MacUser 1996: How to Rig Up the Fastest Photoshop Machine | Hacker News Reader

MacUser 1996: How to Rig Up the Fastest Photoshop Machine

archive.org

128 pointsdoener54 comments

The fastest (Mac) Photoshop Machine in 1996 was not a Mac at all and not running PPC either. Apple never transitioned from 68040 processors in their high end Macs to 68060 processors, instead opting for PPC processors.

But since Photoshop 4 was made for 680x0 processors, the PPC had to run software emulation. So the fastest abomination was a Commodore Amiga 4000 with DMA-transfer SCSI-II drives, third party (super expensive) 68060 CPU upgrade, (super expensive) third party 24-bit graphics card add-on, (awkward installation) running a Macintosh 68k Emulator! (Shapeshifter if I recall.) It could even run dual monitors, with the Amiga OS still running on one monitor and the Mac on the other monitor.

You would run up to Mac OS 7 on this thing.

Another insane tidbit. If you run NetBSD on 680x0 hardware, there is an emulator (Basilisk II) which gives you a virtualised classic Mac running at native speed - no CPU emulation.)

Edit: even with all the expensive Amiga upgrades, if I recall correctly, it was all still a lot cheaper than the near-equivalent Mac... of course, already in 1998, the situation was different. Photoshop on PC was well supported and PCs were fast and the PPC Macs had gained speed, too.

I remember running Mac software on my A4000-040. It was smoother than on real 68k macs, likely due to the co-processors the Amiga had. I didn't use the emulator much as I found most things just as good or better on the Amiga side of things, with an asteroid shooter, Maelstrom, as an exception - loved that game :)

You might be interested in the Maelstrom GPL port.

https://github.com/libsdl-org/Maelstrom

As far as I can tell, Photoshop 3 in 1994 had a PowerPC version? If 68k emulation was a bottleneck, wouldn't they be testing Connectix Speed Doubler's improved emulator as an option alongside more exotic things like the DSP card?

Yes, I think you are right - but the explanation (this is a long time ago, so hazy memories) was the following scenario: you had invested in some Photoshop plugins, or there were some stock filters/plugins inside Photoshop which were not yet optimised for PPC. You wanted them to go fast, so you stayed on m68k for a while until the vendors sorted it out.

Also myth and lore was much more an everyday part of life then. It was hard to come by accurate information. I was really geeky, yet I never heard of the Speed Doubler until today. It indeed seems to be a very clever product, essentially JIT compiling 68k code?!

Correct - the main application would move to a new CPU/OS much MUCH faster than some or all of the filters/plugins, and many more were not part of Photoshop at the time.

You'd have things that are now considered standard tools that would be implemented by a plugin; I recall companies in the early 2000s that still had older Macs around for a particular plugin for a particular customer, as it never migrated.

Even now you have that problem; I have a photoshop plugin for x86 that hasn't moved to a M1 native yet.

The built-in MacOS 68k emulator also used dynamic recompilation starting with the PCI PowerMacs (introduced starting in 1995)

Ah, the brief era of licensed Mac PowerPC clones towards the end of Sculley. Apple (still under Sculley) killed them all by cutting them out of System 8 (Mac OS classic).

4 PowerPC 604 processors in a DayStar Genesis MP 600. Only $16k USD! ($31k in 2023 money.)

It was actually the next CEO, Gil Amelio, that cut off the clone makers.

Except for UMAX, who accepted the new clone terms (that would give Apple a % of sale price instead of a [too-low, assuming that the clones would be a race to the bottom] fixed fee per machine sold) and actually got to distribute MacOS 8 for a while.

Looking back, Amelio had some decent ideas on how to get Apple back on track but he just didn't have that inspiration to get people excited over it all. Things like iMac, the 4 different system tiers, reducing the amount of research projects to a fraction of what they had were all done under his watch. Steve however kicked it into over drive and actually got people talking about Apple again.

Also didn't help that a 2 hour presentation by Amelio felt like weeks!

As I saw it reading his book, he correctly identified what was wrong with Apple to get it to the point it was, and had good ideas to fix that. But he didn't have the respect of the rank-and-file to implement the ideas without being undermined, and his ideas for what to do next were all wrong (the stuff in the last chapter of his book when he rails at Jobs).

> Steve however kicked it into overdrive… Is there a specific action of Steve’s that comes to your mind?

