Tronix wrote on 2021-11-13, 17:43:

BTW, [Jürgen] made his 65f02 CPU running at 100 MHz for direct replacement original MOS 6502 […]
Show full quote

BTW, [Jürgen] made his 65f02 CPU running at 100 MHz for direct replacement original MOS 6502

http://www.e-basteln.de/computing/65f02/65f02/

Pretty sweet! Hopefully soon we may have ones for later sockets

Sphere478 wrote on 2021-11-13, 20:14:

Pretty sweet! Hopefully soon we may have ones for later sockets

I don't think so. The one is pretty simple, they've just probably somehow die shrinked it, to get 100 mhz. But I don't know, didnt analyzed it, how they've improved it.
I doubt they ever get even to 286 level in home conditions, as I've said. I am sceptical to get even to 8086. But forget about socket 3 and socket 7. I mean 486 and Pentium. That's too complicated, some enthusiastic cannot do same work, as thousand of people, engineers, in laboratory and factory in early 90's.

Nexxen wrote on 2021-11-13, 12:13:

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)?

Put a transputer module in a 386 and place a p24t pentium overdrive in the 486 socket

AKA (Already done)

There is also a guy using a transputer on an inboard/386 to run doom and windows 3.1 on his IBM 5150

rmay635703 wrote on 2021-11-15, 00:26:

Put a transputer module in a 386 and place a p24t pentium overdrive in the 486 socket […]
Show full quote

Nexxen wrote on 2021-11-13, 12:13:

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)?

Put a transputer module in a 386 and place a p24t pentium overdrive in the 486 socket

AKA (Already done)

There is also a guy using a transputer on an inboard/386 to run doom and windows 3.1 on his IBM 5150

I meant a new PGA132.
As we are in imagination land, to retrofit newer cores/tech to max out an old architecture.
Pointless exercise but having a 5x86 to work @ 300mhz, a K5-500, 386@ 80mhz... is a 80's/90's old dream of today's old people.

It just makes me smile to imagine a new cpu like that.

Here can be seen how complicated is 4004, just for imagination. I doubt you can go any more complicated in home conditions. So even 8008 would be unimaginable achievement to reach.

15volt logic always impressed me, tri voltage stuff as well, very complicated solutions to very simple logic

I was wondering, is a smaller production process making a small difference for the same design (just upping mhz) or does it just beg for a new design to take full advantage (ok, more stuff on the same area then new ideas)?
Even if you have a 80mhz 286/386, bus limits all the extra speed, what's the point in having same tech and faster cpu? Ok, faster but you are going to get bottlenecks.
A new motherboard design with faster components wouldn't be required?
Ignorant's questions (really).

Nexxen wrote on 2021-11-22, 01:14:

Even if you have a 80mhz 286/386, bus limits all the extra speed, what's the point in having same tech and faster cpu? Ok, faster but you are going to get bottlenecks.
A new motherboard design with faster components wouldn't be required?
Ignorant's questions (really).

I used to have a bell curve showing the relationship between performance and clock speed assuming a fixed FSB speed.

Given large amounts of very fast cache is available cheaply it would make more sense to have a little cpu board with 0 wait state cache used as memory.

CPU Performance would then increase linearly but the other components would be limited by the ISA bus similarly to a high end ISA only 486

rmay635703 wrote on 2021-11-22, 02:03:

I used to have a bell curve showing the relationship between performance and clock speed assuming a fixed FSB speed. […]
Show full quote

Nexxen wrote on 2021-11-22, 01:14:

Even if you have a 80mhz 286/386, bus limits all the extra speed, what's the point in having same tech and faster cpu? Ok, faster but you are going to get bottlenecks.
A new motherboard design with faster components wouldn't be required?
Ignorant's questions (really).

I used to have a bell curve showing the relationship between performance and clock speed assuming a fixed FSB speed.

Given large amounts of very fast cache is available cheaply it would make more sense to have a little cpu board with 0 wait state cache used as memory.

CPU Performance would then increase linearly but the other components would be limited by the ISA bus similarly to a high end ISA only 486

Thanks for the explanation.

A better solution could be to design a new board from scratch, like a x86 with pci express 😀 No limits from the bus. Assuming it is possible.
1080 + 486 350mhz!

