New CPU Accelerator options for 68K Macs?

[JDW]

If you want a faster 68k Mac you can just use a PPC mac.

I've never felt compelled to do that.  The only PPC machines I currently own are my G4 Cubes and PowerBook Wallstreets, mainly because if I want a fast Mac that runs modern software, I will use a modern Intel Mac.  

My primary love is for older 68k Macs, especially the 9" CRT Macs like the SE/30.  The SE/30 runs reasonable well at its stock speed, but I must say I enjoy it more with a 50MHz 68030 accelerator installed.  And its that upgradability that makes the SE/30 king of the compacts!  My Daystar Turbo 040 40MHz PDS accelerator makes it faster still, but there are software compatibility issues even with the cache disabled.  As a result, I tend to use my 030 accelerators more than the 040.  The 030 boards are just like the stock SE/30 only faster. So if there was a new-fangled and modern accelerator for an SE/30, I would ask about the software compatibility.  If compatibility with legacy software was worse than an 040 card, it wouldn't be worth $500 in my eyes.  

This is why I've been most impressed by the work of @Bolle.  He has created a TS Adapter with built-in MacCon Ethernet and he has cloned the socketed PowerCache!  Sure, those two upgrades are specific to the SE/30, and current pricing of the socketed PowerCache clone is high, but the point here is that we retain software compatibility while gaining the ability to free up the PDS card slot for other upgrades.  And all that goodness fits inside the stock case, which I think is important.  

I will admit though that the "concept" of having a 68080 chip in a compact Mac is drool-worthy to be sure.

 

[CC_333]

68080?  Does such a chip even exist??!!

I thought Motorola cut it off at '060?

c

 

[JDW]

68080?  Does such a chip even exist??!!

Details are found on the 68080 wiki here.

 

[uyjulian]

I'm not entirely sure, but I *think* that's basically an entire Amiga on a Raspberry PI. It has an image of the OS and it's also generating its own video. This is the first I've heard about it, so I could easily be wrong though. 

I'm not saying I would say no to a similar mac project, if it fit into one of my compacts, but I don't think that's a simple 68000 replacement. 

Looks like only emulation of fastram, IDE, mmu, fpu, and CPU. Rest is real hardware.


I think the idea of PiStorm is nice, as you could have similar possibilities as RaSCSI but with more bandwidth. It looks like A314 (trapdoor expansion for Raspberry Pi) services can be integrated on the PiStorm.

 

[johnklos]

I'm not entirely sure, but I *think* that's basically an entire Amiga on a Raspberry PI. It has an image of the OS and it's also generating its own video. This is the first I've heard about it, so I could easily be wrong though. 

Nope. All the original hardware is still available and used. The CPU is emulated and other hardware is made available by virtue of direct attachment to the CPU.

 

[Frobozz]

Well, neat! 

Not that I want to manage a rasberry pi inside a compact mac case, but gotta love it. 

 

[NJRoadfan]

My biggest gripe with the Vampires is that they are only for wedge Amigas. The 2000/3000/4000 get no love. As for Macs...... the reason was already mentioned, far more powerful machines were released by Apple and the platform got more support for peripherals over time as well. Heck, Amiga folks were stuffing Sonnet Quad Doublers into their Amiga 4000s since they were cheap and readily available compared to the Amiga specific cards.

 

[Frobozz]

 

Bigbox Vampire Amiga FAQ 




What type of a bigbox Amiga is supported?



The Amiga 1000, Amiga 1500 and Amiga 2000 is supported. Amiga 3000, Amiga 3500T, Amiga 4000 and Amiga 4000T is at the moment not supported. Also Amiga CDTV is not supported but some have getting it to work.

You can use the Vampire V500 V2+ with Amiga 1000, Amiga 1500 and Amiga 2000. You can connect it to the 68000 CPU socket or get a riser card. 

There are also solutions made for connecting the Vampire V500 V2+ accelerator easier inside Amiga 1000 or Amiga 2000 if you want to use the CPU socket.

----

https://amitopia.com/the-big-box-amiga-vampire-68080-accelerator-guide/

If it makes you feel better, I'll trade you your A3000 or A4000 for an A2000 with a vampire  

 

[johnklos]

I'm not a fan of the Vampire.

First, they're trying too hard to make their own idea of an incompatible Motorola CPU successor. What value is there in making a CPU replacement which supports old software while at the same time trying to encourage programmers making new software which uses CPU additions which mean the new software won't run on real Motorola CPUs?

Second, they don't care very much about compatibility. FPU support was a very low priority, and it appears that they're not interested in MMU support. This means that virtual memory is not possible, nor is running NetBSD, GNU/Linux, nor A/UX. This would also mean it'd be more difficult to add memory pools to Mac OS because Mac OS usually requires sequential memory.

