mothergoose729 wrote:

the3dfxdude wrote:

Doornkaat wrote:

What would those chips be good for? I mean you would be able to benchmark a very fast 486 but then what? Games that need faster CPUs can be run on a Pentium or better. Even later games will require additional instructions and at some point the CPU is so restricted by bus bandwidths that increasing clock speeds won't do much for speed.
If you need more power than a real Am5x86 can deliver it's probably best to just change platforms.

Your question is like a mainframe designer asking what good is an 8080 microprocessor in 1975.

That analogy makes no sense.

He is right. A super duper charged 486 would be fun, but it isn't in any way practical - even for the purposes of retro computing.

*Sigh*

I don't see anyone claiming a super charged 486 must be made practical. Only detractors are saying something about practicality.

The point of the analogy is that the 8080 was seen as a TOY. The intel parts up to that point were nothing special, and were not unique in the industry. But what happened was that hobbyists tried it out anyway to find out what it could do. I think that is what was asked, right? Since a 200mhz clock speed does not seem to be possible on existing parts, someone is guessing it would just take bringing the 486 to a new process node. I mean, you if think there is no practical purpose, Intel already did it with the pentium. It's called Atom. To people in the industry that I talked to, they say it was a bad idea and a flop, so why did Intel do it? I guess it should have been practical (sarcasm). I would be trying to use it if it came in a PGA package, and wasn't an SoC...

200mhz is somewhat easy to obtain with temperatures below zero. Good enough for benchmarks.

the3dfxdude wrote:

I don't see anyone claiming a super charged 486 must be made practical. Only detractors are saying something about practicality.

Come on, dude. If you want that thing to become a reality it has to have a practical application justifying its price, otherwise it wouldn't be produced. I get that it's a nice dream to have a super fast 486. No argument there.
But the 8080 did have a practical application in the 70s. People bought it and not just a few.
What's the value of a fictional 486DX10? That's an extremely niche market. You were talking about tracking down some Cyrix engineer and having him explain how to reverse engineer a 486. This is what I was commenting on. I don't think you'd gain a chip that would be good for anything but benchmarks because there are already better working alternatives avaliable at a much lower price than this chip would cost.
I know of stuff like the NuXT but that's done with mostly NOS chips. There was no need to copy a CPU design on a smaller node.

But if you still think I'm a detractor and this is a feasible project, I would actually be happy to be proven wrong.
There are already companies like TechInsights Inc. that will analyse a chip like the 486DX2 for a five figure or low six figure sum. Maybe they have already done so, making the information much less expensive. I suppose they will also help you redesign the chip on a smaller node or direct you towards somebody that can help you with that step. Still to produce the resulting chip you'd have to pay licensing for a lot of Intel patents.
Alternatively either AMD or VIA should own the patents and rights to the Cx486. Maybe they will be open to a licensing agreement.

the3dfxdude wrote:

To people in the industry that I talked to,

You say you have contacts (who btw?), so really, please make this chip happen if you actually think you can! If I can pick this up for 30$ or less, working in a 3.3V motherboard right out of the box I'll definately buy one. Talk to a few people, get a cost esitmate and make a kickstarter.

All I'm saying is I don't see this happening because I don't think there's enough practical use for this chip to generate enough demand to be produced on a scale that makes it come out at a price where it'll be interesting to buy. Again, I would actually like to be wrong here.
At 30$ it's a toy to tinker with. At anything more than 50$ it's starting to become a little expensive for a toy and I'd really have to have a practical reason to buy one. Instead of spending that money I'd rather use my K6-III system and have all that backwards compatibility and more while being able to run any 1999 game on playable frame rates on 100% historical 90s hardware. Or simply use my DX2 alongside my K7.

the3dfxdude wrote:

I mean, you if think there is no practical purpose, Intel already did it with the pentium. It's called Atom.

I really don't think any of the Atom branded CPUs are simply P5 processors on a smaller node.

Overall it's pointless to recreate 486 CPU, because they are not very flexible and has many much better alternatives. And technically we still have stuff like IDT WinChip, which is the beefed up 486DX, performance wise.

Hehe I'll take a few 486 DX69s too :p.. while were at it make a fresh chipset and board capable of hugely faster bus and memory transfers.

Great dream but can't be done financially...yet.... maybe in another 25 years we'll be able to 3d print transistors on a scale of nanometers and make hobby processors. In the mean time the solution is a Pentium 133.. There is no way around the economy of scale, noone is going to tool up and fab a cpu that will sell in the hundreds or few thousands.

Legit am5x86 @ 200mhz was always a dream but I've never seen it done in an ambient environment. Still though, they're easy to run at 160mhz which a silly fast 486!

The Serpent Rider wrote:

Overall it's pointless to recreate 486 CPU, because they are not very flexible and has many much better alternatives. And technically we still have stuff like IDT WinChip, which is the beefed up 486DX, performance wise.

Totally with you. Both on the super fast 486 being pointless and on the C6 being sort of the better 486 core especially for its faster bus and MMX support.
And I don't believe in a new 486 Socket 3 CPU coming out any time soon.
All I'm saying is if the3dfxdude can find a way to make them for 30$ or less I'm buying one because at this point I can afford a pointless toy. 😉