Intel Macs

[Spock]

It's here we have 2 intel Macs, I'm not sure about anybody else but it feels kinda weird. I remember the PPC transition and I don't think I feel the same way about the intel transition, Im sure this kind of thread has been covered but as Classic Collectors that have 68k Macs and Pre-G3 PPC Macs, does this feel the same way as the last transition? And what are Your thoughts?

 

[kretzy]

It's here we have 2 intel Macs, I'm not sure about anybody else but it feels kinda weird. I remember the PPC transition and I don't think I feel the same way about the intel transition, Im sure this kind of thread has been covered but as Classic Collectors that have 68k Macs and Pre-G3 PPC Macs, does this feel the same way as the last transition? And what are Your thoughts?

While I don't collect makes, it does seem a bit strange in that the PB i'm currently using will one day be a collectors item.

 

[macEfan]

While I don't collect makes, it does seem a bit strange in that the PB i'm currently using will one day be a collectors item.

yep, I used to use an apple IIGS and I had never dreamed of it ever being collectable.. of course it is not yet, but hey, its getting there

It does feels weird to hear intel inside about macs ...

 

[wasimyaqoob]

I'm not happy about Apple using Intel cpu's - They should stick with PowerPC, but who the hell am i?

 

[risc]

I'm not happy about Apple using Intel cpu's - They should stick with PowerPC, but who the hell am i?

Yeah because the Intel Core Duo is the worst notebook CPU Apple have ever used. That PowerBook G5 was so much nicer, I remember buying my first portable G5 and thinking WOW THIS... no wait that was a dream!

 

[5300cs]

The 68k > PPC shift didn't seem as final and as absolute this is. This time around it seems more like "let's get all this PowerPC junk out of the way already" whereas the 68k > PPC transfer was like "come come, it's time to move on to the new."

Maybe that's just me ...

Also I don't like the name "MacBook". "PowerBook" has been in the family since 1991. Isn't it bad enough they killed the Happy Mac face??

 

[zap2]

i want a faster iMac, i'm not to sure i like having notebook CPUs in my desktop, while they are fast they seem not as good as the G5(but i know they are)
G5 is to good for notebooks

 

[ahunter3]

I don't mind so much going Intel (it's a good company that has made a long string of very good chips) but I'm unhappy about going x86.

The only reason the x86 instruction set is still cookin' is momentum and market share. The very good chips that Intel (and AMD for that matter) make have to push around this horrid vintage klunky mass of 1980s-vintage code, essentially emulating a 386-DX chip in deep-pipelined RISC-like chips, ripping apart those ugly instructions into piece of equal length and guessing how to execute and retire them. Huge portions of the chip hardware exist only for accomplishing this emulation of a long-dead processor that had its heyday when Windows 3.11 was young.

That's a lot of unfortunate overhead, and only the economies of scale have been driving the research & development necessary to not only do it but do it very fast.

The PowerPC instruction set is much cleaner; if R&D at either Motorola or IBM had put all their resources into making chips that would execute it efficiently, the PowerPC should by all rights have left the x86 in the dust. And the chips themselves have been simpler, more streamlined, which indicates a longer subsequent lifetime.

I don't blame Apple for not only switching to Intel but specifically switching to the exact same processor that the rest of the industry uses: Apple is tired of being burned by the lack of serious R&D on the smaller-market chips that Macs have used result in PCs generally being faster. Every time the Mac would catch up (the first-generation PPC chips were as good as competing Pentiums) or get in front (the G3 kicked butt on arrival, and the G5 beat everything but the fastest Opterons and Xeons), it would then lag there while the x86 architecture caught up and went on past again. This way, the same market forces acting to make Dells fast will make Macs fast.

Problem is, it's a sadly burdened chip-architecture and now we're riding it along with the PC. The same efforts on this family of chips will result in less performance than identical development efforts on a more modern and less legacy-hampered design.

 

[Seasought]

Now Mac users can sub-divide themselves into "Purists" (those using PPC machines) and "Bunnies" (those using the new Intel chips). After all, internal strife always makes things more interesting.

 

[Anonymous Freak]

Nah...

I'm a 'true' purist. I'm posting this on a 68LC040-equipped PowerBook 520c, connected via AirPort. (Using iCab 2.9.8) Who wants this new-fangled 'PowerPC' thingy?

