History Repeating? (68K --> PPC --> Intel --> ARM)

[Bug-Creator]

Ohhhh, ok. Sorry, I thought you were saying that 68k could have been advanced further, Moto just chose not to do that.

Sure the could have done a 070/080/etc, but with the resources they had those would have fallen further behind compared to Pentium2/3/4.

The 060 did perform o.k. clock by clock compared to the 1st gen Pentium but was a bit late and didn't clock as high.

 

[RogerWilco6502]

Sure the could have done a 070/080/etc, but with the resources they had those would have fallen further behind compared to Pentium2/3/4.

The 060 did perform o.k. clock by clock compared to the 1st gen Pentium but was a bit late and didn't clock as high.

Ah, ok. That makes sense.

 

[Amethyst1]

On a side note, does the 68k emulator (for applications) still exist in Mac OS 9?

I have run the Mac version of Prince of Persia (1992, so 68k) on OS 9 - so I'd say yes.

 

[RogerWilco6502]

I have run the Mac version of Prince of Persia (1992, so 68k) on OS 9 - so I'd say yes.

Awesome! That's what I thought. You don't hear it talked about as much in OS 9 literature as you do in, say, System 7 literature it seems.

 

[Amethyst1]

Now when we get to why PPC was dropped a decade later, that can actually be laid at Moto's feet for not coming up with anything substantial after the G4

Yep - by 2005 the G4 was handily beaten by the Pentium M (talking about laptops).

 

[weckart]

Inflation is a bitch.

Apple systems are price competitive with "Windows" offerings for desktop workstations with comparable hardware.

While that has been true the caveat is: Only at the time of launch. In subsequent years, PC sellers will either discount their current stock or offer speed bumps, more RAM or some other sweetener. Apple just draws a line in the sand and the price and specs remain frozen in time. The MP 2013 was probably one of the most egregious examples of this.

 

[Macbookprodude]

Can someone tell me how the 68k performed ? I did not get into Mac until the PPC days of 1999-2000 when i worked as a tech for CompUSA. 1st PM G4 was a 350 sawtooth which I may try to find one as I like this machine. It was my 1st, sadly in 2001 due to financial difficulties I had to get money at that time and I had to sell all of it, sadly. OS 9 + VPC 3.0 - all total I paid at CompUSA almost 1799.99 with my discount.

 

[Anonymous Freak]

Scenario: Intel (IBM in 2004) hasn't been able to bring the thermals down in order to ensure a powerful, thin and light laptop that sips battery. The Core series (G5 in 2004) is simply too hot even at 10nm (90nm in 2004). There have been promises of a cooler-powered CPU with the next 7nm Core architecture (POWER5/G6 in 2004) but delays have made Apple grow impatient and source other CPUs.

eeeehhhhh....

No.

Intel has ultra-low-power CPUs, has even made special ones just for Apple (the custom CPUs in the original MacBook Air, the custom desktop CPUs with the highest-spec integrated graphics that were normally reserved for mobile CPUs for the base model iMac, even the latest MacBook Air has slightly tweaked CPUs that aren't the same as non-Apple equivalents.) If the ARM transition happens, it won't be about "hitting power/performance targets," it will be purely about Apple having complete control over 100% of the system, top to bottom.

 

[DearthnVader]

While that has been true the caveat is: Only at the time of launch. In subsequent years, PC sellers will either discount their current stock or offer speed bumps, more RAM or some other sweetener. Apple just draws a line in the sand and the price and specs remain frozen in time. The MP 2013 was probably one of the most egregious examples of this.

Apple dropped the Quad 2013 and made the 6 core the price of the old quad.

I think they also upgraded the base graphics from the D300 to the D500 at no extra cost.

Given, Apple painted themselves in a thermal corner with the 2013 MP.

 

[Anonymous Freak]

Ohhhh, ok. Sorry, I thought you were saying that 68k could have been advanced further, Moto just chose not to do that.