Making a multi socket board, 80xx, 186-486, could be done? Bus issues? Chipset? ...
Still more ignorant questioning.

I enjoy tracking down the crazy number of "packaged" upgrade processors for various platforms and have amassed a decent collection including the various Cyrix/TI options for 386 machines, Kingston/Evergreen and Intel upgrade chips for the 486, Intel Pentium / Pentium II overdrive processors for older socket 5 and 8 platforms, and of course the "slocket" and socket interposer upgrades for Slot1/Socket 370 machines. They tend to be pretty soundly outclassed by whatever succeeded them, usually at a lower cost to acquire, but do represent a glorious high watermark for performance at the twilight of a given platform's relevance.

I don't know if it would be worth the time/cost to develop a new solution of this sort for legacy platforms, consider an alternative like the EPIA C3 integrated mainboards that brought 486/586 architecture into the Pentium III age of computing, but remains relatively affordable and easy to tweak via software tools to a desired level of performance for maximum backward compatibility. You've also got the Media GX and Geode chips integrated into industrial boards that provide similar capabilities without the the impossibly high bar that exists in developing a custom multi-socket retro mainboard.

There's something on my "would be cool to do maybe" list that's in the parking lot if not the actual ballpark of this. i) Find a 386 clamshell that had a huge lead acid or nicad battery compartment, takes 4Mb of RAM and has a fast screen probably orange plasma ii) repower it with huge capacity, fraction of the size lithium cells, iii) use the saved space to mount some flavor of pi. iv) network it to the pi through best available means v) find an Xserver that runs on the 386 to be a client to X running on the pi. vi) play glquake etc in glorious orange monochrome.

Since prices have become arguably absurd for some vintage video cards, I would be more interested in new video cards for old systems...
New batches of 3DFX compatible cards, at nice prices, for instance. 😉

Not that I need newly designed CPUs, pushing farther than what was possible back then for a certain platform (those would also be welcomed), but compatible "clone" replacements for good old designs, which are either becoming scarse or too expensive.
Slot 1 CPUs higher than 600mhz, for instance, are quite expensive (especially 100mhz bus ones) and hard to come across today.
Same could be said about K6-2+/K6-3+.

I guess when FPGA chips become less expensive, compatible hardware for the cases above mentioned, could be one day achievable.
Sure, we are talking about a niche market though, who would go through the needed I+D and production expenses?

Con 2 botones wrote on 2021-11-22, 19:33:

Since prices have become arguably absurd for some vintage video cards, I would be more interested in new video cards for old sys […]
Show full quote

Since prices have become arguably absurd for some vintage video cards, I would be more interested in new video cards for old systems...
New batches of 3DFX compatible cards, at nice prices, for instance. 😉

Not that I need newly designed CPUs, pushing farther than what was possible back then for a certain platform (those would also be welcomed), but compatible "clone" replacements for good old designs, which are either becoming scarse or too expensive.
Slot 1 CPUs higher than 600mhz, for instance, are quite expensive (especially 100mhz bus ones) and hard to come across today.
Same could be said about K6-2+/K6-3+.

I guess when FPGA chips become less expensive, compatible hardware for the cases above mentioned, could be one day achievable.
Sure, we are talking about a niche market though, who would go through the needed I+D and production expenses?

Voodoo cards had many people in mind, but actually, you need working chips. They will be expensive even newly created. I mean in home conditions. That is manual work basically.
The only two cards are worth to remake ... voodoo4 and voodoo5. There you can be cheaper. Particulary voodoo5 with AGP 4x.

To the making it in factory, that would be unrealistic. You would need a market for that. It is still too small (100 or 200 people wanting to buy it, is still not enough to start mass production). Only by mass production in factory, could it be cheaper (under 100$), otherwise, it would be always far above 100$ per piece.
If you want to make for example 5000 cards (if you get 5000 working chips, maybe some company can make the deal in factory, and make cards... I saw in china new X58 socket 1366 motherboards made out of old chipsets).... but you need to convince some card manufacturer that it is profitable... which is not.
It's big risk to manufacture for example 5000 voodoo cards, when you find only 200 or 300 buyers. It would be big loss for them, and big risks. Also, they dont see a purpouse (you find only minimal people on higher places in companies that make cards, that have so big crush on old hardware, that they would help retro community and feel for them... just so they have better price for voodoo card. Nah, it's unrealistic. No company will start mass produce it and taking risk for nothing. )