Oh, and I already have my Intel Mac... Yes, indeed! A desktop computer made by Apple, running the Mac OS, with an Intel core processor inside. And it can also dual-boot Windows just fine. No PowerPC processor.

What is it? It's a Quadra 610 DOS Compatible. 68040 at 25 MHz, 486 at 25 MHz. Runs Mac OS 8.1 and Windows 95 simultaneously.

 

[ahunter3]

Quadra 610 DOS Compatible. 68040 at 25 MHz, 486 at 25 MHz. Runs Mac OS 8.1 and Windows 95 simultaneously.

If you could get your hands on A/UX, you could run that, too, I think

 

[Spock]

No PowerPC processor.

I really hope Your joking, Windows and DOS were the only things running on that intel chip the Mac OS still runs on the PowerPC chipset not the intel

 

[portent]

I really hope Your joking, Windows and DOS were the only things running on that intel chip the Mac OS still runs on the PowerPC chipset not the intel

It's a Quadra...the Mac OS would run on the Motorola '040; that was before PowerPC.

 

[RacerX]

Addressing the original question of the thread... yes, this does feel a lot like the first transition.

First of all, there were almost no PowerPC native apps when Apple released the first Power Macintoshes... in fact most of System 7.1.2 was run in emulation on those systems.

The first complaints came from people who needed an FPU for their software. The processor that was emulated was a 68LC040... which lacked the built in FPU. So even though Apple had been talking about the incredibly fast floating point abilities of the PowerPC processors, there was no way to access it using 68k apps. Because of this, many apps ran faster on Quadras than on the new Power Macintoshes.

What is sad about the current transition is the fact that we are losing floating point abilities. Intel has yet to put out a processor that can match IBM's processors in floating point operations. But that was never one of Intel's main concerns either, so this is not surprising.


Oh, and yes, there was speculation back then about the ability to run Windows on Power Macintoshes. Microsoft had ported Windows NT to PowerPC, so people were talking about installing NT 3.5 on the new systems.

Also, Apple had not originally completely ruled out going with Motorola's 68060 processor. But Motorola was late (again) with releasing it and it came with the same type of compatibility issues that plagued the release of the 68040 processor. One of the reasons why NeXT had 68040 based systems before Apple was that the 68040 processor had compatibility issues with Mac software designed for the 68000/68020/68030 series processors. I can recall reading in Mac User and MacWorld warnings about making sure that your critical apps worked with the 68040 processor before spending the money for a Quadra 700 or 900.

I still have apps that won't run on a 68040 system. As I recall, my copy of Macromind Director 2.0 won't run on any of my 68040 systems. And Theorist 1.0 complains that it can't find an FPU on those systems (it looks for either a 68881 or 68882 floating point processor).

 

[Spock]

It's a Quadra...the Mac OS would run on the Motorola '040; that was before PowerPC.

Yes Im sorry I forgot, You know what I meant.

 

[blakespot]

...
Problem is, it's a sadly burdened chip-architecture and now we're riding it along with the PC. The same efforts on this family of chips will result in less performance than identical development efforts on a more modern and less legacy-hampered design.

The "Core Duo" is based on the Pentium M. This is a good thing. This is nothing like a dual Pentium 4.

This is technology derived from the Pentium III, a superior architeecture to the P4.



blakespot

 

[Anonymous Freak]

The "Core Duo" is based on the Pentium M. This is a good thing. This is nothing like a dual Pentium 4.

This is technology derived from the Pentium III, a superior architeecture to the P4.

Not necessarily 'superior', just 'more efficient'.

The Core Duo's 'core logic' dates back to the 1995-era Pentium Pro. It has had many new things tacked on, resulting in the Pentium III in 1999. Then, Intel decided to go with an all new core for the Pentium 4. One that was less efficient, but could be scaled to higher clock speeds. The idea was that yes, clock-for-clock, then P4 would be slower than the P3; but the P4 would go so much faster due to its faster clock speed.

Well, the move from a 130 nm to a 90 nm process killed that theory. It turned out that they couldn't squeeze that much more clock speed out of the process improvement, as had been done with every prior process improvement. Combined with power loads topping 100 Watts for the processor alone, made continuing the 'NetBurst' core of the P4 untenable.