On a side note, does the 68k emulator (for applications) still exist in Mac OS 9? It probably does given that I have to ask, I just want to know for certain

Motorola absolutely could have continued - the 68060 was actually released and used in computers (Frequently as an upgrade for Amiga computers; it wasn't 100% compatible with the 68040, in such a way that it didn't work in Mac OS, but did on Amiga as Amiga didn't use any of the incompatible code.) They just chose to go along with Apple and IBM and pursue PowerPC as their primary "computer CPU architecture". But the 68k line didn't die - it just became a "mobile" CPU line. The original Palm Pilot used a modified 68k, and the line continues to be offered today with the "ColdFire" CPU (NXP now owns the CPU line after Motorola transitioned to Freescale, then sold the product line to NXP.) With an embedded CPU at up to 300 MHz, with PCI controller, USB controller, Ethernet, and more. There was even a project to make an Amiga upgrade from it - that would run original Amiga software unmodified.

 

[RogerWilco6502]

Motorola absolutely could have continued - the 68060 was actually released and used in computers (Frequently as an upgrade for Amiga computers; it wasn't 100% compatible with the 68040, in such a way that it didn't work in Mac OS, but did on Amiga as Amiga didn't use any of the incompatible code.) They just chose to go along with Apple and IBM and pursue PowerPC as their primary "computer CPU architecture". But the 68k line didn't die - it just became a "mobile" CPU line. The original Palm Pilot used a modified 68k, and the line continues to be offered today with the "ColdFire" CPU (NXP now owns the CPU line after Motorola transitioned to Freescale, then sold the product line to NXP.) With an embedded CPU at up to 300 MHz, with PCI controller, USB controller, Ethernet, and more. There was even a project to make an Amiga upgrade from it - that would run original Amiga software unmodified.

Awesome! That makes sense, the TI-89 and related calculators use 68k processors

 

[Bug-Creator]

it wasn't 100% compatible with the 68040, in such a way that it didn't work in Mac OS, but did on Amiga as Amiga didn't use any of the incompatible code.)

Yes, no, almost true....

On bootup the 060 has an incompatible MMU/cache setup that would make an Amiga (and a Mac too) crash.

68060 cards had a little code in their ROM disabling cache which would later be reactivated by a special library.

The new AmigaOS3.14 has that code to, so if you have that as a Kickstart-ROM you can run an 68060 in a modified 68040 card without ROM.

All these setups will run ShapeShifter (aka MacOS) without any 060 specific patches.

Or in short if someone had wanted they could have done the same as an CPU-upgrade for 68k Macs.

But the 68k line didn't die - it just became a "mobile" CPU line. The original Palm Pilot used a modified 68k

DragonBalls are literally 68000s, thats like Intel selling you a (superlow power) 80286 today.

ColdFire was highly incompatible in ways that can't be 100% trapped in SW or HW. Stuff like reused opcodes, missing opcodes that don't trigger a fault and a different format for FPU data.
Also performance for these was always way below PPCs available at the same time.

The Amiga-CF card you are most likely referring to was 100% vapor ware. The guy just placed a few components on the PCB with no clue how to make them work in HW or SW.

Another project by Elbox wasn't much better as it could only have worked by running a full 68k EMU, hence superslow and could have used just any other CPU to achieve that same goal.
On top of that it was mounted behind a PCI backplane with no direct access to the AmigaHW, so it would either have to emulate those too (why not just use UAE...) or have a helper OS running the onboard 680x0.


Now, them Atari guys did pull it off with the FireBee but TOS was/is a much simpler OS.

 

[Amethyst1]

DragonBalls are literally 68000s, thats like Intel selling you a (superlow power) 80286 today.

That reminds me of the Quark.
[automerge]1589410782[/automerge]

(the custom CPUs in the original MacBook Air

...which were throttled. FWIW.

 

[rampancy]

Can someone tell me how the 68k performed ? I did not get into Mac until the PPC days of 1999-2000 when i worked as a tech for CompUSA. 1st PM G4 was a 350 sawtooth which I may try to find one as I like this machine. It was my 1st, sadly in 2001 due to financial difficulties I had to get money at that time and I had to sell all of it, sadly. OS 9 + VPC 3.0 - all total I paid at CompUSA almost 1799.99 with my discount.

Here's a text file from 1992 comparing the features and performance of the 68040 vs. the 486. Clock-for-clock, the 040 was superior but Motorola struggled to successfully clock it past 40 Mhz, whereas the 486 had reached 50 and 66 Mhz with the DX2 in 1993.

