@Karlos

Quote:

Karlos wrote: @you will never guess who Quote: The Pentium FDIV bug needs corrective design as a new stepping e.g. the FDIV bug affects the 60 and 66 MHz Pentium P5 in stepping levels prior to D1, and the 75, 90, and 100 MHz Pentium P54C in steppings prior to B5. The 120 MHz P54C and P54CQS CPUs are unaffected. On the subject of how the Amiga audio could be better, this insightful pearl of wisdom could not be more relavent. No wait, that other thing. Irrelevant. Anyway, in case you guys didn't hear it.... Filters. That's what we lacked. I've had the pleasure of playing Paula through an external filter, controlled via MIDI, and while the result is technically paraphonic (4 voices, one filter), the result is massive, especially when using all 4 channels like a single stacked oscillator. A fatter sound has never been heard.

"Paula" is tied to graphics chipset improvements due to the shared Chip RAM and Agnus's DMA controller. Amiga's hardware sound and graphics architecture wasn't partitioned e.g. ECS double scan mode increased Paula DMA audio to 56 kHz.

The whole point why I purchased ECS Denise is for Paula's 56 kHz capability while I have an RTG display.

External MIDI is useless for the game's majority use cases. Atari Falcon's 13,000 to 14,000 unit sales show "must include MIDI" argument as a distraction.

PS, I have a Yamaha MU80 external XG MIDI box and a Yamaha XG MIDI (via USB connection) keyboard.

Last edited by Hammer on 31-Jul-2024 at 06:05 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@cdimauro

Quote:

And you should know that engineers aren't there left alone: there's always a TECHNICAL lead that they belong to and which decides which direction to take.

The technical leadership is weak i.e. considerably weaker than Bill Gates or Elon Musk examples.

The "Commodore The Final Years" book by Brian Bagnall should be a prerequisite subject matter for this topic.

If copyright wasn't restrictive, I would have posted entire chapters from Brian Bagnall's book.

Quote:

You see yourself: NEW revisions -> NEW chips... Quote:

That's your outsider assertion, not Commodore engineers'.

Quote:

I, even from the outside, can see FACTs like... NEW chips being developed. As even YOU proved.

From Commodore engineers' point of view, ECS is not new.

This is not a dictionary debate.

Quote:

In this context, as I've said, what's important is how many of them were developed. You reported at leat FOUR revisions for ECS, and I bet that there are other revisions, which may involve other chips.

ECS Agnus and ECS Denise work for either A3000's Fat Gary or A500/A2000's Gary.

Lisa and Alice work with AA3000/A4000's Fat Gary or A1200's AA-Gayle.

There's no functional change with Paula.

Denise 8373 R1/R2/R3/R4 revisions do not fundamentally change 4 color double scan resolution modes. 8373 Phase II's Genlock fix is useless for most Amiga gamers.

Quote:

Again, the important thing is about the NEW chips: that was the discussion about.

From Commodore engineers' point of view, ECS is not new. You keep inserting yourself into a Commodore's internal matter.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Hammer wrote: @Karlos Quote: Karlos wrote: @you will never guess who On the subject of how the Amiga audio could be better, this insightful pearl of wisdom could not be more relavent. No wait, that other thing. Irrelevant. Anyway, in case you guys didn't hear it.... Filters. That's what we lacked. I've had the pleasure of playing Paula through an external filter, controlled via MIDI, and while the result is technically paraphonic (4 voices, one filter), the result is massive, especially when using all 4 channels like a single stacked oscillator. A fatter sound has never been heard. "Fat Paula" is tied to graphics chipset improvements due to the shared Chip RAM and Agnus's DMA controller. The rest of your post is irrelevant.

I've clearly PROVEN on my article that you could have up to 14 (FOURTHEEN!) audio channels using the same shared Chip RAM of an OCS system...

Of course, you need to change Agnus in order to have more DMA channels.

And change Paula to use a pair of 18-bit DACs to mix all channels, instead of having 14 x 8-bit DACs + 14 x 6-bit PWRs.

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: And you should know that engineers aren't there left alone: there's always a TECHNICAL lead that they belong to and which decides which direction to take. The technical leadership is weak i.e. considerably weaker than Bill Gates or Elon Musk examples.

You see it...
Quote:

The "Commodore The Final Years" book by Brian Bagnall should be a prerequisite subject matter for this topic. If copyright wasn't restrictive, I would have posted entire chapters from Brian Bagnall's book.

What you've reported 'til now is enough.
Quote:

Quote: You see yourself: NEW revisions -> NEW chips... Quote: That's your outsider assertion, not Commodore engineers'.

Guy, do you know that I've worked at Intel? It's that little (!) company which is producing... chips. And designing chips and masks was the same at Intel, do you know it?

When I was there I've received chips with new revisions. New chips, since new masks were produced.