W.x. wrote on 2021-11-22, 20:45:

Voodoo cards had many people in mind, but actually, you need working chips. They will be expensive even newly created. I mean in […]
Show full quote

Con 2 botones wrote on 2021-11-22, 19:33:

Since prices have become arguably absurd for some vintage video cards, I would be more interested in new video cards for old sys […]
Show full quote

Since prices have become arguably absurd for some vintage video cards, I would be more interested in new video cards for old systems...
New batches of 3DFX compatible cards, at nice prices, for instance. 😉

Not that I need newly designed CPUs, pushing farther than what was possible back then for a certain platform (those would also be welcomed), but compatible "clone" replacements for good old designs, which are either becoming scarse or too expensive.
Slot 1 CPUs higher than 600mhz, for instance, are quite expensive (especially 100mhz bus ones) and hard to come across today.
Same could be said about K6-2+/K6-3+.

I guess when FPGA chips become less expensive, compatible hardware for the cases above mentioned, could be one day achievable.
Sure, we are talking about a niche market though, who would go through the needed I+D and production expenses?

Voodoo cards had many people in mind, but actually, you need working chips. They will be expensive even newly created. I mean in home conditions. That is manual work basically.
The only two cards are worth to remake ... voodoo4 and voodoo5. There you can be cheaper. Particulary voodoo5 with AGP 4x.

To the making it in factory, that would be unrealistic. You would need a market for that. It is still too small (100 or 200 people wanting to buy it, is still not enough to start mass production). Only by mass production in factory, could it be cheaper (under 100$), otherwise, it would be always far above 100$ per piece.
If you want to make for example 5000 cards (if you get 5000 working chips, maybe some company can make the deal in factory, and make cards... I saw in china new X58 socket 1366 motherboards made out of old chipsets).... but you need to convince some card manufacturer that it is profitable... which is not.
It's big risk to manufacture for example 5000 voodoo cards, when you find only 200 or 300 buyers. It would be big loss for them, and big risks. Also, they dont see a purpouse (you find only minimal people on higher places in companies that make cards, that have so big crush on old hardware, that they would help retro community and feel for them... just so they have better price for voodoo card. Nah, it's unrealistic. No company will start mass produce it and taking risk for nothing. )

Oh, I know it is unrealistic. We were just "dreaming", like the OP.

There´s a man (in Ukraine, I believe), can´t recall his name right now, that builds new Voodoo 4 or 5 cards. Not cheap as expected, it is a craft work.

W.x. wrote on 2021-11-14, 15:06:

Sphere478 wrote on 2021-11-13, 20:14:

Pretty sweet! Hopefully soon we may have ones for later sockets

I don't think so. The one is pretty simple, they've just probably somehow die shrinked it, to get 100 mhz. But I don't know, didnt analyzed it, how they've improved it.
I doubt they ever get even to 286 level in home conditions, as I've said. I am sceptical to get even to 8086. But forget about socket 3 and socket 7. I mean 486 and Pentium. That's too complicated, some enthusiastic cannot do same work, as thousand of people, engineers, in laboratory and factory in early 90's.

65F02 is faster because it's in an FPGA. FPGAs are pretty quick, and it's especially helped since the 65F02 loads all the external ROMs into an internal 64 kilobyte memory when it starts, and then it uses the internal memory instead of whatever (48KB?) of RAM may be on board. The internal memory is very very fast and explains why it can run at 100MHz. It only goes slow when it has to access external I/O, and it even write-through caches video memory when used in an Apple II.

286 level is extremely achievable on an FPGA, something like 386 is also achievable. You'd have bus speed bottleneck problems to a degree since you couldn't pull off the same trick with only using on-FPGA memory, but even if you used the same FPGA chip as the 65F02, you could design in a ~32 kilobyte L1 cache, and if your system had a MDA/CGA/Tandy card (4KB/16KB/32KB Video RAM) you could do the same trick with write-through caching the entire video memory.

