Working from home at 25MHz: You could do worse than a Quadra 700
arstechnica.com
arstechnica.com
I too fell for this numerous times. In fact I actually went back to an Acorn RiscPC 600 for a bit once instead of a modern windows machine. It was what I considered to peak productivity for me. Built in assembler and programming language, decent quality desktop applications and no distractions.
After about 5-6 days it was back on eBay because quite frankly I realised what a pain in the ass it really was. 99% of the stuff I was actually doing was collaboration and it's really difficult doing that on vintage computing platforms. On top of that the hardware is almost always like a hand grenade with the pin out. It's going to die, but you just don't know when.
I cite this as an example for the comedy timing: https://www.youtube.com/watch?v=TU55-7dWMi0
Now emulation I can get behind because when you are completely fed up of the idea you had it's easier to dispose of it than getting rid of it on ebay :)
> the machine will always be hobbled by its technical limitations.
> I would never go as far to say that the Quadra 700 should have a place on your desk
Maybe it is, maybe it isn't. Are we talking Zoom or IRC here?
If you need to do the sort of work where you just need to concentrate with no distractions, a "vintage" platform is a superior choice. A few prominent authors swear by their obsolete word processors for example. Many people (myself included) find it hard to focus and resist procrastination on any modern Internet-connected system. But I can get my head down in First Word Plus on my Atari ST (or Word 5.1 on a classic Mac) and upload the file later for formatting in Word and sharing by whatever means.
Much as I have fond memories of IRC, collaboration requires the other people to use the same tools as you, and you don’t always get to choose what that is.
Indeed - my point being that whether a vintage platform is viable or even superior for your workflow is very situation-dependent, it's not "all or nothing" nor "one size fits all".
Hell I am reading HN now when there is something I could be working on. I should go into the other room and boot up the ST.
Makes sense. Other than perhaps latency, if software was truly better in previous decades, you can just run old software on new hardware [1] and get the best of both worlds. Given infinite time, you could probably find a way to boot directly into some kind of hypervisor that runs RiscOS or whatever full-screen. While also running LVM underneath it for online backups. Oops, a new software feature.
It's related to my pet peeve about Gemini: You can just choose to do all the stuff Gemini does with your own safe subset of HTTPS. If you have the political power and dev hours to maintain Gemini, you can also just maintain HTTP/1.1 with TLS.
Running Acorn hardware won't fix software bloat created by anyone but yourself, and running Gemini won't fix HTTP bloat created by anyone but yourself. Might as well run good software on good hardware, instead of running bad protocols or bad hardware out of spite.
[1] I'm on the young end of millennials, so my nostalgia is for Windows XP. But there's a valley for retro nostalgia - It's okay to use modern Linux, because you need security patches. It's okay to use Windows 10 because your employer forces it. It's okay to run an Amiga or RiscOS ironically, because it's cool and funny. But Windows XP? No! Evil security hazard! Deprecation means the software became bad overnight! Why won't you upgrade??
No, but it does help building a community of like-minded people. And that is really the core idea of gemini, isolationism is sort of a feature there. I see gemini more like a flag under which people can rally under, the technical aspects are incidental.
(note: I don't use/run gemini so I might be mis-representing them here)
Of course, it would be great to have the Risc OS experience while doing modern computing tasks, so I'm pleased to hear that the lessons it brought to the world have not been forgotten.
The first is that XP has a tendency to, by default, open a lot of forward facing holes and services that to my knowledge most other "old timey" OSes don't. I'm not familiar with riscos or amigaos, but I do know old versions of OS X ship with basically no attack surface, while XP (IIRC) will enable SMB and guest access automatically.
The second is that unlike those other OSes, XP was and is very widely used, and there are exploits which target XP and a range of other commonly used Windows machines. Unlike basically any other niche retro hardware, Windows is both 1) pervasively networked and 2) very widely attacked, and lacks basically any modern mitigation for exploits, so a simple buffer overflow in a networked service can easily become a RCE.
