Apple Cancelled the Turbo Mac. Let's Build One.
HTML-код
- Опубликовано: 19 авг 2024
- 💾 Check out the Action Patreon! / actionretro
In 1985, Apple was rumored to be working on a Turbo Mac, an upgraded original Mac with a 68010 instead of a 68000, and up to 2 megs of RAM instead of 128K or 512K.
Well, that Mac never came out, and instead we got the Mac Plus with the same ol' 68000.
However, the 68010 is pin compatible with the 68000 in some packages... and I found one! Combined with the 2 meg RAM upgrade we found in our Mac 128K, we should be able to upgrade it to match the specs of that rumored Turbo Mac.
Or we'll completely ruin it. Who knows! Come find out!
🍎 ROMinator for 128, 512 and Plus: en.infinitypro...
🍎 68kMLA Thread: 68kmla.org/bb/...
🍎 JDW Builds a ROMinator: • ROM-inator Kit Build &...
══════════════════════════
💾 For more vintage Apple stuff, please subscribe: www.youtube.co...
💾 Support these retro computing shenanigans on Patreon! / actionretro
══════════════════════════
Check out my Amazon page with links to my tools, adapters, soldering equipment, camera gear and more: www.amazon.com...
══════════════════════════
💬 Come talk about old computers on the BitBang Social Mastodon! bitbang.social
══════════════════════════
#Apple #Macintosh #128K
Imagine a Mac with non-upgradeable RAM and inadequate storage. Thankfully Apple would never make that mistake again.
ha
LOL
yeh... about that...
heh. like so many things, they were just ahead of their time. I am honestly sometimes surprised that most PCs still have upgradable RAM, esp since all the other memory isn't.
@@neeneko is literally the only way You can have a PC, with upgradable ram
Is ilegal not to in México
That's why macs are sold as "entertainment electronic devices" or "profesional audio equipment" but you will never see an Apple ad talking about home computers
The Turbo Mac morphed into the Mac Plus.
The instruction cache only gave marginal improvements to most code.
The compelling thing about the 68010 was that it fixed the absolute internal state train-wreck that happened when a page fault occurred. It was a bug that was discovered late in the 68000's development cycle, and was punted; which was a shame because Motorola had already designed two different MMUs that were intended to go along with it (the 68451 segmented mmu, and the 68851 paged mmu).... This only mattered if you were intending to implement virtual memory and memory protection, which were considered important for running operating systems like UNIX, so Apple just quietly punted, and decided to tackle this with the Macintosh II.
As a weird digression, even though the 68000 couldn't properly handle page faults itself, workstation vendors built all sorts of hardware work-arounds to this problem. Sun and Apple implemented custom MMUs, the latter being used in the Lisa, and was one of the reasons that the Lisa ran at 5MHz. The most over-engineered solution though, came from Apollo Computer, where their first Domain workstations contained a second slave 68000 which ran one instruction behind the master CPU, when a page fault occurred, the hardware would halt the master, copy the processor state from the slave, and resume execution. It had no other function. ;)
Hahaha HAHAHA!!!
The instruction cache was only used in loops. Which was not something found in most software except for micro-controller functions. It was otherwise not really worth the effort to design for it given how late it came to market and how much more expensive it actually was.
@@wishusknight3009 With most stuff happening on Mac being done in software I'd say that there were numerous loops in the 'driver' code of the system.
Didn't read the sources, but it seems sensible.
Problem is 68010 has 6 byte instruction cache - not much, thats like 2 instructions only.
Morbed.
If you want to test the rominator with the original CPU, just stick the 68k into a machine pin socket and then put that into the original socket.
I love how the 68000 revolutionized computing in the 80s... I've seen a similar mod on a Sega MegaDrive/Genesis, where they swapped the original 7.6 MHz 68000 with a 10 MHz 68010, and unsurprisingly almost the entire game library work as good or even better (less hickups sometimes) than the original.
Does it fix the lag in Sonic games when Sonic gets hurt and loses his rings?
The megadrive mod was the first thing I thought about when he started talking about swapping the 68000 for the 68010.
@@commentarysheep of course it fixes it... But if you don't have access to an 68010, yo can get almost the same results (or even better ones) swapping the CPU crystal clock and bump the original CPU around 10 MHz without effort. If you push it higher it becomes incredibly unstable (12-14 MHz). I did it a long time ago and it was amazing to play Sonic (1,2 &3) or Thunderfoece IV without any lag.
I also modded that MegaDrive to work in NTSC mode, and it was an amazing experience.