Or in short if someone had wanted they could have done the same as an CPU-upgrade for 68k Macs.

Daystar Digital was actually going to release a 68060-based accelerator card for 68K Macs in 1994 (I even remember reading about it in MacWorld) but apparently Apple nixed those plans.

 

[repairedCheese]

On a side note, does the 68k emulator (for applications) still exist in Mac OS 9? It probably does given that I have to ask, I just want to know for certain

I have run the Mac version of Prince of Persia (1992, so 68k) on OS 9 - so I'd say yes.

Oh, this didn't just exist in Mac OS 9, because of Classic, you could continue to use your early 68k Mac apps all the way to 10.4. The real limiting factor in my experience is that particularly once you get before 1990, games sort of generally expect a 16 color mode, and most Macs you can install 10.4 on don't have one. Because they're so old, they often think you're stuck at something lower and fail to start.

 

[Amethyst1]

Here's a text file from 1992 comparing the features and performance of the 68040 vs. the 486. Clock-for-clock, the 040 was superior but Motorola struggled to successfully clock it past 40 Mhz, whereas the 486 had reached 50 and 66 Mhz with the DX2 in 1993.

The Am5x86 reached 133 MHz in 1995 - Star Trek on one of those would be fun

 

[Bug-Creator]

Clock-for-clock, the 040 was superior but Motorola struggled to successfully clock it past 40 Mhz, whereas the 486 had reached 50 and 66 Mhz with the DX2 in 1993.

The listed clock on the 68040 was a bit odd, a 40MHz would need an external 80MHz clock with some parts running at that speed.
Sometimes used for overclocking 060 in Amiga cards that supported both 040 and 060. Put it in 040 mode and have the 060 run at 2x clock of the card (RAM, logic, SCSI) and you can reach 80 sometimes even 100MHz with a late mask 060 and ample cooling.

Daystar Digital was actually going to release a 68060-based accelerator card for 68K Macs in 1994 (I even remember reading about it in MacWorld) but apparently Apple nixed those plans.

Pretty much what I said, you have to prevent the OS/ROM to activate/use 040 features, which doesn't really work without replacing/adding/patching the ROM. No surprise that Apple's legal department might have had a word with you about that at that time.

 

[RogerWilco6502]

Oh, this didn't just exist in Mac OS 9, because of Classic, you could continue to use your early 68k Mac apps all the way to 10.4. The real limiting factor in my experience is that particularly once you get before 1990, games sort of generally expect a 16 color mode, and most Macs you can install 10.4 on don't have one. Because they're so old, they often think you're stuck at something lower and fail to start.

Awesome! Thanks for the info

 

[crammedberry]

With each of Apple's previous transitions 68k->PPC PPC->x86 it required that the new CPU could run the old code at acceptable speeds while developers ported their code.

Arm can not run x86 code at anywhere near "acceptable" speeds.

Tho, it would seem Apple is going to go that way because they can fab their own Arm cpu's.

I don't think Apple will dump x86, likely the Air the Mini and the MacBook will get Arm, and maybe the low end iMac, while the Pro line will retain x86.

Apple will do it just to make more profit, not because Arm is better for desktop macOS, and for that reason I think it's a mistake.

I remember Rosetta being quite slow when it came to emulating some PPC code and apps (sometimes more than twice as slow). That didn't stop Apple back then. The difference between the speed of high-end PPC chips (G5s) and Intel CPUs (Core singles and duos), while still present, also wasn't that extreme. The transition was based on roadmap more than present performance. The G5s delivered performance that matched Intel's offerings but were too power hungry to put in a laptop, and development for the G4s was pretty much frozen in time. This is a familiar situation, with constant delays from Intel and virtually flat performance gains over the past decade, especially compared to the decade prior. Sure, there have been some power efficiency improvements on Intel's side, but even those gains haven't matched Apple chips' efficiency. So this is a familiar situation all over again. The A13 (a hexacore chip with two high-performance cores) also outperforms most lower- to mid-range Intel chips and is only outmatched by the high-end chips, and the iPad Pro (A12x, an octacore chip with four high-performance cores) consistently outperforms Apple's Macbook Air and part of the Macbook Pro lineup. Some will argue that the benchmarks are not a real-life representation of performance, but nevertheless the performance is there when it comes to Apple's A-series chips. This is also performance we see with thermally limited designs like the iPhones and iPads. The performance of the next generation of these chips will most certainly outperform all but the highest-end Intel offerings if given adequate cooling and (more than likely) more cores, as Apple has done with the iPad Pro – an A14X laptop/desktop series processor would likely feature more high-performance cores than seen in the iPad Pro.

