Which Mac was the last that could run 9.22?

[Cory5412]

One note, just for accuracy's sake, there's a Quicksilver and Quicksilver 2002, and the MDD was originally launched in 2002, there's a 2003 revision with firewire 800 (not compatible with OS 9) and a further revision from 2003 which removes firewire 800 and adds the single 1.25GHz option and adds back in OS 9 compatibility. Sometimes that machine jokingly gets referred to as the "XPress Edition" (because Quark took longer than almost anybody else to build an OS X version of their software.)

The main thing that differentiates the QS'02 and makes it require newer OS install media than the dA/QS (which, again, are AFAIK literally identical) is that some components of the Ethernet interface changed and it requires newer drivers. The technical specs are otherwise basically the same and any dA can be upgraded with QS, QS'02 or third party parts to run at the faster speeds.

Here's the reasons why I'd choose a dA/QS/QS'02 for OS 9 over an MDD:

1) guaranteed ability to boot and install and run from the eMac 2003 install CD, at http://macintoshgarden.org/apps/emac-g410-ati-restore-discs (download #9) - this might work with the MDD, but I haven't personally verified it.

2) the fan control works a little better on the older systems. Mac OS 9 support on the MDD was a little slapdash, Apple did it because they had to, but you can tell the MDD is explicitly intended to be an OS X computer. If I remember correctly, ultimately what happens is the driver loses track and the machine runs the fans at fast speeds all the time, which, because the MDD has several fans, can be a little loud.

3) Most (not all, but most) MDDs come with dual CPU configurations and dual CPUs are advantages to very few OS 9 programs. Any OS 9 program that *can* use dual CPUs will almost certainly be a better experience if you just used it in OS X. Especially given that the MDD can run 2GB.

4) Mac OS 9 can't reasonably use over a gig of RAM. It can probably boot with it installed but IME most workflows don't need more than 256. I was editing DV with 576 megs of RAM back in the day.

Mostly, in situations where you've got both kinds in front of you, my personal recommendation is to pick the QS if you're looking to focus on 9 and pick the MDD if you're looking to focus on X, but either will run either "fine".

I personally wouldn't buy any of these machines based on Zip drive availability - at absolute worst, you can/should use a USB or Firewire zip drive or run a drive without the bezel. On the MDD, you should be able to mount a 3.5" device in the second optical bay, and you can just fold down the flap when you're putting media in. Notably, I personally have no recollection of ever seeing a Zip bezel specific to the MDD, so I'm guessing this is what the MDD's Zip750 installation kit  did, but I could be wrong, I looked, but not very hard.

 

[Cory5412]

In reality: these particular preferences are mostly aesthetic. I don't know of any situation that will cause an MDD to be less reliable or less performant than a QS. A couple say they are "noticeably" faster than the QSes are, but I just don't see it, myself. It will bench faster, but given that I don't usually notice responsiveness differences between my fast G4s and my "slow" G3s, or even fast 604es, I might not be the best authority on this. Both machines are on the absolute top of the heap though. I suspect even for someone who says they can notice, they are within spitting distance.

If you want something faster and an MDD is available conveniently and inexpensively, I would say: get it. At absolute worst, if you don't like how it performs or runs in OS 9 or if you don't like the fan noise, you can use OS X on it and keep using OS 9 on your G3.

 

[Trash80toHP_Mini]

In reality: these particular preferences are mostly aesthetic. I don't know of any situation that will cause an MDD to be less reliable or less performant than a QS.

Yep, that's why I put the aesthetic considerations relating to the USB FDD/LS-120 installation thing up front and then detailed how to do side by side comparisons on everymac.

QS and MDD processors moved L2 onto the die which is significantly faster than backside L2 in the DA. YMMV

MDD'03/FW400 system bus is 25% faster than those of the DA/QS twins. That's a pretty significant upgrade right there, YMMV

MDD bumps ATA-66 to ATA-100, which should make an SSD upgrade zippier, probably a statistically significant amount for some benchmarks over putting the same SSD in the earlier models. IRL? YMMV

They hopped up performance a bit after the QS styling makeover of the DA, but for the most part it was spiffing up the trim on the same sheet metal chassis even for the MDDs. But when all is said and done, Apple appears to have been biding its time until the new model G5 was ready for the streets. They had a TON of bughunting and optimization that needed to be done to X before the G5 shipped. OSX was just awful at first and it's debatable at which rev it became usable, certainly not in the DA generation and not for the QS in my book.

I think I'll pull the trusty DA out of Mothballs, install my extra QS CPU in there and see how the old girl handles on the dance floor when I get a chance. []

 

[Cory5412]

Nobody's going to dispute that the MDD should turn in higher benchmarking numbers than the dA/QS, for a variety of reasons. I will dispute that in Mac OS 9 the differences are meaningful enough for day-to-day use unless you're doing a couple of extremely high end tasks, at the absolute upper echelons of what Mac OS 9 was ever capable of.

