VOGONS


New processors for old sockets.

Topic actions

Reply 41 of 134, by W.x.

User metadata
Rank Member
Rank
Member

I've read only beginning of thread... my points:
You can print and DIY lots of things, but not too microscopic. The smaller, and more complicated is something, the more machinery and chemistry, you need. And more complicated is thing, the more people you need, and very genial ones. So, chip like Intel 4004 is possible doing at home by DIY, maybe by one genial , rich individual, or maybe small team, but I think, 8008 is highest, where you can get. You wont get even on xt, or 286 chips, in house conditions... and his workshop for that chip that was in video mentioned here, was almost like mini-factory.

And here we are... for anything complicated, you need A FACTORY to make. Basicaly, factory is only cover for very complicated machines, and special conditions, and to gather teams. Many people with various functions. Only those can make more complicated chips, on smaller process. So even for 286, you would need factory I would say. There is lots of problems, to make a more complicated chip, and you need also lots of planning (for example, you need AMD team and factory, to reverse engineering 386, 486... like they did in 80's and 90's ... to make own 386, and 486. You cannot do it as individual, because those chips are already too complicated.

Can it change? Maybe... if there will be machines, that are capable doing so, and you'll be too rich, and too educated. You need to know how to process small things, you need technology, you have to know work with it, only one small mistake, and chip is damaged. I cannot imagine someone, who would do it at home... and when you gather team of people, and buy those expensive machines... then what. You make one chip, one 486... it would be expensive as hell (thousands of dollars)... it will be uneffective, if you buy such expeinsive equipment, and do only 1 or 2 chips. They were profitable (for example AMD), because they bought for millions equipment, make factory, and then... made 10000 chips, or 50000 chips... so it returned the money. But as a hobbist, you won't get anything. Noone pay you 1000$ or 2000$ for you 486 at 200 mhz. And as I said, you won't be able even make 486... I say, 8008 is highest what you can make at home. Otherwise, you need regular team, regular factory, etc... and that would cost millions and millions... so you just cannot make one or two chips. In other words... theoreticaly it's possible, but you need to invest millions of money, just to make few chips, that noone buys, because price will be too high. And if you sell them from 50$ or 100$, then you lost all money that was investition, to equipment, and hiring personal. In other words, 486 at 300 mhz would costs you for example 10 million dollars to make. 😀 Or even more. That's how I can say, noone will do it.

So here we are, new chips are not possible.
If you can modify somehow existing chips, that is other story, but for sure, you cannot do it inside the chip. You will damage it. It has many layers, you can do it again only very early and simple chips.

So no, it is not possible, it is not profitable, it will not happen in reality. What you saw in youtube video, where someone did at home something close to 4004, is highest, where someone will get. Unless machines for making such chips, doesnt drop by price for 5000$ or 10000$, I doubt someone will invest more, to make such chip, only to make interesting video, or post it on forums.

In other words, you would need to convince Intel, to make chips, from old architecture, but even then it would be lots of work (they have old designs on higher nm technology, they would need to shrink die, with new transistors and technology, it would again needed engineers to make... and they costs money to pay... lots of money... Intel would ask you "for what you need 14nm 486 ? How much you pay? How many chips you want to make?" ... and there is, where you would end... cause you answer would be... I need 1 or 5 or 10 chips, and I dont have money. 😀 And I need it for retro community, for hobbiests, so I can look good and cool to have 486 at 300 mhz. And that would be end of conversation with Intel, I would say. 😀
So, in reality, 486 made with lesser nm technology as it is, at 300 or 400 mhz, will never happen. You would need to be madman and billionaire, to make even one or several pieces, you would need to pay those big companies to make them and it would be problem to convince them to make such stupidity. 😀 So I would say, you would need A LOT of money to offer them to make such thing (something like Elon Musk when he decide to travel to moon or something, yes it's possible, you just need millions of $$$). Long story short... in my opinion, it will never happen. We will not see new 486, faster than 200 mhz, made by new technology (completly new chip), unless some madmen decide to throw millions of $$$ into trash.