Attaboy. I went the other way and realized that I wanted to buy some very modestly priced emulator-on-a-SoC for an old gaming console near and dear to my heart. That gave me the best of both worlds -- form factor and tactility of the real console with cost, speed, battery life of a present day android device. I must say, for a $40 Android device, I use it a lot more than I thought I would.
I can't help but be curious: What are you talking about?
It ran System 7.1. It was a beautiful machine but I couldn't get any real work done on it -- the browser (IE for Mac) was too slow, and ClarisWorks was too simple.
I believe there are pivot points in computer technology. A machine just after a pivot will last a long time, while a machine just one year earlier will age badly.
My main desktop was a Dell Inspiron 530 Core 2 Duo from 2005-2020. The Core 2 Duo was a long lasting chip, and I could watch YouTube, browse the web on the latest Firefox, etc. with no problems at all the way up to the start of the pandemic this year (I recently upgraded to a 2014 Dell i7, which I'll likely keep for another 10 years). If I had a Core Solo, I'd likely have dumped it.
I eventually sold the Quadra (and the Mac II, Mac SE, the Sun SPARCstation 1 pizzabox, and all the other vintage machines I had in my college apartment). They were fun to own for a while, but ultimately impractical to keep around.
I do note that there was never a Core Solo desktop CPU. There were handful of single-core Conroe Celerons released, but all main "Core" series parts were at least dual-core (i.e. Core 2 Duo or Core 2 Quad)
I have the spiritual successor, the M1 mini now.
* https://ark.intel.com/content/www/us/en/ark/compare.html?pro... * https://ark.intel.com/content/www/us/en/ark/compare.html?pro...
https://everymac.com/systems/apple/mac_mini/specs/mac_mini_c...
It's a nice boost for certain workloads that thread well, but it's not a huge boost for single threaded workloads except for certain specialised stuff where there are newer instructions to take advantage of.
Contrast that to a roughly comparable time frame: 486sx 60Mhz -> Athlon 1Ghz.
Short of a major revolution, I expect this CPU/Memory will last me just as long.
Of course at the time almost no one had really seen video on a personal computer so to the vast bulk of the movie going audience it was just futuristic.
Also in order to film that scene the video card had to be slowed down to 24 frames/sec and synced to the camera, looked great on film but was probably headache inducing to the people present
(running at 3x would likely have resulted in banding)
We had done a lot of early video-work prior to quicktime actually being released, we had hacked together a motion-JPEG card as a daughter card on one of our unaccelerated cards that could play video in a window, I suspect this was based on that (but could be wrong, it was a long time ago!).
We certainly hacked on the drivers to slow down the refresh rate, but that was just a matter of changing some numbers in the video timing controller - in a time prior to multi-sync displays it may have involved getting a bespoke display manufactured.
I don't know of any archive, here's a few stories:
- when we first went to Sony to buy 19 inch monitors we asked them for 15k/year, they laughed at us, they had never ever made that many, unlike TVs 19 inch monitors were made by hand on a single manufacturing line, the didn't believe there was a market for that many , in the end we sold 25k+
- when we first sold monitors to Australia/NZ they were crap, people returned them, turns out all monitors were aligned in Tokyo facing east .... southern hemisphere ones needed to be aligned in a special magnetic faraday cage
- we first demoed quicktime in a window at MacWorld on an HDTV monitor, one of only 5 in existence in the US at the time - most people had never seen video in a window, or a screen that big before - the monitor was so heavy it required a forklift
- Premiere was written by a Supermac employee, partly in his spare time, to work with our video capture card, we sold it to Adobe, he quit and followed it very soon after
- one April fools the boss carefully locked his door and set traps (people went in thru the false ceiling) next morning he started work, his screen slowly got greener and greener, he rebooted his machine, it happened again, he started removing suspicious inits from his system folder, rebooted again, it did it again .... just before he wiped his harddrive and did a full install people offered to put his original video card's rom back in ....
I doubt it. It probably just looked like film, no?
Remember that when film plays back the projector shows you the same image for most of the 1/24 of the time before switching to the next frame
Filming monitors and TVs was and is a very common thing. You can probably get away with filming a 30 fps monitor if your shutter speed is tight enough and you're ok with a few pulldown frames sneaking in.