In short: you can only launch Dreamweaver MX or render 1998's npr.org so instantly.

In particular, I'm thinking about Virtual PC, DV video editing (remember: Mac OS 9 was never really qualified for HD video work, that was exclusively within the realm of OS X), DVD Studio Pro, photoshop and DAW tasks.

Notably here, DV video editing, due to the compression being used, didn't really need that much disk throughput to work well. DV streams were about 13 megabits per second. That's not trivial, exactly, but the only interface in any G3+ Mac that can't do that is USB 1.1.

Notably, most of this stuff moved onto OS X pretty quickly. The app we named the 2003 re-release of the 9-capable MDD, QuarkXPress, was capable of running on '030 Macs in under 16 megs of RAM (though, your particular documents would impact what you wanted very heavily, and there were a number of different possible workflows with XPress.) HIgh end creative work benefited from Mac OS X almost instantly. The main reason to keep using OS 9 is if you couldn't or didn't want to upgrade app versions or if you had an app (namely: QuarkXPress) whose vendor was badly dragging its feet on an upgrade.

Almost anything that was either released as Carbon (runnable on both 9 and X) or dual-released (Office 2001 and v.x) was better on X than it was on 9, and that's even before you get into issues like multiprocessors, being able to reasonably use more than a gig of RAM, overall stability (as in: the OS isn't crashing all the time for no good reason) and better ambient multiprocessor support (in particular: OS can sort processes onto least busy processors, even if individual tasks aren't threaded.)

I often credit the iMac G3 and the overall G3 lineup for "saving Apple" but it's absolutely Mac OS X that made it worth switching to Macs if you weren't already used to and comfortable with the platform's many foibles.

I personally have more nostalgia for Mac OS 9, but I'm not under any false impressions that it's in any way objectively better than OS X, perhaps save anybody who was power-using spatial Finder.

QS and MDD processors moved L2 onto the die which is significantly faster than backside L2 in the DA. YMMV

You can run QS processors on a DA board. But, there's also G4 vs G4E issues. In some universe where I had a lot more money than I do to throw into this kind of thing, I'd take a pair of well-endowed QS'02s and a similar MDD and just try all the processors, likely with a few benchmarking suites to get an idea of what relative performance should have looked like in both OS 9 and OS X.

 But, that's a lot of money, time, and space that I don't have.

That's a pretty significant upgrade right there, YMMV

I would love to see proof that it's actually a meaningful upgrade - that it impacts the speed at which anything can get done or that it has a meaningful impact on usability. A side-by-side video of someone using both machines and pointing out spots where one is faster or slower than the other would be extremely instructive, for example.

Because, like, my 133MHz bus machines don't generally feel a lot faster than my 100 or 66MHz ones, in any given day-to-day, so I'm... skeptical.

One thing I'll note is that one of the person who posted saying they could feel it, had a QS and an MDD with almost every potential upgrade stuffed in, so perhaps if you pour hundreds of dollars into these machines, it makes a difference.

MDD bumps ATA-66 to ATA-100, which should make an SSD upgrade zippier, probably a statistically significant amount for some benchmarks over putting the same SSD in the earlier models. IRL? YMMV

I'll have to look again, but I could've sworn it was ATA100 to /133, or perhaps it's /66 to both /100 and /133 buses, as IIRC the MDD has a couple additional mounting points for hard disks over the previous machines, but: at /150MBps, a PCI SATA Card should be faster than either of these machines' onboard IDE buses.

Again, I know perfectly well these differences will show up in benchmarks. I don't know that I believe it'll show up in (*most) day-to-day workflows with Mac OS 9.

 

[Unknown_K]

I find OS 9 gets pretty fast (user experience) with a fast G3 setup and a good video card. On the mac side of things a super fast OS 9 machine gets limited by the drivers and code apple used. OSX on the same machine will seem faster just because of the better memory management and just the HD subsystem running full out. FSB does matter if the OS can uses it.

On thew PC side of things if you say stick Windows ME on a last generation (supported) Pentium 4 3.0ghz with native SATA and DDR support and a ATI X850 pro AGP card it runs almost too fast. Yes I have one of those.

 

[Trash80toHP_Mini]

I personally have more nostalgia for Mac OS 9, but I'm not under any false impressions that it's in any way objectively better than OS X, perhaps save anybody who was power-using spatial Finder.

I won't give you any argument at all there. I look at using fifteen+ year old OS, apps and hardware like dirt track racing in beaters as opposed to Formula 1 insanity. Heck, I'm setting up a Windows 98se machine out of nostalgia to run PowerPCB for Windows 95 to see what it was like to do circuit board design with full blown pro tools available five years after I'd done design, prototyping masks and rubytiths to send off for process camera based PCB production on my signmaking equipment. Sure I could do it with any number of packages available today, but living in a time capsule stretching between 68000 and G4 has the right feel.