Reply 42 of 134, by Sphere478

User metadata
Rank l33t
Rank
l33t

Okay, this just occurred to me, you guys are talking about making interposers for running other processors, but what about this,

Amd I’m told had smp on the k6 but it was never implemented on any motherboards right?

What about a interposer with a fpga that emulates intel smp to translate to k6 smp?

Maybe we can finally realize the dual k63+ dream?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 43 of 134, by rmay635703

User metadata
Rank Oldbie
Rank
Oldbie
Doornkaat wrote on 2021-11-12, 17:59:
BitWrangler wrote on 2021-11-12, 15:26:

(I actually have no idea [...] whether it was just a geek urban legend.)

We can safely assume it was. 😉

Transputer existed

Pentium can be forced into 32bit read/write

My mention of a small board to drop in the cpu socket with cache as ram would solve many of the interfacing issues between a 486 and pentium

Might need to take a look if a real low level circuit diagram with microcode exists for the transputer upgrades might give some clues how you can use glue logic to deal with pipeline differences

Reply 44 of 134, by Doornkaat

User metadata
Rank l33t
Rank
l33t
rmay635703 wrote on 2021-11-12, 23:56:
Transputer existed […]
Show full quote
Doornkaat wrote on 2021-11-12, 17:59:
BitWrangler wrote on 2021-11-12, 15:26:

(I actually have no idea [...] whether it was just a geek urban legend.)

We can safely assume it was. 😉

Transputer existed

Pentium can be forced into 32bit read/write

My mention of a small board to drop in the cpu socket with cache as ram would solve many of the interfacing issues between a 486 and pentium

Might need to take a look if a real low level circuit diagram with microcode exists for the transputer upgrades might give some clues how you can use glue logic to deal with pipeline differences

The story was 'somebody didn't want to wait for Pentium Overdrives to get less expensive so they made an adaptor'.
While it is certainly possible to create a device that translates between the 486 board and the Pentium CPU this device would either
-cripple the Pentium's performance to the point where the CPU upgrade would become a downgrade
-or (even today but more so back then) far exceed the cost of a new Socket4/5/7 board (keep in mind the person would have to buy a Pentium CPU regardless).
I'd assume that if this was in any way feasible there would have been a commercial version of this adaptor to upgrade 486 systems to even later Pentium (MMX?) CPUs, not just one hobbyist's diy project.
This is why I say it is implausible for the story to be entirely true.

BUT legends are always based on something. In this case I suspect that
-confusion between POPD5V63/83 vs. POPD5V133,
-adaptors for 3.xV 486 CPUs on 5V motherboards,
-rare adaptors for S5 Pentiums in S4 boards
all blended into one glorious adaptor that allowed a nifty hacker to just stick his new CPU in his old system and save some money.
This is why I think that it is plausible how a story like that would come to be despite the fabled artifact never having existed.

Of course I can not prove this adaptor didn't exist. (Burden of proof isn't on my side anyway. 😉) To me it's a question of plausibility and while the story leaves a lot of questions that make it implausible we can easily come up with an explaination on how the story emerged based on confusion about other upgrade options.
That's why it's much more likely to be fiction than fact.

Reply 45 of 134, by HangarAte2nds!

User metadata
Rank Newbie
Rank
Newbie

I haven't read all these posts because there seems to be a lot of tangents, however, I wanted to add my $0.02 and say I think there is rather more need of new sockets for old processors. And AT cases. That is a real problem these days. Original ones are not cheap. The only one I have is proprietary so no shuffling around, sadly. Looks like my new 486 is going on an open mining frame for starters 😥 I will be modding a late '90s ATX case to fit an AT board but that is not a high priority atm.
Luckily I have an ATX SS7 board for my MMX.
Hardware from the '80s through the mid-90s is not a big focal point for my hardware collecting. I have a 386 and building a 486. I have 486 and Pentium laptops. Those satisfy my need to play DOS games on period hardware. However, it would be nice to have the option to build other AT based machines without having to spend a fortune. Pentiums and 486s are still pretty cheap but the motherboards are really expensive. There is absolutely no reason why someone couldn't put Socket 3 and SS7 boards into production in 2021. I would love to have an ATX Socket 3 motherboard with fully integrated I/O and adjustable bus speed, just so I could not have to deal with old PSUs and cases. I think my interest in playing around with 486 overclocking would go way up.
Sadly, there is probably not enough interest. But wait a few years and maybe that will change. When the cost of new manufacture motherboards on a not so big scale is significantly cheaper than working originals, it is likely to happen.
In theory, the CPUs can be remade too and I bet someone somewhere can turn out a 350nm die shrink of a 600nm CPU from 25 years ago... Patents are expired.