>Remember that when film plays back the projector shows you the same image for most of the 1/24 of the time before switching to the next frame
Is this true? Isn't it 50/50 or so? Take a look at a rotary disk shutter. Film is pretty delicate and you can't pull it through that fast.
Don't Copy That Floppy on a LCIII via a Nautilus subscription CD.
It was like a minor version of what I suspect the moon landing felt like. I watched with my jaw down thinking about the day all video is on computer.
Boots in literally seconds, cold boot is longer due to memory check.
The computer you grew up with couldn't deal with taking, manipulating and storing photos of your phone's camera(s).
Sure, your phone could boot faster, but you also need to compare to the capabilities of both machines. Plus there's no need to reboot your phone on a daily basis anyway.
> the computer you grew up with couldn't even handle the colour palette and resolution of the screen that you now carry around in your pocket
the other way of looking at this is that the phone is enormously more powerful, can render hi-res full colour depth 3D scenes at 60fps that would take minutes per frame on a 90s computer, can store and transfer amounts of data that would have been unimaginable back then
so why is it so slow to boot up?
A computer from the 80s probably rebooted multiple times a day.
How much effort is someone going to put into an event that happens once a month vs investing that same time into something like optimizing battery life or taking better pictures?
Also, in my personal experience with Android/Linux mobile devices, phones in 2020 boot significantly faster than phones in 2010.
Let's say you bought a 100-baseT network card in 1995. The features were fixed in metal, and so were the bugs. Even if a new revision of the board or ROM came out, you still had the exact same card. If a flaw was found in one of the chips they used, chances are someone would need to program around it. Settings were probably done once using DIP switches and never touched again. You probably never even knew if a new revision came out.
The wifi card you use today is going to boot up in a pretty dumb state. It'll identify itself on the bus and not much else. Your computer's kernel is going to poll every device on the bus to see who's out there. The wifi card driver is going to match the ID of the wifi card to one it knows and then upload firmware. Then the card will need to briefly reboot into the new firmware and identify itself to the driver again. Then the driver is going to configure it.
It may take a couple of seconds for all this to happen, and it needs to happen for almost everything. Even your CPU which booted the machine in the first place might get patched microcode uploaded when the machine boots.
Nothing is just a device anymore, and nothing is just a device driver.
Ultimately, though, the device enumeration and initialization is a tiny fraction of the startup time on most computers, be they desktops, laptops, servers, phones, tablets, or what have you. Usually the actual source of long startup times is the incessant need for these machines to spin up oodles and oodles of background services doing who-knows-what, and it's remarkable how much faster a machine boots when these services are pared down to more reasonable minima. And worse, it's these background services that often make modern computers feel so slow even after they're booted up.
I worked with (3D) graphics and computation intensive processing as well as modelling packages and the experience was horrible compared to today.
I have particularly fond memories of the workflow we used for a film project back in the late 90s. The lab only had so many editing stations and they were shared across multiple groups. So for each editing session we first had to copy the project from multiple CDRs onto the editing machine (which took ages on the 4x SCSI CD drives), edit, then burn the project to a set of CDs - performing various rituals to ensure the ROM image wouldn't be corrupted by a butterfly coughing in the corner or something - clean up the HDD so the next group could get working and hope the CDRs were even readable.
The editing experience itself wasn't great either: forget about real time scrubbing or full resolution previews. Today you can edit films on a phone much quicker, more comfortably and with significantly higher quality. No disappointment there for sure.
Same goes for 3D modelling packages, rendering and just plain number crunching in Mathematica, Maple or MATLAB.
Word processing is the one area that simply is so primitive by its very nature, that it's trivial to reach peak-efficiency without throwing tons of compute power and tech in general at it. If word processing is all you do, you don't need more than a glorified electric typewriter.
For many, many other applications - and yes, that even includes just taking and sharing pictures in real time - today's technology is vastly superior to anything the 80s and 90s had to offer.
But yes, it is impressive that my $500 phone blows away the $50,000 workstations I started on.
IMO this is the only vitally important point. If my phone’s screen was 640x480 and 8-bit colour, it would still be fantastically useful.