 

[Franklinstein]

For me it's a question of value, mostly: do I need to spend $500 on top of whatever else has been sunk into the device (assuming prior max RAM upgrades and SCSI2SD or whatever) just to... make Asteroids run so quickly that I basically die as soon as the screen loads? To be able to scroll through a Lotus 123 spreadsheet so fast the screen is just a blur? To (attempt to) render FaceBook or other modern website on a 512x384 b&w screen (or external high-res screen, for some reason)? I mean, what's the utility here? Why would I need a quantum leap in performance and tons of new features for my SE when I can just use a faster Mac that already does that stuff? 

If I was going to buy an accelerator all I would want is a choice of proper 68k-based cards for various Macs: clone some Daystar or MicroMac or Interware cards and re-release them, possibly with enhancements (maybe an '060 version? Some Interware cards have pads suggesting this was in consideration but may have been scrapped due to incompatibilities between the '040 and '060's respective ISAs). I'd definitely like a QuadDoubler with a L2 cache built-in. And really, once you start getting into an FPGA with expanded RAM, video, Ethernet, etc built-in that basically just uses the Mac as a host for power and peripherals, it kind of ceases to be an "accelerator" and more a "new computer wearing a retro skin suit" and the modding community already has this covered with their various iPad or Mac mini-based hacks. Plus I'd be more inclined to buy several different 68k accelerators for several different machines rather than just one bonkers FPGA accelerator as a gee-whiz sort of thing (if there were multiple versions available I assume it would be the same FPGA base on a different machine-specific form factor so there would be little point in buying more than one).

 

[Byte Knight]

So... two questions:

1) What's the level of interest in something like this? Would you plunk down $500 for an accelerator/RAM upgrade? 

$200-$300 - yes

$500 - hard to justify spending that much...

 

[Kai Robinson]

So circle back to the idea - for 68000 based macs, to go to an 020 or 030 - much simpler, lower cost of entry, easier to adapt. 

 

[CharlieFrown]

hey guys, any specific reason to use 030 in compact Macs still limited to 1bit video? Can you name any software other than flightsims that would utilize more power or demand better CPU than 68000? 

TBH  I always thought it would be nice to use MC68HC000 based turbo which runs as fast as 50 mhz https://retro.7-bit.pl/?lang=en&go=projekty&name=WICH508I anyway stable @ 33 MHz  https://sordan.ie/product/583/turbo-card-hc533-33mhz-8mb-ram-ide-amiga-500/

It would make the system and apps snappier. 

 

[Frobozz]

For why more speed on B/W Macs, I would say: 

• development (THINK C, etc.)
  
• general system snappiness. Try using miniVmac at like 4x speed, then a mac 128/512/plus/classic for general tasks. You'll notice a big usability difference.
  
• (not speed, but memory): having more than 4MB would be nice. Not 64 GB, but 8 or 16 would be quite nice. 
  

For 68000 overclocked vs 68030, I would maybe suggest (not an expert here):

• Apparently a bit more stable (at least compared to wicher). I haven't experienced any stability on my 50mhz 68030. 
  
• having FPU is nice for lots of things, including, apparently toolbox routines that use it for drawing/etc.
  
• Not sure why not do it, cost is comparable? Doesn't seem to be a lack of 68030s?
  
• Speed. Both those shown above are running a lot slower than the 68030TK and TF534 (on same Amiga hardware).
  

A couple of projects I've seen:

• matze's 68030TK (open source)
  
  
  "68030 Turbocard for 68000-DIP processors"
  
• https://gitlab.com/MHeinrichs/68030tk
  
• (note that they solved the 64MB RAM issue mentioned in the readme; the new ones being built have option of 4MB or 64 MB)
  

[*]TerribleFire TF534

• 
  
  https://linuxjedi.co.uk/2020/10/31/repairing-a-terrible-fire-534/ (see snapshot of benchmark at bottom of page)
  
• https://www.8bits4ever.net/product-page/tf534
  
• Apparently the TF developer has left the community and deleted the project though, so this line of products is probably stuck forever where it is, with no new support coming. 
  
• The TF series could apparently be adapted to work with the Atari ST computers, but I don't know the details. 
  

 

[Dog Cow]

hey guys, any specific reason to use 030 in compact Macs still limited to 1bit video? Can you name any software other than flightsims that would utilize more power or demand better CPU than 68000? 

My Sabina TCP/IP stack has better ping latency and TCP ACK latency on the SE/30 than any 68000 Mac.

 

[cheesestraws]

My Sabina TCP/IP stack has better ping latency and TCP ACK latency on the SE/30 than any 68000 Mac.

Ah, a critical metric.

For why more speed on B/W Macs, I would say: 

I think one of the reasons why this hasn't happened is a cultural difference between classic Mac and, say, Amiga hobbyists.  I am, socially speaking, on the edge of the hobby, so perhaps my sample is off here, but classic Mac people mostly seem to be more interested in what could have been done at the time rather than attempting to accelerate the machine past all recognition.  So I suspect that an all-new super-fast accelerator might, once the novelty has worn off, meet with less long-lasting success than, for example, @Bolle's recreations of historical accelerators.

This is partly, I think, because the classic Mac didn't really "die" in the same way the Amiga did.  If one wants to run classic MacOS on a 1.5GHz G4 processor one can (just), without needing any hardware modifications at all.  There are already loads of ways to run classic MacOS fast, and when one does, the first thing that becomes obvious is just how creaky it is when you attempt to do so.