Reply 46 of 134, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
Sphere478 wrote on 2021-11-09, 06:24:

Can we have a faster processor for socket 3 than the pentiun over drive

Then you build a Pentium/K6 system on Socket 5/7.

Sphere478 wrote on 2021-11-09, 06:24:

, socket 7 faster than the k6-3+

Then you build a Pentium 3 or Athlon System.

Sphere478 wrote on 2021-11-09, 06:24:

, a faster processor for 370 than the tualitin?

Then you build a Pentium 4 or Athlon XP system.

None of these platforms are rare, expensive or difficult to get. This is why there is very little desire from people to build faster CPUs for older platforms.

As others have pointed out, at some point, the platform becomes a major bottleneck and it needs to be replaced to make the computer be quicker.

Reply 47 of 134, by 386SX

User metadata
Rank l33t
Rank
l33t

To expect homemade CPUs I suppose it's impossible even on papers but companies that made alternative adapted CPUs for older sockets were always interesting just like others adapters. Things like the Cyrix 486DLC or the Overdrive CPUs concepts were amazing to boost perfectly functional old computer in the 90's I suppose. Not only as a benchmark reason.
For the Socket 7 I suppose even with a powerful CPU, the chipset/ram speed would still be an old and slow bottleneck anyway and as said at that point any later configs would have much more sense. At the end the Socket 7 lasted even too long.

Reply 48 of 134, by Nexxen

User metadata
Rank Oldbie
Rank
Oldbie

One question, could a pentium core be adapted to fit a 386 system? 486s were (ok, with limits)
What modifications would be required (at least)?

PC#1 Pentium 233 MMX - 98SE
PC#2 PIII-1Ghz - 98SE/W2K

Reply 50 of 134, by BitWrangler

User metadata
Rank l33t
Rank
l33t
canthearu wrote on 2021-11-13, 08:41:
Then you build a Pentium/K6 system on Socket 5/7. […]
Show full quote
Sphere478 wrote on 2021-11-09, 06:24:

Can we have a faster processor for socket 3 than the pentiun over drive

Then you build a Pentium/K6 system on Socket 5/7.

Sphere478 wrote on 2021-11-09, 06:24:

, socket 7 faster than the k6-3+

Then you build a Pentium 3 or Athlon System.

Sphere478 wrote on 2021-11-09, 06:24:

, a faster processor for 370 than the tualitin?

Then you build a Pentium 4 or Athlon XP system.

None of these platforms are rare, expensive or difficult to get. This is why there is very little desire from people to build faster CPUs for older platforms.

As others have pointed out, at some point, the platform becomes a major bottleneck and it needs to be replaced to make the computer be quicker.

For real, this.

There's already a way for all these platforms to put something on it that shows up the weaknesses of the platform, and yes, it's probably gonna be the priciest option of what to put on that platform. Start thinking about new silicon though, and the fact you'll have a hard time gathering a few hundred people to care enough to spend any money on the ultimate super CPU for it, and then you'll realise the million dollar foundry cost would have to be borne on that narrow and tight fisted market. So sure, 300mhz 486, $2000 a pop if you can get 500 deposits. Then you realise that all the ppl yelling "make new ones, make new ones" were all the people who didn't wanna pay $100 for a dx5-133 or spend $200 assembling a basic Pentium system.