The fact that 4G cellular internet is higher bandwidth than the system bus on my first Mac (Performa 5200), the fact that every image the camera takes us too large to fit into that Performa’s RAM even when it was fully upgraded, the fact that a single Netflix film takes up more storage on my phone than my Performa had in total, is an impressive feat; but I use most of the impressive technical performance to share dumb photos of weird cooking experiments [0], for the purpose of social connection with distant friends, and I could manage those connections in other ways if the tech was not there.
In practical terms, what I care about is the instantaneous responsiveness — the fact that the screen lights up when I lift the device, the way the system unlocks with facial recognition so I don’t have to enter a passcode, the way most apps are about a second away from use, that they auto-save, that crashes are rarer.
And that is because, as you say, there is no need to boot up the phone whenever you want to use it.
[0] e.g. https://travellingcurious.wordpress.com/2020/12/05/greenies-...
I'm also not quite sure about the capabilities. A 2014 $150 entry-level phone like the Motorola MotoG G4 has eMMC controller, SD-card interface, two separate camera systems, WiFi, BT, 4G modem, GNSS (GPS/Beidu/GLONASS), USB 2.0 (client), USB OTG (host), accelerometer, gyro, proximity sensor, compass, FM-radio plus the usual stuff (DRAM controller, GPU, audio subsystem, battery controller, etc.).
That's a lot of components that the OS needs to initialise and load drivers for on boot. And all that needs to happen on 1 GB 533MHz LPDDR2 RAM and 4 Cortex A7 cores at 1.2 GHz.
Just to give a point of reference: that's less computing power than a $35 Raspberry Pi 4 and significantly slower memory and storage.
A 10 year old laptop would likely feature a dual core Intel Core i3, i5, or Core i7 with 35W up to 75W TDP clocked at between 2.5GHz and 2.9GHz, dual channel DDR3 memory running at ~530MHz with a 4x multipler (i.e. 2133MHz effective vs 1066MHz on the phone) and your BIOS is most likely set to "fast boot", meaning it'll skip 90% of the hardware initialisation anyway. I also doubt your 10 year old laptop would last that long on a 2070 mAh battery (my own 9 year old model has a ~44000 mAh battery - just for reference).
TL;DR yes, even a vintage muscle car is still faster at 0-100 (or 0-60 if you prefer) than a 5 year old compact car. That shouldn't come as a surprise if you compare the specs.
No login, no passwords.
There are apparently some DNA sequencer machines that only interface with older macs. The linked article is just as an 'exhibit A'... It doesn't go into the technical details as to why that is so. I can't find the article on why virtualization is difficult for this case but it was very interesting. Cooperative multitasking (mac system 5, system 6, system 7 ish era) meant that the running task had the responsibility to yield. If the task was something like 'sample the value of x instrument register every 10 ms' (or something like that) then it could do the quick maths based on the base system clock rate and know exactly how many cycles to sleep between each sampling. In other multitasking scenarios where the os has it's own things it wants to do, the program cannot deterministically know how many cycles to sleep. What if the process will be arbitrarily delayed or interrupted, so some samples are taken 17 clocks apart and some samples are 19? With the priority being JITTER FREE, it doesn't help if the os thinks, ahh i have a million clock cycles this second, and only 1000 things to do, I'll just do some housekeeping... so there are background processes using bus resources, network interrupts, etcetera, all batched up and dispatched because the os thinks it has time to procrastinate this one task that just wants to sleep and sample a register and sleep... but sleeping for EXACT duration.
As another option, it would be nice if modern OS's could dedicate cores on multi-core systems to specific real-time processes. Then the OS can still do whatever it needs, while the real-time process gets guaranteed CPU treatment.
Preemption is evil in a multicore world. Still necessary in spots, but evil nonetheless.
I used to work in a Human Genome Project lab. Two of the 1990s workhorse DNA sequencing machines, the Applied Biosystems models 373 & 377, connected to Mac machines of similar vintage (my recollection is we had a lot of Power Mac 7100s, or at least machines in the 7xxx form factor).