An FPGA 286 could readily hit at least 25MHz in cheap and readily available FPGAs (Spartan 6 is what I'm thinking about), I'd expect something like 40-50MHz real top speed. All this would be helped vs the original system's speed by whatever L1 caching could be done, and trying to fit a 287 in the FPGA could also improve speed a lot. If you wanted, you could clock-double (or triple, quadruple, quintuple) vs the external bus.

Basically it would be possible to construct a 286DX/DX2/3/4/5/6 on a Spartan-6 using similar methods to the 65F02. FPUs are big, so maybe it wouldn't fit a cheap FPGA, but on a 25MHz board you could at least have a 25MHz 286 with instructions that take fewer cycles than a standard 286, and a boatload of built-in cache.

As far as silicon fabrication rather than FPGAs, you won't be able to make a CPU more complex than a 4004 in your garage, even if you do have a lot of money to burn. You could, however, pay a fab service. There are a few that are intended for people to experiment with hobbyist chip fabrication. Google had a thing a bit back, where they'd let you submit your (open source) chip design and if you were one of the winners, they'd fabricate it on a 130nm process for you for free, aka roughly the same fab process that Tualatins were made with. I believe they're working on trying to open it up to everybody now, for affordable prices.

Error 0x7CF wrote on 2021-11-22, 23:07:

65F02 is faster because it's in an FPGA. FPGAs are pretty quick, and it's especially helped since the 65F02 loads all the extern […]
Show full quote

W.x. wrote on 2021-11-14, 15:06:

Sphere478 wrote on 2021-11-13, 20:14:

Pretty sweet! Hopefully soon we may have ones for later sockets

I don't think so. The one is pretty simple, they've just probably somehow die shrinked it, to get 100 mhz. But I don't know, didnt analyzed it, how they've improved it.
I doubt they ever get even to 286 level in home conditions, as I've said. I am sceptical to get even to 8086. But forget about socket 3 and socket 7. I mean 486 and Pentium. That's too complicated, some enthusiastic cannot do same work, as thousand of people, engineers, in laboratory and factory in early 90's.

65F02 is faster because it's in an FPGA. FPGAs are pretty quick, and it's especially helped since the 65F02 loads all the external ROMs into an internal 64 kilobyte memory when it starts, and then it uses the internal memory instead of whatever (48KB?) of RAM may be on board. The internal memory is very very fast and explains why it can run at 100MHz. It only goes slow when it has to access external I/O, and it even write-through caches video memory when used in an Apple II.

286 level is extremely achievable on an FPGA, something like 386 is also achievable. You'd have bus speed bottleneck problems to a degree since you couldn't pull off the same trick with only using on-FPGA memory, but even if you used the same FPGA chip as the 65F02, you could design in a ~32 kilobyte L1 cache, and if your system had a MDA/CGA/Tandy card (4KB/16KB/32KB Video RAM) you could do the same trick with write-through caching the entire video memory.

An FPGA 286 could readily hit at least 25MHz in cheap and readily available FPGAs (Spartan 6 is what I'm thinking about), I'd expect something like 40-50MHz real top speed. All this would be helped vs the original system's speed by whatever L1 caching could be done, and trying to fit a 287 in the FPGA could also improve speed a lot. If you wanted, you could clock-double (or triple, quadruple, quintuple) vs the external bus.

Basically it would be possible to construct a 286DX/DX2/3/4/5/6 on a Spartan-6 using similar methods to the 65F02. FPUs are big, so maybe it wouldn't fit a cheap FPGA, but on a 25MHz board you could at least have a 25MHz 286 with instructions that take fewer cycles than a standard 286, and a boatload of built-in cache.

As far as silicon fabrication rather than FPGAs, you won't be able to make a CPU more complex than a 4004 in your garage, even if you do have a lot of money to burn. You could, however, pay a fab service. There are a few that are intended for people to experiment with hobbyist chip fabrication. Google had a thing a bit back, where they'd let you submit your (open source) chip design and if you were one of the winners, they'd fabricate it on a 130nm process for you for free, aka roughly the same fab process that Tualatins were made with. I believe they're working on trying to open it up to everybody now, for affordable prices.