On a side-step, Intel figured out that the original Pentium 4 was horrible for use in a notebook. It was hot, and to make it run cool, you had to run it slower than a Pentium 3. So they developed the 'Pentium 3 M', which was a Pentium 3 with a larger L2 cache. This totally blew the P4 out of the water in Apple's now famous 'performance per watt' benchmark. Yes, the highest end P4 were still overall faster (They also had SSE2,) but once you clocked the P4 down to run reasonably in a notebook, it became a loser.

So you had the choice between the high clockspeed, high bus speed, high power usage, small cache Pentium 4 and the low clockspeed, low bus speed, low power usage, large cache Pentium 3 M. Intel saw that the old P6 core (the name of the core in the Pentium Pro through P3,) could still have life. So their Israel division set about to modernize it. They added drastic power saving features (Enhanced SpeedStep,) a positively HUGE L2 cache (1 MB, when other processors were only at 256 kB,) added SSE2, and gave it the same bus as the Pentium 4. This resulted in the 'Pentium M'. This made it competitive again. This is where the 1.4 GHz Pentium M could tie or beat a 2.4 GHz Pentium 4.

Well, the Pentium 4 has gone through a few more revisions, gaining, among other things, HyperThreading, 64-bit, SSE3, and a faster bus (from 400 MHz to 800 MHz and 1 GHz on some 'Extreme Edition' models.) The Pentium M lagged, only gaining a slightly faster bus (from 400 MHz to 533 MHz.) Then came dual-core.

The Pentium 4/Pentium D dual-core systems are pretty much two completely separate processors slapped in a single package. Effectively identical to if they had used two separate sockets. The mobile dual-core solution was again redesigned from scratch for mobile use, just like the original Pentium M design had done. Still missing are HyperThreading and 64-bit, but we now have SSE3, and a faster (667 MHz, still not 800 MHz or 1 GHz, though,) bus. And a dual-core design that was meant just for mobile use. Sharing a single 2 MB L2 cache allows the system to shut down one whole core, but still get the speed boost of a 2 MB cache.

It's not that the NetBurst (Pentium 4/Pentium D) architecture is worse, it's just less efficient than the P6 (Pentium Pro -> Core Duo) architecture. It was a good idea when Intel thought it up. They honestly thought it would carry them for a decade; that process improvements alone would help with power consumption, and give speed boosts. They were wrong.

 

[blakespot]

Not necessarily 'superior', just 'more efficient'.

The Core Duo's 'core logic' dates back to the 1995-era Pentium Pro. It has had many new things tacked on, resulting in the Pentium III in 1999. Then, Intel decided to go with an all new core for the Pentium 4. One that was less efficient, but could be scaled to higher clock speeds. The idea was that yes, clock-for-clock, then P4 would be slower than the P3; but the P4 would go so much faster due to its faster clock speed.
...

As pointed out by ArsTechnica, the long-pipe design of the P4 that allowed it to run at a much higher clock than the P3 was driven my marketing, not a desire for as robust a CPU as possible:

I was one of the original dissenters from this school of thought, and in my P4 vs. the G4e series I tried to make a plausible technical case for why the P4's designers had made some of the design decisions that they did. I ultimately managed to convince myself and not a few others that the P4's deeply pipelined design was, in fact, performance-driven and not marketing-driven.

That was then, and this is now. As it turns out, the P4 bashers were right. Revelations from former members of the P4's design team, as well as my own off-the-record conversations with Intel folks, all indicate that the P4's design was the result of a marketing-driven focus on clock speeds at the expense of actual performance and scalability.

It's my understanding that this fact is pretty widely known within Intel, even though it's not publicly acknowledged.

So while a long-pipe, high-clock CPU can, indeed, be a real performer, the underlying engineering - in my opinion - is weak, as compared to a CPU designed to crank through numbers by virtue of its per-clock power. Can engineering work be improved by pressure from Public Relations and Advertising to create a CPU allowing the company to tout a not-so-critical spec of the processor?

I view this evolution of the Pentium M as a return to intelligent CPU design.



blakespot

 

[Anonymous Freak]

As pointed out by ArsTechnica, the long-pipe design of the P4 that allowed it to run at a much higher clock than the P3 was driven my marketing, not a desire for as robust a CPU as possible:

So, let's dredge up this old thread...