IIRC the sequencing instrument had an interface to plug into the LocalTalk port on the Mac, but I might be wrong on this, it's been 20+ years. And the software needed the classic MacOS.
The ABI 373 and 377 are so old and obsolete at this point that you might find one in your local science museum. I know my city's science museum has a display with an ABI 373 from my old lab.
ABI's successor Sanger sequencing machines, models 3700 (1998) and 3730 (2003), had Windows-based companion computers and connected to the instrument via Ethernet port, so you didn't have as much lock-in to super old hardware.
really? there's email clients for MacOS 8.1 that can understand IMAP and SMTP over TLS1.2? Because otherwise you'll be using some old thing that only supports plaintext authentication.
the most secure way to do email on such a thing would probably be to ssh to another local machine on your LAN and run a terminal based email client from that.
I suspect that, like other pre-SysVR4/Solaris-y unixes, compiling anything newer than GCC 2.95-ish will be pretty much impossible. But, depending on your interest/curiosity, it looks like there's some fun stuff to play with.
You would invoke Commando from the shell for a command and it would bring up a dialog to help you construct the command line, which would be written back to the shell.
There's a screen shot of it for ls here [1], and mount here [2].
[1] http://toastytech.com/guis/aux3.html
[2] https://retrocomputingforum.com/t/running-a-ux-apples-at-t-u...
(Also, OS/400 commands all follow a very strict naming convention, such as the example WrkActJob in the link. The consistency of the words used and the abbreviations for those words were intended to make learning easier.)
You see a bit of this today in PowerShell's command auto-completion for built-in commands, PowerShell script files, and cmdlets (user-defined commands written as PowerShell script functions or .NET CLI methods). The graphical PowerShell Integrated Scripting Environment (ISE) even pops up windows that let you interactively fill out forms to enter command options, just like Commando: https://blog.netwrix.com/2018/02/21/windows-powershell-scrip...
Regrettably, PowerShell ISE is deprecated, and the recommended replacement of Visual Studio Code with the PowerShell extension isn't an exact one.
Original Macintosh software like WriteNow feels really fast on it on System 7.
> on System 7
Dual boot, or VM?
EDIT: Yep,
> The Booter also will certainly fail if Virtual Memory is enabled, so turn that off too while you're snooping around in the Memory control panel.
- http://www.netbsd.org/ports/mac68k/booter-manual/index.html
So terribly true. Feel the same way about my old 14 MHz Amiga.
Just saying.
I still have the Powerbook, but it won't boot. I regret ever selling my PowerMac 9500 though.
I also regret getting rid of my PowerMac 7600. Still have my SE/30 however! And a network card I have been meaning to install in it but can’t find the right screwdriver. With that I think I could do some “real work” on it.
Can you imagine if they tried something like that today? The cheese grater will not even turn on with the case removed
The early PPC Macs were good machines, and the brief flirtation with licensed clones at least allowed me to buy a Mac-compatible computer (Power 100) that I wouldn’t otherwise have been able to afford.
The G4 was amazing, but right from the start Motorola couldn't get acceptable yields on the higher-clock versions, and that kept happening. The IBM-sourced G5 was a power hog that was clearly not meant to be a consumer CPU.
The first computer I owned was a G4 Power Mac, I ordered with the student discount the day after it was announced. I got the middle 400MHz version. Apple was forced to abandon the high-end 450MHz version shortly after launch.
If Motorola could have manufactured the G4 according to plan, Apple probably never would have switched to Intel.
But regardless, that entire pre-G3 era was the nadir of the platform. Those were dark days.
And it screamed. Faster than anything before it.
The fact that emulated 68k code could call native PPC was nuts to me but it was how most of the classic OS worked on PPC, with native parts being slowly rewritten over time.
The Intel move still made lots of sense, but that didn't make PPC unuseable.
Had these phones never existed, I'd get around to the Amiga that's in the garage.
Sigh.
We weren't allowed to touch the computer or even get near it, and at the time that really pissed me off, but as an adult I totally get it now.
I had to first boot MacOS6, and then jump to Linux via some kind of loadlin for Mac.
That was around 2000.