Hi All!
Something occurred to me ages ago but thought I'd ask on here seeing as there are probably people on here who've been in the music game long enough to remember.
How did the Amiga 500 and the Atari ST computers back in the early 90's handle things as regards music sequencing with what would be a very low memory, ram and processing speeds in comparison to today? Just seen on Wikipedia the Amiga 500 had 7.16 Mhz processor, wow!
Thanks!
How Did Amigas and Atari ST's Handle Stuff?
-
Creativemind
- Posts: 4897
- Joined: 17 Jan 2015
- Location: Stoke-On-Trent, England, UK
Reason Studio's 11.3 / Cockos Reaper 6.82 / Cakewalk By Bandlab / Orion 8.6
http://soundcloud.com/creativemind75/iv ... soul-mix-3
Atari ST released in 1985, a big difference from 1990 which had computers hitting 66MHz and rapidly getting faster.
Keep in mind it doesn't take a whole lot of processing power to send MIDI messages and do some simple sequencing. An Arduino UNO, an 8-bit, chip can do it running at 16MHz. MIDI communicates at a baud rate of 31.250 kbps, for a chip running at 16MHz and doing not much else it's not even pushing the limits of the processor.
For an Atari with the Motorola 68000 CPU I doubt it was stressing it. Keep in mind that there wasn't multi-tasking, it did one thing at a time. No switching between huge browsers and DAWs and streaming music.
Keep in mind it doesn't take a whole lot of processing power to send MIDI messages and do some simple sequencing. An Arduino UNO, an 8-bit, chip can do it running at 16MHz. MIDI communicates at a baud rate of 31.250 kbps, for a chip running at 16MHz and doing not much else it's not even pushing the limits of the processor.
For an Atari with the Motorola 68000 CPU I doubt it was stressing it. Keep in mind that there wasn't multi-tasking, it did one thing at a time. No switching between huge browsers and DAWs and streaming music.
-
fieldframe
- RE Developer
- Posts: 1038
- Joined: 19 Apr 2016
Great question. 
I would say it was a combination of low-overhead resources, close-to-the-metal programming, and some specialized hardware.
On the resource side, as you may know, most of the music you would hear on the Amiga was created in trackers, in MOD format, which was kind of like a MIDI file with built-in samples. And these samples were tiny – you might have only one sample for an entire piano or synth lead, and it would be well below CD quality. Though of course, it's exactly the sound quality we try to get today when we use a bitcrusher!
On the programming side, operating systems on the Amiga and Atari were way less complex than modern ones. You had to do a lot more on your own to program for them, but you could also hyper-optimize your playback of tiny samples enough to play eight tracks of them on an 8 MHz processor.
On the hardware side, it wasn't actually that 8 MHz CPU doing the heavy lifting, but some custom hardware. On the Amiga, it was the famous Paula chip, so far ahead of its time that it took over a decade for the PC to catch up. Although once the PC did catch up, it had done so through sheer CPU power, which opened up the world of soft synths, leading to Rebirth and the world of synthesis we have today.
I would say it was a combination of low-overhead resources, close-to-the-metal programming, and some specialized hardware.
On the resource side, as you may know, most of the music you would hear on the Amiga was created in trackers, in MOD format, which was kind of like a MIDI file with built-in samples. And these samples were tiny – you might have only one sample for an entire piano or synth lead, and it would be well below CD quality. Though of course, it's exactly the sound quality we try to get today when we use a bitcrusher!
On the programming side, operating systems on the Amiga and Atari were way less complex than modern ones. You had to do a lot more on your own to program for them, but you could also hyper-optimize your playback of tiny samples enough to play eight tracks of them on an 8 MHz processor.
On the hardware side, it wasn't actually that 8 MHz CPU doing the heavy lifting, but some custom hardware. On the Amiga, it was the famous Paula chip, so far ahead of its time that it took over a decade for the PC to catch up. Although once the PC did catch up, it had done so through sheer CPU power, which opened up the world of soft synths, leading to Rebirth and the world of synthesis we have today.
-
adfielding
- Posts: 959
- Joined: 19 May 2015
- Contact:
It's also worth bearing in mind that, while the clock speed of both machines was relatively slow, they also had two factors working to their advantage:
First up - there was very minimal processor overhead in the form of background tasks. TOS with the Atari ST (and I'm assuming AmigaOS) weren't true multitasking operating systems back in the day, so if you were running an application then the entire computer's resources would be focused on that one application. Things get a little more confusing with things like desktop accessories on the Atari ST and MultiTOS and stuff like that, but in the early days they were very much focused on running one application at a time. It still amazes me that STs were often configured with 1mb of RAM, which is less than one instance of Notepad uses in Windows nowadays.