Yeah, the 'NetBurst' architecture was designed during the 'dark days' at Intel, when Intel also thought RAMBUS would win the memory wars over DDR. We know how that turned out. (If you don't know what RAMBUS is... Well, that makes my point for me.) It wasn't that marketing forced these decisions on engineering, it was just 'hinted' by marketing that high clock speeds would be really great. The NetBurst core did have some other real innovations; some of them introduced specifically because of the long pipeline, to cut down on its negative effects (Trace Cache,) some totally unrelated (SSE2.)

RAMBUS, on the other hand, was one of those things that from a deep-down technical perspective, on paper, looked better than the alternative. When reality struck, manufacturing problems, licensing issues, and other glitches caused it to not live up to its promise.


To move back to the original topic (which I realized I forgot to answer in my first post...)

Does this feel like the 68k to PPC transition? No.

Pretty much everyone knew that 68k was dead. Even Motorola had no plan beyond the 68060, and that chip wasn't fully compatible with the rest of the line, so it would have taken a lot of effort to get it shoehorned into a Mac. So when Apple announced the new PowerPC, it wasn't a 'well, but the 68k might be better' kind of thing. It was a 'WOW! Real high-end workstation quality chips!' moment. I remember being truly excited over the PowerPC processor. Yeah, it took a couple revisions to work it all out, but the PowerPC did go a lot further than the 68k could have gone.

It helped that a PowerMac could run 100% of previous Mac software (well, with the exception of rare, really old, 32-bit dirty apps.) And the transition was a slow process. The initial Power Mac line came out in March of 1994, less than 6 months after the last compact B&W Mac was discontinued (the Performa 200, a rebadged Classic II.) The Quadra 630 was introduced four months AFTER the first Power Mac, and not discontinued until the Power Macs had been on the market for a year. And the 68040-powered PowerBook 190 was introduced at the same time as the first-PPC PowerBook 5300 in August 1995; and not discontinued until September 1996. That was a full two-and-a-half year transition, with a year and a half between desktop and laptop intros. Heck, the 'professional' desktop model came first, with the consumer models trailing at the end.

Compare to the Intel transition, where you can't run Classic apps, the PowerPC line still appears to have some life left in it, it looks like the transition will be over in under a year, and the professional desktop looks like it'll be the last to transition.

One of the biggest visible differences was the Graphing Calculator. It came on the Power Macs, and was a great way to show off the power of the PowerPC. The 68k's couldn't do that kind of work, it took a PowerPC! There's no equivalent this time; it's more just Steve saying "Trust me."

(Heck, one of my first test when my MBP arrives will be to install the OS X version of Graphing Calculator, and compare it to the Windows version on my 2.26 GHz Pentium-M and to my 1.25 GHz eMac, which will be running it natively. (They don't have a universal binary yet.)

 

[RacerX]

Does this feel like the 68k to PPC transition? No.

Pretty much everyone knew that 68k was dead. Even Motorola had no plan beyond the 68060, and that chip wasn't fully compatible with the rest of the line, so it would have taken a lot of effort to get it shoehorned into a Mac.

It should be noted that Apple put some serious consideration into going with the 68060 processor in 1993/94.

How serious?

The logic boards of 6100, 7100 and 8100 systems were designed to run either the PowerPC 601 or the MC 68060. Apple wasn't completely sure if they were ready for PowerPC yet and the deciding factor was if Motorola was going to be able to deliver on time.

Motorola didn't have any plans beyond 68060 because they were still working on getting the 68060 into production. Plus the 68k line was losing customers pretty fast. Silicon Graphics left after the 68020, NeXT stopped making hardware with the 68040, and Apple had plans to move to the PowerPC line. All of Motorola's energies were directed at the 68060 for reviving the series... which obviously didn't happen.

It should also be noted that the backwards compatibility issues faced with the 68060 were very similar to the types of issues Mac users had with the 68040 when it was introduced. I can remember reviewers in most of the Mac magazines of the time cautioning people about rushing into using Apple's Quadra line. If you weren't 100% sure that your apps would run correctly on the new 68040 processors, staying with your current system (or moving up to a system like the Macintosh IIci) might be a better choice (I don't recall anyone pushing the IIfx as a Quadra alternative back then... mainly because of the odd SIMMs it used).

At any rate, Apple did hedge their bets back then... and we all know what happen (Motorola couldn't deliver ).

 

[Maxwell Smart]

I'm looking forward to this transition, I see it as Apple finally coming onto the same playing feild as PCs, and then being able to show them up in not only looks, but speed and choice of operating system as well. And FINALLY it will be so simple to compare a PC to a Mac Just my $0.02...