A good example is that Apple had something like 350 different projects. Amelio ended up cutting down to around 50 to be more focused. Steve cut it down to just 12. The idea of making the business hyper lean and hyper focused at the same time.

Amelio started to focus on things like reducing the amount of SKU's and pushing hard on Education. Steve took that and actually communicated this to the world.

And that was the big thing Steve managed to do. Capture peoples imagination and put out the image that Apple wasn't dead in the water. By leaning hard on those few things that made them different at the time - like the iMac.

I thought it was Jobs.

The clones licensed System 7, so jobs took the next version, branded it 8 and forced it out, cutting them off.

Amelio identified that the clone deal was bad (fixed license cost instead of based on % of machine price) and rebranded to MacOS 8 in order to renegotiate to get a better clone deal.

Spindler made the original clone deals assuming they would be chasing aggressively low pricing as seen in the PC market but instead the clones ate Apple's high-end market, replacing Apple's high-end margin with cheap OS licenses. So Amelio wanted to renegotiate the deals to be based on the value of the machine so that Apple would not lose from a clone maker selling a high-end machine instead of Apple.

Jobs as iCEO continued the negotiations (after all, NeXT was also an OS-only company at that point so Jobs wasn't necessarily sold on the idea that Apple had to sell hardware) but nobody aside from UMAX bit on the more expensive MacOS 8 license (so yeah there was one clone they had a MacOS 8 license - you can find UMAX MacOS 8 CDs out there) so Jobs gave up, and there were no other clones.

So Amelio started it and Jobs finished it.

Yes, Amelio's departure coincided with the release of System 8 in July 1997. (The NeXT deal and Jobs' return was in February.)

To be fair, System 8 was more than just a service update for System 7: it introduced the Platinum UI, the Appearance API, a multithreaded Finder, and various Copland features. (This was quite a rich update, maybe second only to System 7. It would have been a bit strange to just give this all away as System 7.7.)

> System 8 was more than just a service update for System 7

A lot more, eventually. System 8.6 brought a fully pre-emptive kernel with multi processing (http://macos9lives.com/smforum/index.php?topic=3656.0)

Problem was that, for backwards compatibility, all applications ran in a single pre-emptively scheduled task that cooperatively scheduled between them. You could write additional tasks that got scheduled preemptively, but they couldn’t make all system calls (https://developer.apple.com/library/archive/documentation/Ca...)

7.6 felt like Snow Leopard. Great stable OS after the flakiness of low 7s.

Then 8 felt like a huge shift to the modern. Yes, would have been weird to be 7.7. An ideal break point.

Seems odd the clones licensed by OS version, not duration.

9 didn’t have the same change feel.

System 9 was really just about some compatibility layers in preparation for OSX, as I remember it.

(And, yes, 7.6 was a major step in terms of stability. There was also some pre-PPC, pre-OpenDoc version of System 7 that was pretty stable, but with all those major integrations that came with the further iterations, things became a bit "delicate".)

IIRC Mac OS 9.0.x had significant changes for users directly booted into it, as did 9.1 (though to a lesser extent). I think it was 9.2.X that was mostly focused on usage with Classic.

Yes, might be right there. 9 introduced the changes to allow multiple-users on a Mac, with a log-in screen.

Going off fuzzy memory.

And then Classic/Carbon/Cocoa came around.

I had PowerTower 200, and I loved that thing. Felt so powerful after our old LC.

https://everymac.com/systems/powercc/powertower/powertower20...

Loved my Power Tower! BeOS, Rhapsody DR 1 and 2. But then I started running Linux on it in college and never looked back.

To be fair, PowerPC was dramatically faster at the time. Akin to the jump with M1.

Although it was supposed to be, PowerPC wasn’t the leap forward that the M1 was, partially because the system software was still largely run in 68k emulation and partially because Intel poured infinite money into making x86 faster to remain competitive with RISC processors.

> Although it was supposed to be, PowerPC wasn’t the leap forward that the M1 was

Just as with M1, it depends on your workload.

For instance, in the old school PBS show, The Computer Chronicles, they mention a prepress shop using the first PowerPC Macs that saw one of their workflows go from hours to minutes.

https://www.youtube.com/watch?v=Ic0dkf1iFOY

Speaking from my own impressions, the UI was noticeably faster. There were many applications - mp3 playing, for example - that only worked on PowerPC.