Also, both the Amiga and ST had much more specialised/esoteric hardware designed to perform certain tasks. I guess you could argue that the same is true now with video cards and stuff like that, but the ST came with MIDI ports built in. That was a pretty big deal back in the day. The STe also introduced a DMA chip that allowed for 50khz audio output which I totally didn't appreciate at the time but, in retrospect, was a pretty big deal. Amigas had a decent blitter from day one which lent itself well to gaming, something which wasn't introduced on the Atari side of things until the Mega ST and STe came out later.
Basically - they ran one thing at a time and had a bunch of specialised components designed to support the CPU. It's why games consoles back in the 16-bit era (and, to a lesser extent, 32-bit era) got away with low clock speeds - they had a bunch of components designed to handle specific/specialised tasks common to games of the time. If you're working with one (or a small number of) hardware sub-set then it makes it much easier to exploit the hardware on offer. For a more recent example - that's why games developers were able to push so much out of the Xbox 360 over its lengthy life span, despite it being based on hardware from 2005.
Anyway! That's my take on it. I'm not an expert, some of this stuff is probably inaccurate, but I hope it gives you a better idea as to why people were able to push more out of hardware at the time.
First up - there was very minimal processor overhead in the form of background tasks. TOS with the Atari ST (and I'm assuming AmigaOS) weren't true multitasking operating systems back in the day, so if you were running an application then the entire computer's resources would be focused on that one application. Things get a little more confusing with things like desktop accessories on the Atari ST and MultiTOS and stuff like that, but in the early days they were very much focused on running one application at a time. It still amazes me that STs were often configured with 1mb of RAM, which is less than one instance of Notepad uses in Windows nowadays.
Also, both the Amiga and ST had much more specialised/esoteric hardware designed to perform certain tasks. I guess you could argue that the same is true now with video cards and stuff like that, but the ST came with MIDI ports built in. That was a pretty big deal back in the day. The STe also introduced a DMA chip that allowed for 50khz audio output which I totally didn't appreciate at the time but, in retrospect, was a pretty big deal. Amigas had a decent blitter from day one which lent itself well to gaming, something which wasn't introduced on the Atari side of things until the Mega ST and STe came out later.
Basically - they ran one thing at a time and had a bunch of specialised components designed to support the CPU. It's why games consoles back in the 16-bit era (and, to a lesser extent, 32-bit era) got away with low clock speeds - they had a bunch of components designed to handle specific/specialised tasks common to games of the time. If you're working with one (or a small number of) hardware sub-set then it makes it much easier to exploit the hardware on offer. For a more recent example - that's why games developers were able to push so much out of the Xbox 360 over its lengthy life span, despite it being based on hardware from 2005.
Anyway! That's my take on it. I'm not an expert, some of this stuff is probably inaccurate, but I hope it gives you a better idea as to why people were able to push more out of hardware at the time.
-
Some Desperate Glory
- Posts: 171
- Joined: 15 Jan 2015
- Location: San Francisco
Not to be pedantic, but AmigaOS was the first consumer operating system with preemptive multitasking (even MacOS didn't have it). It was one of the reasons there were so many crashes (Guru Meditations). AmigaOS wasn't great about protecting memory spaces and left it largely for the applications to play nicely, which many of them did not, so it wasn't uncommon for application A to write data to application B's memory space which invariably led to a crash. What you wrote about the blitter was correct, and the key reason Amiga ran circles around the competition for games (at least until the competition caught up as the Amiga stood still).adfielding wrote:It's also worth bearing in mind that, while the clock speed of both machines was relatively slow, they also had two factors working to their advantage:
First up - there was very minimal processor overhead in the form of background tasks. TOS with the Atari ST (and I'm assuming AmigaOS) weren't true multitasking operating systems back in the day, so if you were running an application then the entire computer's resources would be focused on that one application.
Back to music, I think the key reason computers in the 80s could "handle it" is that unlike the DAWs of today the sequencers of the 80s and early 90s primarily just dealt with the notes and didn't do any audio tasks. When I was writing music in the early 90s, I would record vocals on a 4-track tape deck with one of the tracks printed with a synchronizing time code. The sequencer on my Amiga (and them Mac) then played various hardware synths and samplers to generate the music. I would then mix this all down (a few synths and three tracks of vocals) to a DAT tape for my master.