 

[Gorgonops]

It beats out QEMU on an i5-6600 or so and on anything where I've run it (i5-2400, but I should give it a go on my i5-7300u as well, which should turbo really well and has platform improvements.)

Not to play thread necromancer too badly here, but I'm curious: what kind of benchmark results *do* you get from QEMU? Just this weekend I decided it was time to try Qemu as an OS 9 engine for the first time, and, well... I will say it's less annoying to set up than Sheepshaver. Performance-wise though I'm not quite sure about it.

Running Qemu 3.1.0 from Homebrew on a Mid-2015 Retina with a 2.2ghz i7 I'm getting CPU scores around 1400 in MacBench 4. (I'd love to run MacBench 5 but google-fu is failing me on finding the CD disk image.) FPU scores are particularly miserable, only around 260 compared to the baseline PowerMac 6100. (I think this is the result of QEMU not using the native floating point hardware? BasiliskII and I *believe* Sheepshaver have the option of doing that, which particularly in BasiliskII's case has the effect of producing benchmark-breaking scores.) According to the very limited tests I can run in MacBench 5 without a CD that only translates to B&W G3-ish territory. Perhaps there's some more aggressive emulation switches I can flip.

Also, have you had any luck with sound? Not getting a peep, looks like maybe I need to try an experimental build?

 

[Cory5412]

They're "ok" - it's slower than a real G4, on my i5-2400 it's probably comparable to a g3/300.

In MB4, with 6100 scores set at 100:

8600/300:

• CPU: 411
  
• FP: 750
  

G3/300:

• CPU: 999
  
• FP: 748
  

QEMU i5-2400

• CPU: 1053
  
• FP: 325
  

I also can't get sound to work, but I'm less worried about that. I was looking at it as a possible supplement to vtools and testing location, plus admin and monitoring from work where I can't very reasonably keep an os9/ppc mac. I also thought about building a set-up qemu disk image for vtools access.

The main thing that has me annoyed is mouse access is so weird. Plus I can't get command or modifier keys to work, so it's very cumbersome to use compared to the way I'd use a real Mac, and I'm not even really a Finder power user by any means.

 

[Gorgonops]

So it looks like your scores are pretty much in line with mine. (Or, at least roughly what I'd expect with the different host CPUs.)

I haven't played with it much at all yet, so I can't say if modifier keys, et al, work any better on OS X. I guess this is one of those dancing elephant situations; it's not notable so much for how well it dances but that it manages at all.

 

[nglevin]

There's a couple of places to point to about the software-based floating point implementation in QEMU. Past a certain point, old Emaculation threads go over much of the material rediscovered here about where QEMU is and where it needs to go. That was nearly two years ago.

The latest prerelease builds of QEMU 4.0 address accelerating Altivec instructions, maybe that touches on floating point as well. I'll have to ask around to see how to get that building from source, but I'm in the middle of starting something work-life related so that's not going quickly.

 

[jimjimx]

For what it’s worth, I have a mini that I put os 9 on the other day (from macos9lives, v8), but haven’t played with it too much.

Problems.

it will auto boot into the last selected os 10 ver. selected in the os 10 startup sys prefs (I have both 10.4 and 10.5).

always have to hold down option key to boot into 9.

audio is all up.

www, classilla works, but horribly frustrating slow.

protools gives me a DAE error that I cant find information for anywhere.

I know they’re still working on fixing it and making improvements to it, as well.

It's better, smaller / faster than the DT G3 I’ve been using.... but I’ll still keep it for now to make floppies for System 6 & 7.

 

[pcamen]

For what it’s worth, I have a mini that I put os 9 on the other day (from macos9lives, v8), but haven’t played with it too much.

For the unfamiliar ,  can you clarify what v8 means and what version mini?

 

[jimjimx]

Mac mini g4, 1.4 ghz, 1gb ram, and v8 is the un offical vers. of os 9 from “macos9lives” that runs on a mini.. They’re doing great work there. 

 

[jack]

The original iBook G4 (12" 800, 14" 933/1.0) will boot OS 9. No newer versions will.

I'm pretty sure there's no iBook G4 that will run OS 9, at least in a supported manner. I've got the original 12" 800 MHz, and I've managed to use the various "unsupported G4" install discs. You can get a booting system. But graphics drivers and wifi drivers are lacking, making the thing pretty unusable.

 

[Cory5412]

Good catch - I just looked again and I was wrong on that one. Because that post is in my own list of things to make into wiki pages or future info-pages, I'm going to make the update in that post, as well.

So, All iBook G3s will but no iBook G4s will.