 

[Macbookprodude]

Could PPC have been saved ?

 

[weckart]

Could PPC have been saved ?

From what?

 

[rampancy]

Could PPC have been saved ?

Not likely. Motorola (and later Freescale) soon discovered that the embedded market was much larger (and more profitable) than Apple's share of the desktop market, and devoted their resources into developing their PPC chips to that end. One can see this in how the immediate successor to the PPC 74xx series -- the e600 -- was developed for embedded telecom systems. Ever since then, Freescale's PPC chips have targeting embedded markets like aerospace, automotive, and industrial applications.

On the other side, IBM wanted to futher develop its POWER series of PPC CPUs for its traditional market of servers and high performance computing, which again, didn't align with Apple's goals for power-efficient computers that maximized performance per watt, which could be put into thin and light mobile devices. (Hence Steve Jobs complaining about how they couldn't cram a G5 into a PowerBook.)

Computers like the MacBook Air, the controversial 12" MacBook, or the current iPad Pro are what Apple has always wanted to build, and as Steve himself said back in 2005, such products wouldn't have been possible (let alone market competitive) with either IBM or Freescale's PowerPC roadmap.

For that situation to have changed, Apple (as well as other consumer platforms that relied on PPC like the Amiga) would have had to gone through a massive desktop marketshare renaissance back in the later 90s and early 2000s. This would have lead to an installed customer base that would have incentivized the investment of more resources into developing PPC chips and a PPC roadmap more to Apple's liking.

 

[macgeek18]

Arm is an interesting thought. But here is my look at it. Arm is a good platform, but as someone who runs a PineBook Pro daily with highly optimized bleeding edge Manjaro, It's going to take A killer chip with killer optimization to be usable for most people. Desktop operating systems and ARM are not the most friendly together.

 

[2984839]

Arm is an interesting thought. But here is my look at it. Arm is a good platform, but as someone who runs a PineBook Pro daily with highly optimized bleeding edge Manjaro, It's going to take A killer chip with killer optimization to be usable for most people. Desktop operating systems and ARM are not the most friendly together.

I think it's doable. The Pinebook Pro is 1/5 the cost of the cheapest laptop Apple would release, so there's a lot of room for performance improvements.

sidenote: I ordered a Pinebook Pro back in April and can't wait for it to show up.

 

[Macbookprodude]

Not likely. Motorola (and later Freescale) soon discovered that the embedded market was much larger (and more profitable) than Apple's share of the desktop market, and devoted their resources into developing their PPC chips to that end. One can see this in how the immediate successor to the PPC 74xx series -- the e600 -- was developed for embedded telecom systems. Ever since then, Freescale's PPC chips have targeting embedded markets like aerospace, automotive, and industrial applications.

On the other side, IBM wanted to futher develop its POWER series of PPC CPUs for its traditional market of servers and high performance computing, which again, didn't align with Apple's goals for power-efficient computers that maximized performance per watt, which could be put into thin and light mobile devices. (Hence Steve Jobs complaining about how they couldn't cram a G5 into a PowerBook.)

Computers like the MacBook Air, the controversial 12" MacBook, or the current iPad Pro are what Apple has always wanted to build, and as Steve himself said back in 2005, such products wouldn't have been possible (let alone market competitive) with either IBM or Freescale's PowerPC roadmap.

For that situation to have changed, Apple (as well as other consumer platforms that relied on PPC like the Amiga) would have had to gone through a massive desktop marketshare renaissance back in the later 90s and early 2000s. This would have lead to an installed customer base that would have incentivized the investment of more resources into developing PPC chips and a PPC roadmap more to Apple's liking.

Thank you. I wonder if there was a G5 PowerBook prototype that Apple was trying to build ? And now, 20 years later we still are using our PPC Macs. Incredible how they are all working great today for most things.