As Adam said, the Atari had a built in MIDI port which was huge and why it dominated the electronic music scene for years. While you could buy external MIDI ports for the Amiga, the Amiga didn't had hardware buffering on their peripherals and therefore the timing was kind of dodgy (until more expensive interfaces came out). This is the main reason I moved to Macs for my music).
For computer-generated music (like Amiga tracker pieces) it was a bit different, typically only a small number of very small samples were kept in hardware and very creatively sequenced. You really couldn't fit very much audio into the RAM of a stock Amiga 500 (512 kB RAM) so people didn't try (for the most part).
Still nostalgic about the old days, writing songs with my Amiga 500, Korg M1, and Ensoniq ASR-10 sampler.
Just a small note to add to the other replies. The Atari ST did actually crash regularly. The MIDI timing was very stable but the software was still prone to crashing. There's a bit of a tendency for people to glorify how good the Atari's timing was but forgetting how slow and crash prone they were.
-
HepCat
Fascinating responses.
I used to make music on Amiga trackers - first Soundtracker, then Noisetracker, Games Music Creator and MED, then l went 8-track and got 8 Channel Soundtracker and Octalyzer, and eventually OctaMED when l got it on an Amiga Format cover disk.
As others said: the Amiga A500 was less stable than computers today - l lost a handful of songs via crashes (Guru Meditations). Mind you, l also lost a fair few songs by absentmindedly writing over existing songs on floppy disk.
The "Voices 8" 8 Channel Soundtracker by Shining, gave decent sound quality (compared to Octalyzer, although Octalyzer offered slightly more functionality re: sample editing).
One thing l noticed was if you played a pattern (= a page of manually typed music data) really fast, the sound quality actually felt improved, l mean, almost going on CD quality. Must be some kinda auditory illusion?
Was about to get about 10 songs max., on a 1Mb floppy disk (although l think the floppy disks were actually more like 1.4Mb?).
[FYI, l had an Amiga A500 eventually with 512kb upgrade to give 1Mb RAM.]
I used to make music on Amiga trackers - first Soundtracker, then Noisetracker, Games Music Creator and MED, then l went 8-track and got 8 Channel Soundtracker and Octalyzer, and eventually OctaMED when l got it on an Amiga Format cover disk.
As others said: the Amiga A500 was less stable than computers today - l lost a handful of songs via crashes (Guru Meditations). Mind you, l also lost a fair few songs by absentmindedly writing over existing songs on floppy disk.
The "Voices 8" 8 Channel Soundtracker by Shining, gave decent sound quality (compared to Octalyzer, although Octalyzer offered slightly more functionality re: sample editing).
One thing l noticed was if you played a pattern (= a page of manually typed music data) really fast, the sound quality actually felt improved, l mean, almost going on CD quality. Must be some kinda auditory illusion?
Was about to get about 10 songs max., on a 1Mb floppy disk (although l think the floppy disks were actually more like 1.4Mb?).
[FYI, l had an Amiga A500 eventually with 512kb upgrade to give 1Mb RAM.]
-
Sinistereo
- Posts: 128
- Joined: 15 Jan 2015
I can't speak for Atari, but the Amiga had dedicated chips for handling video and audio. The audio chip ("Paula") had four 8-bit PCM audio generators (2 left, 2 right) that had partially asynchronous access to dedicated RAM. So the Motorola 68000 wasn't doing all of the heavy lifting for either audio or video.
AmigaOS was indeed revolutionary, as was the Amiga architecture. Full NTSC and PAL video generation with genlock and frame-synced audio was previously unavailable outside of dedicated broadcast rigs. This is why the Amiga became the de facto standard for video and digital art for many years... until Mac caught up and Apple's better-managed business (Commodore was a nightmare!) took over that market segment.
AmigaOS was indeed revolutionary, as was the Amiga architecture. Full NTSC and PAL video generation with genlock and frame-synced audio was previously unavailable outside of dedicated broadcast rigs. This is why the Amiga became the de facto standard for video and digital art for many years... until Mac caught up and Apple's better-managed business (Commodore was a nightmare!) took over that market segment.
-
WongoTheSane
- Moderator
- Posts: 1851
- Joined: 14 Sep 2015
- Location: Paris, France
I agree with dioxide, crashing is the main thing I remember from all that era.
Regarding raw CPU power, as has been noted, handling MIDI is an easy task and you would run out of hardware synths long before you ran out of tracks/memory/cpu.
Fun fact (in hindsight): the Atari didn't work with most MIDI cables (the cheap ones) which had 4 of the pins soldered by pairs (1-4 and 3-5 IIRC). You had to unsolder them before use. A soldering iron was an indispensable tool for a musician at that time.