Any compatibility breaks?
@@solarstrike33 from what I can remember, there where a few titles that had graphical or sync glitches. I'll search the original video and share the link here 🤗
Technically i didn’t click on your video. RUclips just knows what videos it should automatically play next. RUclips knows I don’t own a Mac but still find them interesting.
Sean, Cmd+Shift+3 will give you a screenshot PICT file that you can save onto another disc, instead of having to hold a smartphone's camera up to the screen.
This rumored Turbo Mac, was almost certainly a Mac Plus, just not a really precise leak. Both RAM upgrades (1MB standard) and new FDD matches. Maybe Apple indeed considered 68010 but scrapped it as it indeed didn't provide a performance boost.
to use the rom board, add a stack of pin headers atop the CPU socket, it will raise the whole expansion board up and allow you to clear the rom board
Dang, I love this channel. I'm mad this channel wasn't made 10 years ago though.
Or even earlier it could have stopped the Macquarium from becoming a thing in the early 2000s
Remove the rom sockets from the romulan board. Solder the roms directly on. Then do the same on the motherboard, remove the sockets on the mainboard. Solder the romulan directly to the MB. It’ll fit.
Love the "Callahan Auto" shirt. Tommy Boy was such a great movie!
Did you check if the ROMs had the right pinout to jam into the two free ROM sockets on the memory expansion card?
The wonders off the internet... someone before you have had the same exact issue and has created a form thread... that has remained unanswered.
and is still, to this day, 1 year and 2+ weeks later... >_
6:20: That's a decoupling cap, hence the weird beveled end. It was popular to slip them in under sockets like that, since it avoids traces on the PCB and gets the capacitance right where needs to be.
You can get special thin pin headers that fit in IC sockets, they will help if you are going to build the ROM relocator board.
So far, so good. And ending with a cliffhanger?!?! For shame! LOL! \m/
😂
With this success, I've ordered a 68010 for my Macintosh Classic and hope to replicate your success!
I did that same experiment back 1992 with a Mac 512KE with a 4mb Dove Board and System 6.0.4, System 6.0.7 and System 7. The Calculator DA would crash on the Mac. The Calculator on System 4 with the old ROMs is not the same as it is on System 6 with the newer Enhanced ROMs. Also ComicWorks (Art program) also crashed. Then again ComicWorks crashed on System7.
Thee 68010 only had a 3Byte cache which sped up tiny loops, nothing more. It did not improve much as promised on the 68k Purple Manual. Similarly on the Atari ST and Commodore Amiga the same results of crashes but with different programs.
Hey buddy the extender cables might work if you twist them around like a braid. It will reduce the cross talk on the wires
The motherboard for the Mac was rather sketchy. Somewhere I read that Steve Jobs wanted the RAM chips to be spaced farther apart as to line up with all the other chips. Didn’t work, they needed to be crammed together to keep the signals sharp and clean enough to work.
5:32 - Some of the later accelerators and addon boards actually won't slide out, they're too tall. A safe way to get the board out is to gently pry outward the left or right rail and just extract the board that way.
the 68020 accelerator board that was bodged into my 512K is like that, it's annoying.
This in one of the MacEffects Cases wouldn’t just be so perfect, ultimate power and ultimate looks!
Esoteric upgrades ftw!
A Mac with double set of brows thicker than @ElectroBOOM ones is a rare find indeed.
I think the major advantage of the 010 was that the 68000 had a page fault bug that prevented it from working properly with a PMMU. IIRC the 010 fixed that, but most of the 68k machines of that era didn't need > 8mb ram so nobody bothered with the 010. Then the 020 came out and obsoleted the 010.
I think some of the early Sun workstations used an 010 with a PMMU (before that they used a *pair* of 68000s running the same code a clock cycle behind one another so that if one of them page faulted it could fail over to the other CPU and avoid the bug; what a hack!!! -- edit: I was incorrect, it was the Apollo DN100 that did this). There were surely others, but I can't think of any.
I thought that the 010 had something like a primitive instruction cache. {Pause to look it up...}
Yes. Wikipedia says the 010 adds "Loop Mode": "...which accelerates loops consisting of only two instructions, such as a MOVE and a DBRA. The two-instruction mini-loop opcodes are prefetched and held in the 6-byte instruction cache while subsequent memory read/write cycles are only needed for the data operands for the duration of the loop. It provided for performance improvements averaging 50%, as a result of the elimination of instruction opcodes fetching during the loop." "In practice, the overall speed gain over a 68000 at the same frequency is less than 10%."