 

[Kai Robinson]

For why more speed on B/W Macs, I would say: 

• development (THINK C, etc.)
  
• general system snappiness. Try using miniVmac at like 4x speed, then a mac 128/512/plus/classic for general tasks. You'll notice a big usability difference.
  
• (not speed, but memory): having more than 4MB would be nice. Not 64 GB, but 8 or 16 would be quite nice. 
  

For 68000 overclocked vs 68030, I would maybe suggest (not an expert here):

• Apparently a bit more stable (at least compared to wicher). I haven't experienced any stability on my 50mhz 68030. 
  
• having FPU is nice for lots of things, including, apparently toolbox routines that use it for drawing/etc.
  
• Not sure why not do it, cost is comparable? Doesn't seem to be a lack of 68030s?
  
• Speed. Both those shown above are running a lot slower than the 68030TK and TF534 (on same Amiga hardware).
  

A couple of projects I've seen:

• matze's 68030TK (open source)
  
  
  "68030 Turbocard for 68000-DIP processors"
  
• https://gitlab.com/MHeinrichs/68030tk
  
• (note that they solved the 64MB RAM issue mentioned in the readme; the new ones being built have option of 4MB or 64 MB)
  

[*]TerribleFire TF534

• 
  
  https://linuxjedi.co.uk/2020/10/31/repairing-a-terrible-fire-534/ (see snapshot of benchmark at bottom of page)
  
• https://www.8bits4ever.net/product-page/tf534
  
• Apparently the TF developer has left the community and deleted the project though, so this line of products is probably stuck forever where it is, with no new support coming. 
  
• The TF series could apparently be adapted to work with the Atari ST computers, but I don't know the details. 
  

You are correct in that Stephen Leary (Terriblefire) has removed his entire github repo for the TF Projects, due to some members of the community trying to profit off his work, stealing his unreleased prototypes and arguing that it was fair game to do so - so in turn he washed his hands of it all and walked away. Fair play to the man. However, I did personally contact him about this as i already had backed up all of his GitHub repo and Youtube channel - he stated that if i have the sources, anyone is welcome to them, to use them, modify them etc - just to retain original credits as per GPL v2. 

Rather than jump straight in with the TF534 or TF530's, however, i'd suggest the TF520 be the first port of call - it's the simplest of them all, with just a single XC9636XL CPLD involved to handle bus translation. There's no MMU, there's no FPU - just straight 68020 in the 68000 socket. This would be an excellent STARTING POINT. You can then go on to the TF530's - personally i'd suggest removing IDE and the SRAM, and adopt an SDRAM controller so you can use more modern sources of memory - 68030 with SDRAM, it's not as radical as you might think. 

These 'accelerators' were designed more as a personal project, but also can be used on Atari ST's, not just the Commodore AMIGA's If you want to start with one - start there. 

One question is - does anything on the Mac use the 6800 peripheral interface pins on the 68000? Does it need the 'E' clock for anything? If you can eliminate those - then moving to an 020 becomes easier. 

 

[techknight]

One question is - does anything on the Mac use the 6800 peripheral interface pins on the 68000? Does it need the 'E' clock for anything? If you can eliminate those - then moving to an 020 becomes easier. 

Yes. the E clock is required. the E/VMA/VPA is required on Macs, because of the VIAs. So this logic has to be implemented in the accelerator, Something I was trying to research back in 2014 when I was motivated to do such project for the macintosh portable. 

The portable is kinda weird. They kinda use E/VPA/VMA but they kinda dont. Strange, they were in the process of handing that logic over to GLU by the time that design came out. 

 

[moldy]

I'm new to the forum so first of all I would like to say hi  

I've recently start digging a bit into the accelerator topics. I found quite some references for 68000->68030 accelerators within the Amiga community - apart from TerribleFire's and Matze's designs that @Frobozz mentioned, you might find the original LUCAS article useful:

http://kai.supramania.com/Electronics-Projects/AMIGA/LUCAS.pdf

http://amiga.resource.cx/exp/lucas

So far, it gave me the best explanation what is needed to run the CPU asynchronously w.r.t. to the bus clock of the rest of the system.

For those looking into accelerators that would replace a slow 68030 with a higher clocked one, I found this project:

https://github.com/Sakura-IT/decelerator4030

I even asked last weekend, but it stopped before running asynchronously:

https://github.com/Sakura-IT/decelerator4030/issues/4

It still seems to me like a good starting point that I'll consider looking into.

@CharlieFrown , there's a big thread about Wicher 508 in the Amiga forums, let me know if you would need some translation, I'll be happy to help!

 

[Kai Robinson]

Here's one to start with - runs a 68000 at 16MHz, includes 16K of Cache and a 68881 FPU...

https://drive.google.com/file/d/1Op7TzK1pxRjqcT68MoZE1G4HNkS5GJNe/view?usp=sharing

All the GAL equations and schematics included.