There's nothing "on the table" here, all platforms have some performance overlap at the top and bottom ends, so extending that overlap takes you from a 486 that feels slow in comparison to a Pentium 100, to a 486 that feels slow in comparison to a Pentium 166. If you want the genuine feeling of the chipset bottlenecking underneath you put 80ns FPM RAM in a pentium board with all the wait states CMOS supports, set the bus to 50, the multi high and have at it. There you go, now you have your P200 effortlessly performing at the "slow in comparison to a Pentium 166" level that a 5x or 6x multiplier 486 would.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 51 of 134, by Sphere478

User metadata
Rank l33t
Rank
l33t
HangarAte2nds! wrote on 2021-11-13, 08:10:
I haven't read all these posts because there seems to be a lot of tangents, however, I wanted to add my $0.02 and say I think t […]
Show full quote

I haven't read all these posts because there seems to be a lot of tangents, however, I wanted to add my $0.02 and say I think there is rather more need of new sockets for old processors. And AT cases. That is a real problem these days. Original ones are not cheap. The only one I have is proprietary so no shuffling around, sadly. Looks like my new 486 is going on an open mining frame for starters 😥 I will be modding a late '90s ATX case to fit an AT board but that is not a high priority atm.
Luckily I have an ATX SS7 board for my MMX.
Hardware from the '80s through the mid-90s is not a big focal point for my hardware collecting. I have a 386 and building a 486. I have 486 and Pentium laptops. Those satisfy my need to play DOS games on period hardware. However, it would be nice to have the option to build other AT based machines without having to spend a fortune. Pentiums and 486s are still pretty cheap but the motherboards are really expensive. There is absolutely no reason why someone couldn't put Socket 3 and SS7 boards into production in 2021. I would love to have an ATX Socket 3 motherboard with fully integrated I/O and adjustable bus speed, just so I could not have to deal with old PSUs and cases. I think my interest in playing around with 486 overclocking would go way up.
Sadly, there is probably not enough interest. But wait a few years and maybe that will change. When the cost of new manufacture motherboards on a not so big scale is significantly cheaper than working originals, it is likely to happen.
In theory, the CPUs can be remade too and I bet someone somewhere can turn out a 350nm die shrink of a 600nm CPU from 25 years ago... Patents are expired.

What’s the name of that place that is selling tx ide and what was it, new 8086 motherboards?

Lemme search

https://yeokhengmeng.com/2020/05/review-of-a- … oard-nuxt-v2.0/

Here it is, maybe we could talk the creator of this thing into tackeling socket 7/socket 3?

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 52 of 134, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
cyclone3d wrote on 2021-11-11, 16:43:
Would there really be any reason for a faster S370 CPU? […]
Show full quote

Would there really be any reason for a faster S370 CPU?

You already have up to 1.4Ghz which can be overclocked.

Past that, you can go to Socket A, which can net you up to around 2.3Ghz+ on a KT7A with a simple socket wire mod and modded BIOS.

Then if you want even faster with AGP and full ISA support, you can go with something like a PIAGP s478 or LGA775 system though the LGA775 SBC is impossible to find.

Now if you could have a CPU that could have very granular speed control on the hardware size... that would be excellent.

I doubt that having (much) faster s370 CPUs would add as much as unlocked multipliers would.
At best I could see something like (I'm just tossing up something here) a 2GHz 1MB Cache Pentium 3 work, but it would be just as 'useless' as a 500MHz 486.
It wouldn't really help much since the rest of the hardware would probably just hold it back.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 53 of 134, by BitWrangler

User metadata
Rank l33t
Rank
l33t
Tetrium wrote on 2021-11-13, 17:15:

a 2GHz 1MB Cache Pentium 3 work, but it would be just as 'useless' as a 500MHz 486.
It wouldn't really help much since the rest of the hardware would probably just hold it back.