 

[Anonymous Freak]

The logic boards of 6100, 7100 and 8100 systems were designed to run either the PowerPC 601 or the MC 68060. Apple wasn't completely sure if they were ready for PowerPC yet and the deciding factor was if Motorola was going to be able to deliver on time.

Yeah, but you can't just drop one in. (I've also heard rumors that the PowerBook 500 series *CAN* accept a 68060 as a drop-in replacement, but the OS isn't compatible.)

Motorola didn't have any plans beyond 68060 because they were still working on getting the 68060 into production. Plus the 68k line was losing customers pretty fast. Silicon Graphics left after the 68020, NeXT stopped making hardware with the 68040, and Apple had plans to move to the PowerPC line. All of Motorola's energies were directed at the 68060 for reviving the series... which obviously didn't happen.

'Betting the farm' is the appropriate analogy here. Processor companies plan 2-3 generations out. They put it all on the 68060 because they knew the 68k line was dead. Sort of like Intel and the Pentium D. They know that the NetBurst core is dead, so they just threw together one last chip.

edit: And when I say 'dead', I mean new products in mainstream 'PCs', including the Mac. The 68k line has had a very long life, as it is still used to this day in some PDAs; although ARM is taking over there. And the Amiga topped out with a 68040, but there were 68060 upgrades available, because Amiga built 060 compatibility into the ROMs from the start.

It should also be noted that the backwards compatibility issues faced with the 68060 were very similar to the types of issues Mac users had with the 68040 when it was introduced. I can remember reviewers in most of the Mac magazines of the time cautioning people about rushing into using Apple's Quadra line. If you weren't 100% sure that your apps would run correctly on the new 68040 processors, staying with your current system (or moving up to a system like the Macintosh IIci) might be a better choice (I don't recall anyone pushing the IIfx as a Quadra alternative back then... mainly because of the odd SIMMs it used).

At any rate, Apple did hedge their bets back then... and we all know what happen (Motorola couldn't deliver ).

heh... Apple sure likes hedging their bets, eh?

And I don't remember the 68040 worries being that big. I remember the transition to 32-bit clean being a MUCH bigger deal, with stubbornly staying in 24-bit mode for no good reason until Apple finally did away with it altogether. I think the Intel transition is, by far, the most 'worrisome' to some purists, though; simply because it isn't a radical improvement in speed, but it is a radical departure requiring new software for full compatibility. (At least for MOST things, the PowerPC was faster in emulation than a true-blue 68k. And even where it wasn't, it was faster than MOST 68ks at least. Only higher-end Quadras tended to beat the early PPCs. Now, we've got cases where the low-end one-year-old iBook can beat a MacBook Pro in many PPC-code tests. (At the time of the PPC intro, a one-year-old Classic II would have been blown away in everything by the PowerMac.)

 

[Cooknn]

Once all the major app's are universal and we start seeing Mac Pro's rocking with the latest and greatest Intel CPU's we'll all be very thankful for the transition. IMHO that is

I guess that's pretty much of a no brainer...

 

[RacerX]

And I don't remember the 68040 worries being that big. I remember the transition to 32-bit clean being a MUCH bigger deal, with stubbornly staying in 24-bit mode for no good reason until Apple finally did away with it altogether.

The problems were significant enough to slow Apple's introduction and adoption of the processor.

Both the Quadra 700 and 900 were introduced in October of 1991. By February of 1991 NeXT had already moved their entire line (NeXTstation, NeXTstation Color and NeXTcube) to 68040 processors (introduced by Jobs in September of 1991). Not only that, the NeXT line of systems were priced comparably to Apple's line of systems at the time.

NeXTstation (68040 at 25 MHz, 8 MB of RAM, 105 MB hard drive, 2 bit (black & white) 17" display, Ethernet) $4,995.00
Macintosh IIsi (68030 at 20 MHz, 5 MB of RAM, 80 MB hard drive, 8 bit 12" display, LocalTalk) $5,097.00

NeXTstation Color (68040 at 25 MHz, 12 MB of RAM, 105 MB hard drive, 16 bit (color) 17" display, Ethernet) $7,995.00
Macintosh IIci (68030 at 20 MHz, 4 MB of RAM, 80 MB hard drive, 8 bit 13" display, LocalTalk) $7,897.00