The benchmark program must not have used any 2-instruction loops that would benefit from the speed-up, but in a broader selection of software, there might well be some.
very interesting, i hadnt considered that the benchmark might not hit the specific differences in the 010
Is that the reason why I keep seeing sun machines as dual cpus? To workaround a hardware bug??
@@markinius8866 It was the really early ones. Like the sun2s I think. But I may be full of poo; I can't remember for sure.
@@Lee_Adamson_OCF I always thought pre-sparc sun machines were multiprocessor aware haha
The 'sieve' (rhymes with Steve) test is the Sieve of Eratosthenes algorithm, which is a way of finding prime numbers
I for one thoroughly enjoy your work.
Sweet guitar licks brother!!! Needs more cow bell though :)
Why is Void Linux and PowerPC in the hashtags?
oops lol
I wish the Mac Plus had 4bit grayscale.
You could put a new 40 pin socket that has square legs into the stretched holes to mount the old CPU.
Also might the possible to find an IDC cable that extends the ROM chips.
Then it will probably be too tall to fit inside the metal frame.
Edit: if you also wanted to use the RAM expansion board that is. But it could lift the expansion above the rominator.
Hey that's Hertzfeld on the magazine cover.
I realise what's going on with those black stickers on the mac around the screen. Probably velcro for a privacy filter.
I clicked the video and now we're in it together?!
Aw crap! 🤣
“You clicked on this video, we’re in this together…”
Well played, sir, well played. BTW, you had me at Turbo Mac…
ah fraq - a cliff hanger ending :-(
one thing about the 68010 is that it had a system mode / user mode transition instruction that worked, so all the dudes that wanted to bring up a Unix where the kernel executed in a system priviledge mode and user land programs in less priviledged user mode, would use the 68010, which there were a few such PCs or workstations that sported the 68010
So it's been nine months. Have you ordered the parts yet? Tell me you atleast have the parts?
“TurboMac. Another attempt to design a next-generation Macintosh, called TurboMac, was led by Burrell Smith, who designed the circuitboard on the first Macintosh. Working in a corner of Bandley 2, across the street from Macintosh headquarters, Smith took over a series of cubicles that he called TurboTown, where he developed an all-in-one concept that was roughly twice the size of the first Macintosh, with a faster microprocessor, two Custom chips designed by Smith that would do the work of a dozen off-the-shelf components, a 12-inch gray-scale display, a built-in hard disk and a Snow White case designed by Stephen Peart. Without a full-page display, however, it would not be the product that Apple needed. Some urged Jobs to convert TurboMac into a modular CPU with a separate 15-inch display and formed a committee, packed of with former Lisa engineers, to advise. Yet Burrell Smith, the mayor of TurboTown, vetoed that plan. "Since everyone on a committee wants to feel that they've made a contribution, committees will always undermine the best parts of an idea with an endless series of compromises," says Stephen Peart. As a result, TurboMac was cancelled, which prompted Burrell Smith to resign from Apple, declaring that he would never again drive his car through the inter-section of DeAnza Boulevard and Stevens Creek Road, which cuts through the heart of Apple's campus.” - AppleDesign by Paul Kunkel
I'm glad you still always write thonk
2018, I thought it was gonna be 2009 or something. 2018 was a like a week ago.
It's like using a 68010 on my Amiga... Feels pointless but helped with whdload
Reminds me of when I upgraded my Core Duo iMac to a Core 2 Duo, not a huge upgrade on paper, but it made a big difference.
FWVLIW: Stacking one or more turn-pin sockets between the RAM board and the motherboard 68K socket to give the ROMinator clearance will do for testing. Will that fit back in the MAC? Maybe not, but you never know! Some stabilising standoffs for the RAM board would be a good idea.
If that's not viable for reassembly it may be possible to mount the ROMinator under the motherboard... being VERY careful not to mirror the lines, assuming there is sufficient clearance underneath. In the long run and remote adaptor would seem least janky.
"Buggy features"
Thats the neat part! It's a feature!
A buggy feature!
Looks like there are two EPROM sockets, second high and low on the upgrade card, what where they intended for? I "assume" IF they would accept the OS eproms someone would have tried that already so if there not for that then what for?
Also if you want to raise the upgrade card more people often use stacked sockets, probably will interfere with sliding it in to the case tough.
Also the guy who designed the red eeprom expander could maybe help you layout it in a better way to fit the upgrade card if you asked.
