I don't have those on hand.

@@ceruleanserpent387 it should! But it is having an issue with that right now having to do with the cache.

CF cards always gave me issues on PCs as a boot drive tbh, even ones meant for internal use. Had a lot more luck with the cheap SD card to IDE adapters, funnily enough on Macs it's kinda the opposite situation - just goes to show how temperamental old machines are with hardware lol.

I have had surprisingly good luck with CF cards. Pretty much anything I have that runs an IDE drive has a CF card these days. My VLB 486 has run the same CF card for going on 10 years and it has a ton of time on it. I love them for these quick tests because I can just pop it into my modern machine, copy a program, and then slot it back into the machine without hassle. My AST 486 All-In-One also runs a CF card. Some of the issues that people run into could be old CF cards? I struggle to think of a CF card in my shop that has died or corrupted data. Better go knock on some wood....

@@amirpourghoureiyan1637 Even more mondern machines? My Voodoo Pentium 3 rig has run the same 32GB CF card for 5+ years.

@@DKJones96 yeah I've used Sandisks and Transcend cards but they just don't want to boot on PCs like my Athlon and Pentum 4 builds, they work fine on my G4 and G3 systems though. The CF cards are a lot more expensive where I live than SD cards so I've just used them for a long time now.

Congrats! You won the giveaway! Email me to my email address in the About section of the channel and I will work out getting your processor shipped!

Swap is done but getting a 16H error. I think it's not liking the overclock so I'm working to get some 40MHz crystals in by the end of the week. Might use a function generator to make the 40MHz happen before that just to continue troubleshooting. In theory, without enabling the cache it should act exactly like a 386... It mostly posts though so it was a successful transplant!

That is strange and sad. This overclock is not to far away and the M/B can do 25MHz bus. I assume you have a 5v 486SXLC2-40. Try 40Mhz and then increase the clock on the fly to 50 if you can. With my TI486SXLC2-50 (3.3v, a common chip), the 2x multiplier never worked but this is a M/B issue I think, and my max there was 1x40Mhz. With the cache disabled this was actually a bit slower than a 386sx 40MHz. Your situation is different. I am curios about the results. Now because this M/B has cache I hope you do not face problems enabling the internal cache. Oh and if it works 2x20 = 40MHz with all enabled. it is more or less safe to multiply all predictions with 0.8 (maybe a bit lower?). So mine is 16.5fps something. This is not 100% true as memory refresh is usually triggered by the reference clock, so the overhead is higher on lower bus speeds.

@@atheatos I got it! It was a BIOS issue and someone on the VOGONS forum uploaded a bios dump of their board which was essentially the same thing but a different brand and an AMI BIOS. I think this chip will probably do 50MHz without much coaxing and I will try that soon. I'm off next week so, at worst, I'll have everything uploaded next week. I do have a question though. The motherboard supports enabling and disabling of the cache but I have to use the utility to do it and I've read that the multiplier setting is where the internal cache enable register is but can't find the right utility or what other register value I need to change to actually enable the internal cache... What did you use for your SXLC2?

Nice! These are good news. On my board I tried many Bios but never managed to enable the 2x mode (chipset revision problem?). I will wait now for your the results :) , this is very interesting. 50MHz should be possible, maybe more. Basically all these "Cyrix" CPUs have the same configuration registers. So the tools to set them are the same. The only difference between 1x and 2x CPUs is that the bit that is used to select cache mapping, is instead used to enable 2x mode. So: "cyrix.exe -c[d] Two way assoc. [direct-mapped] cache. -cd for SXL 2x mode" In Bios when you enable cache it sets these registers to some predefined values. These values are rarely optimal, so it is good to use the tool either-way. Check my videos on 486DLC and 486SXLC2 for more. In general you have to play a bit with different settings based on your config. Make sure A20 line is connected to your CPU and maybe optional the Flush signal. Maybe also play with the KEN input for more performance.

That is good news, waiting for the results now :) Regarding configuration... All these CPUs use the same configuration registers. So the config tools are the same. When enabling cache in the Bios... the resisters are given some presets that usually are not optimal. The difference between 1x and 2x CPUs is that the bit responsible for setting cache configuration is repurposed for enabling x2 mode. So: "cyrix.exe -c[d]" Two way assoc. [direct-mapped] cache. -cd for SXL 2x mode. Check also my video on the 486sxlc2. Happy testing :)

Some early games had trouble with system 7. I run into that a lot trying to play some earlier system 7 and 8 games in even 9.

@@DKJones96Yeah, we weren’t even running System 6 on the 512K.

I wish. That cache card is tight and there is zero room for sockets unless I design a card to replace it. I could design a card to replace it...

Could be. The cache result being lower is odd but it could be a rare situation where cache gets in the way of an already difficult situation for the processor.