I was speaking only about making new chips. Not adaptors. I know, that through adaptors many tricks are possible, but you need higher chips for it. You won't make new chips. I agree, in home conditions you wont get past 4004. This photo tells it all (Number of employee in Intel after making 4004). You won't cheat it by 2-3 men in garage 😀

W.x. wrote on 2021-11-23, 00:04:

Error 0x7CF wrote on 2021-11-22, 23:07:

65F02 is faster because it's in an FPGA. FPGAs are pretty quick, and it's especially helped since the 65F02 loads all the extern […]
Show full quote

W.x. wrote on 2021-11-14, 15:06:

I don't think so. The one is pretty simple, they've just probably somehow die shrinked it, to get 100 mhz. But I don't know, didnt analyzed it, how they've improved it.
I doubt they ever get even to 286 level in home conditions, as I've said. I am sceptical to get even to 8086. But forget about socket 3 and socket 7. I mean 486 and Pentium. That's too complicated, some enthusiastic cannot do same work, as thousand of people, engineers, in laboratory and factory in early 90's.

65F02 is faster because it's in an FPGA. FPGAs are pretty quick, and it's especially helped since the 65F02 loads all the external ROMs into an internal 64 kilobyte memory when it starts, and then it uses the internal memory instead of whatever (48KB?) of RAM may be on board. The internal memory is very very fast and explains why it can run at 100MHz. It only goes slow when it has to access external I/O, and it even write-through caches video memory when used in an Apple II.

286 level is extremely achievable on an FPGA, something like 386 is also achievable. You'd have bus speed bottleneck problems to a degree since you couldn't pull off the same trick with only using on-FPGA memory, but even if you used the same FPGA chip as the 65F02, you could design in a ~32 kilobyte L1 cache, and if your system had a MDA/CGA/Tandy card (4KB/16KB/32KB Video RAM) you could do the same trick with write-through caching the entire video memory.

An FPGA 286 could readily hit at least 25MHz in cheap and readily available FPGAs (Spartan 6 is what I'm thinking about), I'd expect something like 40-50MHz real top speed. All this would be helped vs the original system's speed by whatever L1 caching could be done, and trying to fit a 287 in the FPGA could also improve speed a lot. If you wanted, you could clock-double (or triple, quadruple, quintuple) vs the external bus.

Basically it would be possible to construct a 286DX/DX2/3/4/5/6 on a Spartan-6 using similar methods to the 65F02. FPUs are big, so maybe it wouldn't fit a cheap FPGA, but on a 25MHz board you could at least have a 25MHz 286 with instructions that take fewer cycles than a standard 286, and a boatload of built-in cache.

As far as silicon fabrication rather than FPGAs, you won't be able to make a CPU more complex than a 4004 in your garage, even if you do have a lot of money to burn. You could, however, pay a fab service. There are a few that are intended for people to experiment with hobbyist chip fabrication. Google had a thing a bit back, where they'd let you submit your (open source) chip design and if you were one of the winners, they'd fabricate it on a 130nm process for you for free, aka roughly the same fab process that Tualatins were made with. I believe they're working on trying to open it up to everybody now, for affordable prices.

I was speaking only about making new chips. Not adaptors. I know, that through adaptors many tricks are possible, but you need higher chips for it. You won't make new chips. I agree, in home conditions you wont get past 4004. This photo tells it all (Number of employee in Intel after making 4004). You won't cheat it by 2-3 men in garage 😀

But that would have all been done by hand for the most part... now with way more advanced tools and faster computers, things more complex can be made much easier.

There are design tools, etc that can help out a ton. Plus if making something x86, because the instructions are already known it is feasible that somebody could design one from the ground up.

Wow, nice discussion.

I was thinking, isn't the increase in mhz going to yield a linear increase in performance?
186,286,386? Some FPGA could come with L1 + L2 cache, are the new performance results predictable?
486 with higher frequency could get larger L1 and L2.

Isn't a BIOS necessary for these new cpus? I can't get around the idea of some support updates.
Or being internally clocked x2/3/4... doesn't need?

Nexxen wrote on 2021-11-23, 10:51:

I was thinking, isn't the increase in mhz going to yield a linear increase in performance?