I've already got one, for some reason they added an extra megabyte of cache and stuck it in a laptop though. Pentium M Dothan core.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 54 of 134, by whaka

User metadata
Rank Newbie
Rank
Newbie
Deksor wrote on 2021-11-09, 23:35:

Just imagine your old IBM 5150 running doom with the original board (and a futuristic computer shoved in the CPU socket)

there's a big difference between pc and amiga, the amiga had the ability to have his local bus extended.
this can easily make use of the accelerator local fast ram, local drive controller, graphics and all what you can imagine on an accelerator card, at full speed.
you'll only be crippled for chip ram, and chipset access. in short all what's dependent of the chipset bus.
in other words, amiga had the ability of 2 separate busses, but working together.

pc can't do that, as far i know.
that's why for upgrade you only saw hybrid cpu like sxlc, overdrive. all of them are "downsized" to fit the original cpu bus.
so, even if you can use a pi or whatever for replacing a 8088, with let's say a pi emulating a 1Ghz 486, you'll always be heavily crippled by the motherboard bus for ram access, graphics, etc...

but hey... maybe i'm wrong, and some magic can be done 😀

Reply 56 of 134, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
BitWrangler wrote on 2021-11-13, 17:21:
Tetrium wrote on 2021-11-13, 17:15:

a 2GHz 1MB Cache Pentium 3 work, but it would be just as 'useless' as a 500MHz 486.
It wouldn't really help much since the rest of the hardware would probably just hold it back.

I've already got one, for some reason they added an extra megabyte of cache and stuck it in a laptop though. Pentium M Dothan core.

Great, put it into your best Tualatin capable board and run some benchies on it for us to to oogle at 😁

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 57 of 134, by Sphere478

User metadata
Rank l33t
Rank
l33t

I wonder if a thread like this might be able to get the attention of one of those manufacturers in china that did stuff like the pentium golden tiger.

If they built cool stuff for old hardware like this I would buy it. A new super overclockable socket 7 motherboard for example, 🤷‍♂️

Does anyone here have any connections with manufacturers like that?

Tetrium wrote on 2021-11-13, 17:49:
BitWrangler wrote on 2021-11-13, 17:21:
Tetrium wrote on 2021-11-13, 17:15:

a 2GHz 1MB Cache Pentium 3 work, but it would be just as 'useless' as a 500MHz 486.
It wouldn't really help much since the rest of the hardware would probably just hold it back.

I've already got one, for some reason they added an extra megabyte of cache and stuck it in a laptop though. Pentium M Dothan core.

Great, put it into your best Tualatin capable board and run some benchies on it for us to to oogle at 😁

They were socket 479

Which can be adapted to socket 478, so you could actually put some very nice hardware behind that.

That would be a fun build, maxing out the last remnants of the pentium 3 heritage.

There was some pentium 3 stuff on the xeon side that might be of use for such a project if I recall

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 58 of 134, by Deksor

User metadata
Rank l33t
Rank
l33t
whaka wrote on 2021-11-13, 17:36:
there's a big difference between pc and amiga, the amiga had the ability to have his local bus extended. this can easily make u […]
Show full quote
Deksor wrote on 2021-11-09, 23:35:

Just imagine your old IBM 5150 running doom with the original board (and a futuristic computer shoved in the CPU socket)

there's a big difference between pc and amiga, the amiga had the ability to have his local bus extended.
this can easily make use of the accelerator local fast ram, local drive controller, graphics and all what you can imagine on an accelerator card, at full speed.
you'll only be crippled for chip ram, and chipset access. in short all what's dependent of the chipset bus.
in other words, amiga had the ability of 2 separate busses, but working together.

pc can't do that, as far i know.
that's why for upgrade you only saw hybrid cpu like sxlc, overdrive. all of them are "downsized" to fit the original cpu bus.
so, even if you can use a pi or whatever for replacing a 8088, with let's say a pi emulating a 1Ghz 486, you'll always be heavily crippled by the motherboard bus for ram access, graphics, etc...

but hey... maybe i'm wrong, and some magic can be done 😀

One way it might be doable (maybe it's not as clean or maybe that's impossible ?) would be to treat the xt motherboard as just an ISA backplane ?

Trying to identify old hardware ? Visit Ultimate Retro - Project's thread The Ultimate Retro project - a stason.org/TH99 alternative

Reply 59 of 134, by BitWrangler

User metadata
Rank l33t
Rank
l33t

... or mechanical support.. hello Powerleap renaissance. 🤣

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.