Over time Intel became faster, but there was a period of ~1-3? years when PPC was faster. AFAIR :)

SMP was sort-of included mid System 7. I doubt many programs took advantage of it because the programming model isn't like async GCD-based with origins in NeXTSTEP (NS...) like it is now in macOS X+. I'm guessing there wasn't much to take advantage of multiple processors except specialized professional apps like Photoshop, Illustrator, PageMaker, QuarkXPress, LightWave, Maya, and not much else. That's the only reason they could justify their prices, effectively competing in a similar space as NeXTSTEP was.

I'll take two please

No personal checks and no refunds.

Cash or money orders only.

Delivery will be in 6 to 8 years. :P

A meta-recollection related to this (and takes me back to those days) --

The magazine issue linked here has 208 pages. I remember that in the years to come, you could tell how desperately the Apple / the Mac community was doing by the gradually thinner and thinner page count of the magazine on the newsstand.

And then maybe it grew back a bit and stabilized (but never reached this kind of heyday again) as all the ads shifted online at the same time that Apple began to recover.

I remember the same thing with magazines for 8-bit machines and later the Commodore Amiga as the user bases and readerships declined. But not just the thinner page count - sometimes your favorite magazine also switched to thinner (cheaper) paper!

I remember my dad reading Mac User and Mac World while I subscribed to Mac Addict

I miss this style of quirky magazine layouts. The web has made everything follow the same 12 column grid template with just one or two sans-serif fonts.

RIP Houdini cards.

https://en.wikipedia.org/wiki/Compatibility_card

Montana option ("Power Macintosh 7300/180 PC Compatible") included an almost complete Pentium 166 PC on a card for dual booting Mac and PC MS-DOS/Windows.

I maintain a working collection of every kind of Apple, Reply, and OrangePC card. Nubus, PDS, and PCI. :-)

They’re a lot of fun to play around with.

from the letters section:

>It's true that today there are some jobs that can be done only by mucking with raw HTML, and if you want to do those, you have to code them by hand. But as HTML evolves, so do HTML editors. And one day Real Soon Now, there will be absolutely no need to write HTML code by hand. After all, do you hand-code PostScript to get it to print well?

any day now...

To be fair, except developers that have to write HTML generating web apps most people rent some CMS or use one for free (WP or static ones) and don't have to know HTML, even people that were paid to write HTML in the 90s.

And about Postscript, we generate PDF for that now.

And the quality of "generated" HTML is vastly improved; this is from the era of Frontpage and Dreamweaver, remember; and the latter was vastly better than the former. People would also just edit in Word and export to HTML, and back then you'd get a hell of a HTML file from that!

Via https://mastodon.social/@256/110124267999865910

Wow, sometimes certain things just hit you in the age department. I remember seeing this issue in 1996, and flipping through the pages I recall many of the same ads in MacWorld at the time. That feels so, so long ago.

So. Much. Ads. And this is despite the 1/4th at the end is ads only.

Ads were more useful back then. It was a fast way to quickly see what was available. The internet was far, far smaller in 1996.

> So. Much. Ads. And this is despite the 1/4th at the end is ads only.

They sucked at the time, but they're great now.

Old ads are the best.

They are, but at the time there was so little content overall, that when I was done reading all the articles, I read all the ads, too.

Plus the signal value was much more real back then. If you could afford an ad (however small) in a real magazine with large distribution, you were legit.

Today if I see an ad it just means my filter bubble was shifted by some random youtube video I clicked on.

I actually miss that kind of ads because you would get some brilliantly creative ones. These days, the barrier to entry into the ad space is so low that ads became dumb and garish.

kind of refreshing though, compared to today where most ads are lifestyle adds (some beautiful person doing something), here they feel more creative?

Well, some of/from the big ones are good, sure, but most of them... have nothing with the product. And some of them like on the page 57.

Blast From The Future 20XX Headline Retrospective

> MacUser 2023: How to Rig Up the Fastest Image Generation Machine (archive.org)

Page 41 is awesome. Must have been a fold-out.

Oooooh are we talking dedicated scratch disks or was that more Photoshop ~2000?

That was totally a 90s thing.

Source: I worked at Apple and Apple Dealers in London in the 90s!

DSP coprocessors were hot shit for a minute there.

p. 27

> THE LATEST INCARNATION of Apple's resurgent Newton, the $799 MessagePad 130...

[emphasis mine].

Reply on news.ycombinator.com