Regarding raw CPU power, as has been noted, handling MIDI is an easy task and you would run out of hardware synths long before you ran out of tracks/memory/cpu.
Fun fact (in hindsight): the Atari didn't work with most MIDI cables (the cheap ones) which had 4 of the pins soldered by pairs (1-4 and 3-5 IIRC). You had to unsolder them before use. A soldering iron was an indispensable tool for a musician at that time.
Aw man this thread makes me want to have an Amiga again.
It's worth pointing out that Amigas had 4x 8-bit PCM audio channels (2 channels per L & R output) at the ready which could run independent sample rates. The biggest limitation was that the Paula audio chip was only able to access the system RAM (AKA Chip RAM or Graphics Mem: 512k to 2Mb, dependent on the model) and not any RAM expansions. However in the latter years of the Amiga's lifespan some clever programmers figured out how to stream data from other memory to the Chip RAM in realtime and bypassing the limitation.
It was also possible to push the Amiga's output to 14-bit by exploiting the two channels per output via some clever programming, and mix a theoretically unlimited number of audio channels at the cost of additional CPU overhead. In a nutshell, the better the CPU and RAM the Amiga had, the more you could do with its audio.
It's worth pointing out that Amigas had 4x 8-bit PCM audio channels (2 channels per L & R output) at the ready which could run independent sample rates. The biggest limitation was that the Paula audio chip was only able to access the system RAM (AKA Chip RAM or Graphics Mem: 512k to 2Mb, dependent on the model) and not any RAM expansions. However in the latter years of the Amiga's lifespan some clever programmers figured out how to stream data from other memory to the Chip RAM in realtime and bypassing the limitation.
It was also possible to push the Amiga's output to 14-bit by exploiting the two channels per output via some clever programming, and mix a theoretically unlimited number of audio channels at the cost of additional CPU overhead. In a nutshell, the better the CPU and RAM the Amiga had, the more you could do with its audio.
Last edited by Aquila on 14 Dec 2016, edited 1 time in total.
Fun fact: to keep file sizes small, most Amiga music used short samples at extremely low sample rates. This of course introduced considerable digital aliasing, also known as Ring. The effects of which could be reduced (but not eliminated) by a global 12db low pass filter over the output at the cost of a lower fidelity sound. These artifacts are a major player in defining the good old 'Amiga Sound'.
People either loved or hated the Ring sound. When I was younger I was impervious to it, but listening to Amiga music now (I'm watching some old Demos on Youtube as I type this) I can clearly hear the aliasing on the samples.
People either loved or hated the Ring sound. When I was younger I was impervious to it, but listening to Amiga music now (I'm watching some old Demos on Youtube as I type this) I can clearly hear the aliasing on the samples.
Everything worked so well on the Amiga (with stuff like Protracker) because those programs were written and ran in an extremely low level language. Nowadays you have a whole bunch of layers in your operating system and everything has to pass through those layers. That's also where those weird occasional random errors come from, something goes wrong with passing from one layer to another.Creativemind wrote:Hi All!
Something occurred to me ages ago but thought I'd ask on here seeing as there are probably people on here who've been in the music game long enough to remember.
How did the Amiga 500 and the Atari ST computers back in the early 90's handle things as regards music sequencing with what would be a very low memory, ram and processing speeds in comparison to today? Just seen on Wikipedia the Amiga 500 had 7.16 Mhz processor, wow!
Thanks!
No such thing in Protracker, you worked pretty much directly on the hardware. The amount of samples you could store was depending on the amount of RAM in the Amiga and well, to be able to work somewhat you had to have at least a 512Kb expansion, otherwise it was pretty much unworkable.
I made a bunch of tracks on them but lost most of them due to a harddrive crash, after I painstakingly converted all my Amiga disks to DOS disks and stored my tracks on that harddrive, where I thought they would be safer. How wrong I was.
I have a nice collection of Amiga's now, one 600HD that I still need to repair and a bunch of A500's.
"This is a block of text that can be added to posts you make. There is a 255 character limit."
Amiga had a dedicated sound and graphics chip, that reduced the CPU overhead. Took ages for other systems to come along with it. 4 direct audio channels playing samples and modulation through a chip, so nearly no CPU overhead. Other trackers having more channels needed to calculate it with the CPU that results in a high CPU load. With very small sample you had a kind of graintable synth.
And dont forget, most trackers were coded in assembler that also reduced the amount of code to execute. Today a simple function call produces so much overhead, that could be compared to a play routine of that time...
And dont forget, most trackers were coded in assembler that also reduced the amount of code to execute. Today a simple function call produces so much overhead, that could be compared to a play routine of that time...