NeXTcube (68040 with 25 MHz, 16 MB of RAM, 340 MB hard drive, 2 bit (black & white) 17" display, Ethernet) $11,495.00
Macintosh IIfx (68030 with 40 MHz, 4 MB of RAM, 160 MB hard drive, 8 bit 12" display, LocalTalk) $11,497.00

NeXTcube with NeXTdimension graphics board (68040 at 25 MHz, 24 MB of RAM, 340 MB hard drive, i860 DSP at 33 MHz, 32 bit (color) 17" display, Ethernet) $17,615.00
Macintosh IIfx (68030 at 40 MHz, 8 MB of RAM, 160 MB hard drive, 8 bit (color) 19" display, LocalTalk) $17,196.00​

I think the fact that NeXT (a company known for having over priced hardware) could introduce and adopt the 68040 processor and match Apple's prices while Apple was still selling 68030 based systems is quite telling. It wasn't like Motorola had a deal with NeXT for them to get the processor first (and at a significantly reduced cost). Motorola introduced the processor and Apple wasn't able to used them right away due to software issues.

When Apple finally got around to offering 68040 system, third party developers hadn't had the chance to solve their issues with the new processors yet.

One of the key problems that I recall was problems with the FPU. On most Mac IIs the 68030 was accompanied by a 68882 FPU (the Macintosh II had a 68020 with a 68881 FPU and the Macintosh IIsi didn't come with an FPU on the logic board). The 68040 came with a very fast FPU on the chip itself... which Mac software at the time couldn't find when it search for an FPU. On software that required an FPU, the software wouldn't run, and on software that used an FPU but could run without it, that software would run slowly not attempting to use the built-in FPU.

And that is just looking at math intensive software. This was also the System 7 transition (both the Quadra 700 and 900 originally came with System 7.0.1). I was dual booting my SE/30 between System 6.0.8 and System 7.0.1 around this time to deal with software that wouldn't function right in System 7... and I didn't even have a 32 bit enabler installed at the time (I had 8 MB of RAM which at the time was a major amount for a Mac).

I was following the reviews at the time as many of my friends an I were aware of NeXT using the 68040 processor and had been waiting to see how an 040 Mac would perform. Specially as I knew people in our department that were using the IIfx at the time and would have been the first in line to get a Quadra 700 or 900 (a few of them got them anyways and used both the Quadra and the IIfx next to each other until the software issues were fixed).

 

[Anonymous Freak]

Both the Quadra 700 and 900 were introduced in October of 1991. By February of 1991 NeXT had already moved their entire line (NeXTstation, NeXTstation Color and NeXTcube) to 68040 processors (introduced by Jobs in September of 1991). Not only that, the NeXT line of systems were priced comparably to Apple's line of systems at the time.

I think the fact that NeXT (a company known for having over priced hardware) could introduce and adopt the 68040 processor and match Apple's prices while Apple was still selling 68030 based systems is quite telling. It wasn't like Motorola had a deal with NeXT for them to get the processor first (and at a significantly reduced cost). Motorola introduced the processor and Apple wasn't able to used them right away due to software issues.

Yeah, but NeXT was way more competent than Apple at the time. I hadn't remembered the price similarity, though.

One of the key problems that I recall was problems with the FPU. On most Mac IIs the 68030 was accompanied by a 68882 FPU (the Macintosh II had a 68020 with a 68881 FPU and the Macintosh IIsi didn't come with an FPU on the logic board). The 68040 came with a very fast FPU on the chip itself... which Mac software at the time couldn't find when it search for an FPU. On software that required an FPU, the software wouldn't run, and on software that used an FPU but could run without it, that software would run slowly not attempting to use the built-in FPU.

That's right. I apologize, it's been so long I've obviously forgotten some of these details. That's why the original 68k emulator for PPC only emulated a 68020 with FPU, now that I remember. (Pretty sure it was a '20, not a '30. know it wasn't a '40.)

And that is just looking at math intensive software. This was also the System 7 transition (both the Quadra 700 and 900 originally came with System 7.0.1). I was dual booting my SE/30 between System 6.0.8 and System 7.0.1 around this time to deal with software that wouldn't function right in System 7... and I didn't even have a 32 bit enabler installed at the time (I had 8 MB of RAM which at the time was a major amount for a Mac).

Yeah, that was mostly the 24-to-32-bit conversion (the 32-bit enabler didn't come for dirty systems until a year later, IIRC.)