Anyways grate video and Im sure you solve it, problems are there to be solved! XD
Came to say this! :)
So did it work?
I have a 16MHz 68000...wonder how something like that will work.
A company called Brainstorm sold a 16 MHz 68000 upgrade for early Macs. It worked fine. Not sure if a plain 16 MHz 68000 outside of their upgrade would or not.
Every one of your videos makes me want to start pilfering every garage sale and library/school sell off to find old Macs (and PCs) and dumping them in half my garage and explaining to my wife how I’m preserving tech history. 😂
At first she will fall for it, but as time passes and the tec is accumulating space and dust, she will start to ask hard questions.😅
Hey! What's Mac84 doing in a suit on the front cover of that InfoWorld issue? Somebody's been busy in PhotoShop. 🤣🤣
I just came here to check the specs on the end line for the rotary girder...
Admit it, you've been broken by unreleased Macs. I love this.
Really cool and fun video!! Thanks for sharing and looking forward to your next video
Depending on how tight the clearance issue is you might be able to get around it by just soldering the EPROMs directly onto the Romulator board instead of having them socketed. Of course that would mean you'd have to pick one set of ROMs to stick with, or order a second Romulator and swap the whole board out to switch between the two ROM sets.
Where is the next part?
Did you ever finish that project?
Must admit I've watched a few of your vids but never been a sub, I think this sort of hackery deserves it!
I guess those ROMs don't fit into the expansion board ROM slots?
Exactly what I was thinking too.
Really enjoyable video. I really do love old Macs and modding retro computers is always great fun. 👍
haha thanks!
👍 for more guitar in future episodes 🤘🤘
i don't see another video. did you ever get this together and working?
If you've got flat pins on your chip, can you not twist each to give them some depth?
Never was a mac guy but this is pretty awesome.
thanks!
I'm just curious, is it even using the stock PROM chips with that expansion installed? Kinda looks like it could be intercepting the PROM since it has two chips on the expansion card. I have to wonder if you would have to install the ROMulator on top of the RAM upgrade card. That also leaves the possibility that the monster Mac is using a modded prom to support the extended memory, which means whatever custom prom you use might need patched. I would experiment to see if the machine boots with the monster Mac, but with the stock rom chips removed.
11:52 What about the four identical ROM (?) sockets on the upgrade card? 🤔
Maybe you could use some (trimmed) wire wrap female pin headers as risers to move the ram card up enough to clear the rom card?
I would have just made a socket pin riser for the expansion just to test. This method requires no custom PCB, just some pins and a socket.
Best to disconnect the power cable to the mother board BEFORE discharging the CRT.
New #1 hit! Kay Koba song!
😂😂😂
What are the PROM2 Hi/Lo sockets on the monster mac board for?
Could they be mapped to... the same spot in the address space?
68010? Never heard of it !
So seriously, thats one skanky looking mac.
I know the 68010 is or was a popular mod for the Sega Genesis. It reduces slowdown in some games, but totally breaks compatibility with other games and I think it breaks Sega CD and Sega 32X compatibility as well. Most just overclock the original 68000 instead to something like 10MHz instead
Just put another row of headers or a socket on the expansion board to raise it up a little, i think you were overthinking that a bit.
unfortunately it wont fit back in the case like that.
@@ActionRetro oh fair enough
The expansion board has hi and lo roms, can’t you just replace them?
Those ROMs interact with the stock ROMs from what I've read, but don't replace them
Dropping in 68010s was a popular cross-grade for early Amigas that used 68000s too. I believe there were/are some patch libraries for AmigaOS that make them a tiny bit faster than the equivalent 68000 too, but they aren't much of an upgrade in an Amiga either!
I think you hit the nail on the head with the reason Apple skipped this chip, a lot of other manufacturers did the same. There really was very little advantage to moving to 68010s except to Motorola's bank balance! Apple chose to skip a generation and went for 68020 and 68030s instead which offered more features and true 32-bit right through the architecture, meaning more RAM, more capability and better performance all around.
All these shenanigans are making me want to power up my turbo-charged 50MHz/8MB SE/30 again!
Good gravy, could you clean and retrobrite that Mac's outside in the mean time?
4:00 savage WHAT error?
would love to get my hands on a Mac Classic again and upgrade it with a 68010 that does make it faster, is the 10MHz version actually out there? I think there was some super fast SoC upgrade that was insanely fast that could be used.
A Turbo Mac? I should be needed that.
Okay. He just inserted the processor in to the upgrade. My hardware experience screams “nooooo!!!! At least boot it once without the upgrade.” It’s rare. The chance of blowing something is just too risky.