Not quite. On old systems like 086/088, 186, 286, 386, (and 486 before clock multipliers), the bus clock speed and CPU speed are the same. So, for them increasing the CPU speed does yield a pretty much linear increase in performance, as long as you ignore any performance limited by ISA, which can only really clock up to 8-10MHz safely unless you're picking your ISA cards carefully. In simpler systems (some 8088, some 8086, some 286) where the CPU, Bus, and ISA are all run at the same frequency, a frequency increase is a completely linear performance improvement since it increases the frequency of everything in the system.

You won't reliably be able to exceed whatever Bus speed your board has for its maximum setting. So, after a certain point, you want to increase CPU performance without increasing bus speed.

One way to do this is to make your instructions take fewer cycles to complete. This technique is referred to as "Increasing IPC", increasing Instructions Per Cycle. Many x86 instructions take several clock-cycles to complete. In real CPUs, this was done with the NEC V20/V30, and is why those are known to be faster than their 8088/8086 counterparts at the same clock speed. IPC is improved with pretty much every new CPU generation (except the Pentium 4), and is why a 8MHz 286 system can outperform a 10MHz 8086 system.

Another option is to introduce/increase cache, and reduce dependence on the slow main memory. Even when the Bus and CPU ran at the same speed, DRAM is slow, so going out to memory to fetch a value is going to slow execution down. If you have a cache onboard or on-chip (and the value you want is inside the cache) you can read from the much faster SRAM cache memory and not have the CPU waiting for your slow main memory to respond.

Another option is to increase CPU clock speed without increasing the Bus speed. In real CPUs this happened in the 486 era when Intel started clock-doubling. This especially works well if you have a cache on-chip which can run at your doubled/tripled/etc clock speed, and prevent slow accesses to your now half/one-third/etc-speed Bus. Full-speed on-chip cache in combination with clock-multiplication is what makes the 65F02 fast. It has enough RAM inside it to almost never have to touch main memory, and all the ram inside it runs at 100MHz, vs the previously incredibly slow 1MHz system ram.

Nexxen wrote on 2021-11-23, 10:51:

186,286,386? Some FPGA could come with L1 + L2 cache, are the new performance results predictable?

The new results could be simulated and compared to a simulated version of the old results, but I'm not remotely smart enough to be able to do that. A cycle-accurate emulator would need to be implemented to be able to check, which is pretty much the same amount of work as implementing the device on an FPGA to begin with.

Nexxen wrote on 2021-11-23, 10:51:

Isn't a BIOS necessary for these new cpus? I can't get around the idea of some support updates.
Or being internally clocked x2/3/4... doesn't need?

Probably wouldn't need a BIOS update as long as it behaves mostly like the original chip. NEC V20/V30 don't need BIOS updates. 486DX2 in a system with a bios that only knows about 486DX doesn't need a bios update to clock-double.

cyclone3d wrote on 2021-11-23, 00:40:

But that would have all been done by hand for the most part... now with way more advanced tools and faster computers, things more complex can be made much easier.

There are design tools, etc that can help out a ton. Plus if making something x86, because the instructions are already known it is feasible that somebody could design one from the ground up.

Again, you can design more complex CPU, or even new platform, but you won't make chip in home conditions, that is more advanced, than original one. Your chips would be larger, and thus, generate more heat, etc... How do you want to make more complex chip than 4004 in home condition... with limited budget. Without building a factory, or buying incredible expensive machines.... I've explained it once, it would be profitable only if you would sell thousands of them. For larger factory and machines, you need tons of water and energy. Not only buy machines. You need also personal, that know work with it, you need to buy waffers, pay for water, energy. It would go to millions. Just for making 10 or 20 or maybe 100 chips for 8086, 286 or 386? I just cannot imagine it in reality. Unless you buy machine, that can cost up to 10000$ or 20000$ (for you own money ... but I would guess no such cheap machine exists), and that all with energy, water and cost of waffers, you cannot make new chips, your initial investition needs to be millions to build factory that can produce at least one chip... but then, when you start that power and resource hungry factory, you just need to go and go with new chips, otherwise, it wont pay investition. So to make new 286 or 386 chip, it would possible only if equipment costs several thousands dollars. In reality, noone will pay so much money for enhanced 286 or something, you will need to mass produce it.