Reason12, Win10
I read Calvin Harris produced his first album on an Amiga running a tracker. I'm quite surprised that the Amiga sound chip was 8 bit and less than CD quality. These days even a smartphone app can produce higher quality music with far more tracks and few limitations.
There are a couple of old fanatics still running old Atari STs, apparently they say the midi timing for sequencing external hardware is unmatched and better than modern computers.
There are a couple of old fanatics still running old Atari STs, apparently they say the midi timing for sequencing external hardware is unmatched and better than modern computers.
Timing wise, your windows PC has hundreds of threads running at a time (not sure how many are sleeping), and so it's harder to get precise timing because windows doesn't really know who is best to prioritise for a specific time period.
I've written a basic sequencer for an 8-bit CPU running at 4Mhz, and because it was in assembler I knew exactly how many cpu cycles everything would take - though I also had the option of writing in C and viewing the assembly the compiler generated.
I was accessing hardware directly, so no operating system calls, so timing couldn't be more precise.
I've written a basic sequencer for an 8-bit CPU running at 4Mhz, and because it was in assembler I knew exactly how many cpu cycles everything would take - though I also had the option of writing in C and viewing the assembly the compiler generated.
I was accessing hardware directly, so no operating system calls, so timing couldn't be more precise.
Last edited by avasopht on 15 Dec 2016, edited 1 time in total.
-
HepCat
I came back to lol @ the lower amount of love for the Atari compared to the love for the Amiga that still persists, however:
@ 2:36 - wow that is one funky pitchbend wheel scratch. I will have to use that technique myself one day.
Anyway, not sure what an Atari laptop is, and he doesn't really explain what he's doing. I mean, he's using an Atari, it's a laptop, and it's got flashing lights on the screen, and he's making music, with his keyboard and .... wait, how does it all knit together?
Great to listen to the guy talk anyway.
@ 01:22 - why do you need a Nintendo?EdGrip wrote:There's some Atari Stacy laptop action going on here (still)
@ 2:36 - wow that is one funky pitchbend wheel scratch. I will have to use that technique myself one day.
Anyway, not sure what an Atari laptop is, and he doesn't really explain what he's doing. I mean, he's using an Atari, it's a laptop, and it's got flashing lights on the screen, and he's making music, with his keyboard and .... wait, how does it all knit together?
Great to listen to the guy talk anyway.
-
househoppin09
- Posts: 536
- Joined: 03 Aug 2016
He's using the Nintendo to get those delightful bleep-y sounds. Nothing else sounds quite the same...
The Atari laptop appears to be being used as the main sequencer for the songs. How it all "knits together" is that the laptop is sending out the MIDI data, and then his synths (of which the Nintendo is one) are responding to the MIDI signals and playing the audio of their parts--it's just the standard studio arrangement, except using a very old computer to do the sequencing instead of a modern DAW.
The Atari laptop appears to be being used as the main sequencer for the songs. How it all "knits together" is that the laptop is sending out the MIDI data, and then his synths (of which the Nintendo is one) are responding to the MIDI signals and playing the audio of their parts--it's just the standard studio arrangement, except using a very old computer to do the sequencing instead of a modern DAW.
Gosh this thread takes me back.
I never used my Amiga 500 for music - but I did use an Atari STe, running C-Lab Creator (which ultimately become Apple Logic). The ST had built in MIDI ports so it was great to use as a music computer.
Prior to that generation of computers, the Commodore 64 was great - with the amazing SID chip, early sample playback (Ghostbusters, Impossible Mission), and great music software. I had a little keyboard overlay for it - like this:
My first use of computers for music though was a Yamaha CX-5 Music Computer. It has a Z80 CPU, and a built in hardware 8 voice 4-operator FM synth. The music was all entered step-time from a piano keyboard with music notation
I never used my Amiga 500 for music - but I did use an Atari STe, running C-Lab Creator (which ultimately become Apple Logic). The ST had built in MIDI ports so it was great to use as a music computer.
Prior to that generation of computers, the Commodore 64 was great - with the amazing SID chip, early sample playback (Ghostbusters, Impossible Mission), and great music software. I had a little keyboard overlay for it - like this:
My first use of computers for music though was a Yamaha CX-5 Music Computer. It has a Z80 CPU, and a built in hardware 8 voice 4-operator FM synth. The music was all entered step-time from a piano keyboard with music notation
-
- Information
-
Who is online
Users browsing this forum: No registered users and 20 guests
Share and download patches!
Join ReasonTalk on Slack!