Now I will watch what happens. Fingers crossed.
Exotic…and mildly stressful ways. 🤣
Wait... but did you try any ALU operations beyond addition?
I put a 68010 in my Amiga 1000 back in the day. I still have the chip but not the Amiga.
One of the best videos yet!
14:17 can you just replace your old two roms with new one?
why not add a second riser to the monster mac expansion? and/or adding the rominator to the monster mac board?--are those rom chips socketed as well?
Just get some female headers and stack them on the CPU socket to raise the board up so it clears the ROMs, why complicate things?
Too tall for the cases clearance
You're not hitting it hard enough...
Yeahhh that's definitely a turbo upgrade 🤣
How did you get the telephone software?
Speaking of a freak Mac how's the cursed Mac doing haven't heard about that for a while
Excited, and can't wait
Did the PC world ever have this level of customization? It seems the Amiga and Atari systems of the time also had these types of crazy upgrade paths. Your typical PC though, except for maybe some of the CPU "accelerator" sockets I don't recall anything as crazy for memory.
My Epson Equity 1+ actually had RAM expansion slots and I kept that into the late 90s adding VGA, ethernet, hard disks. But PCs were always built with that in mind it seems to a certain degree.
While fascinating it's also a little sad the lengths other vendors required their customers to hack a machine to get simple upgrades.
Obviously it's much more limited since the PC was effectively a 16-bit platform for a lot longer, while all of those are m68k systems (with at least 24-bit addressing like the 80286 in protected mode). If you couldn't reasonably upgrade using larger SIMM modules, you often had things like 640K memory expansions bundled with hard-drive controller cards. EMS cards were literally bank switching memory mappers to get addon memory beyond the megabyte barrier imposed by the CPU instruction set (and XMS later merely provided a means to copy from and to beyond it in real mode).
x86 CPUs were obviously upgrade-restricted by generation, but even first-gen PCs had 8088 replacement options like the NEC V20 (the Equity II came with a V30 as an 8086 drop-in by default).
Very Cool Sean 68010 Did not know that that was even available. Note on the Rominator K Koba And JDW was kind enough to help me to figure out what was wrong with the first one I built. The second Rominator that I built for the cursed Mac Plus Worked Kinda K Koba said to make sure the pins are seated all the way down on the stock rom sockets no gaps I had to shave down the stock rom sockets after doing that it worked fine but in your situation I don’t know hows that’s going to work hopefully you will find A work around for the Rominator I have the Rominator optional Flash tool that you can down load. And yes you can some really cool thing with the rominator Garth Beagle And Steve Put the Blue SCSI Image On Macintosh Garden For my Mac Plus with the Roninator Flash Tools But it is for the Mac Plus only sys 7.1 But Fantastic Video Sean
Why not just use another socket to raise the RAM expander? They are all male on the bottom and female on the top, or would that create clearance issues when trying to fit the board back into the Mac? Alternatively, the ROM sockets are pin through so you could mount sockets on the bottom of the board and just mirror the pinouts, that way you can fit the ROM expander after you install the board into the system.
I’m not sure how the boards work, but I wonder if you can put the sockets under the board. Somehow… uh… maybe not.
Come for the science, stay for the shenanigans.
You chose to move the object that would be the hardest to do.. all the stray frequencies.. voltage drops would make it harder if not impossible to do.
I do the same crap. lol. The hard way is what I focus on to much.
Nice Tommy Boy shirt :D
Is the 68010 mod possible with a stock Macintosh SE? Are there any benefits to doing so?
Nice shirt, Tommy Boy
Hi i like your videos
I don't know why the 68010 would go any faster if it is just a bugfixed 68000 with extra features not coded for in Mac OS, and at the same clock speed.
ive read that there were optimizations as well
@@ActionRetro Yes, it has a primitive instruction cache. Wikipedia sez that it adds "Loop mode," "which accelerates loops consisting of only two instructions, such as a MOVE and a DBRA. The two-instruction mini-loop opcodes are prefetched and held in the 6-byte instruction cache while subsequent memory read/write cycles are only needed for the data operands for the duration of the loop. It provided for performance improvements averaging 50%, as a result of the elimination of instruction opcodes fetching during the loop." "In practice, the overall speed gain over a 68000 at the same frequency is less than 10%. "
Probably, the benchmark program just doesn't use many (or any) 2-instruction loops that would benefit. But it seems likely that in a more diverse range of software, there would be some such loops that would benefit.