And sometime they changed che implementation of technologies from one revision to the subsequent one: it happened to me with the SGX extension (I had to rewrite some tests for it).
Quote:

Quote: I, even from the outside, can see FACTs like... NEW chips being developed. As even YOU proved. From Commodore engineers' point of view, ECS is not new.

LOL. Of course! They are trying to defend themselves. What do you expect?

As I've said, chips were made the same everywhere: all companies worked that way.
Quote:

This is not a dictionary debate.

I agree: not for people which don't know how they things worked on chip companies. Why do you continue to talk about it, then?
Quote:

Quote: In this context, as I've said, what's important is how many of them were developed. You reported at leat FOUR revisions for ECS, and I bet that there are other revisions, which may involve other chips. ECS Agnus and ECS Denise work for either A3000's Fat Gary or A500/A2000's Gary. Lisa and Alice work with AA3000/A4000's Fat Gary or A1200's AA-Gayle.

And... several new chips.
Quote:

There's no functional change with Paula.

Too complex for those engineers?
Quote:

Denise 8373 R1/R2/R3/R4 revisions do not fundamentally change 4 color double scan resolution modes. 8373 Phase II's Genlock fix is useless for most Amiga gamers.

See above: new chips...
Quote:

Quote: Again, the important thing is about the NEW chips: that was the discussion about. From Commodore engineers' point of view, ECS is not new. You keep inserting yourself into a Commodore's internal matter.

See above: you're defending their lies...

@cdimauro

Quote:

I've clearly PROVEN on my article that you could have up to 14 (FOURTHEEN!) audio channels using the same shared Chip RAM of an OCS system... Of course, you need to change Agnus in order to have more DMA channels. And change Paula to use a pair of 18-bit DACs to mix all channels, instead of having 14 x 8-bit DACs + 14 x 6-bit PWRs.

Nothing is free.

As a Commodore engineer has stated, upgrading the Amiga chipset is like "opening a can of worms".

Alice wasn't upgraded for a reason e.g. backward compatibility.

A simple timing change can wreak havoc with Amiga games since there's no resource tracking.

SAGA's "turtle mode" shows the way forward.

Last edited by Hammer on 31-Jul-2024 at 06:15 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@cdimauro

Quote:

What you've reported 'til now is enough.

It's only the tip of the iceberg.

Quote:

Guy, do you know that I've worked at Intel? It's that little (!) company which is producing... chips. And designing chips and masks was the same at Intel, do you know it? When I was there I've received chips with new revisions. New chips, since new masks were produced. And sometime they changed che implementation of technologies from one revision to the subsequent one: it happened to me with the SGX extension (I had to rewrite some tests for it).

Don't apply Intel's culture to Commodore's internal matter.

Pentium FDIV fixed stepping didn't bring new capabilities e.g. Pentium 60 with FDIV bug is exchanged for newer stepping Pentium 60 with FDIV bug fixed. No free performance upgrade from warranty exchange.

NVIDIA's bumpgate episode with the Geforce 8 family example, my GeForce 8600M GT (G84, NB8P-GS) 256 MB video memory was exchanged for GeForce 9500M GS (G84, NB9P-GE1-A2) 512 MB video memory. The GPU core design is the same i.e. 32 CUDA cores scale.

Some owners of the GeForce 8600M GT have a GeForce 9700M GT (G96, NB9E-GE) exchange. GeForce 9700M GT has 32 CUDA cores scale.

I have a free video memory upgrade with the warranty exchange.

NB9P-GE1-A2 has a newer stepping of G84 that mitigates NB8P-GS bumpgate problems.

Both NB9P-GE1-A2 and NB8P-GS are "G84" has different stepping.

Quote:

See above: you're defending their lies...

That's your assertion.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: I've clearly PROVEN on my article that you could have up to 14 (FOURTHEEN!) audio channels using the same shared Chip RAM of an OCS system... Of course, you need to change Agnus in order to have more DMA channels. And change Paula to use a pair of 18-bit DACs to mix all channels, instead of having 14 x 8-bit DACs + 14 x 6-bit PWRs. Nothing is free.

Even on 1990, after FIVE year that the Amiga 1000 was introduced?
Quote:

As a Commodore engineer has stated, upgrading the Amiga chipset is like "opening a can of worms".

Then the question raises: was there any COMPETENT engineer left for upgrading the chipset? Or... none?
Quote:

Alice wasn't upgraded for a reason e.g. backward compatibility. A simple timing change can wreak havoc with Amiga games since there's no resource tracking.

Yes. And that's why I repeat again that all my proposals are 100% backward-compatible.
Quote:

SAGA's "turtle mode" shows the way forward.

I would say no: it's not needed, if the chipset is initialized in a backward-compatible way (and new features enabled only... on demand).
Quote:

Hammer wrote: @cdimauro Quote: What you've reported 'til now is enough. It's only the tip of the iceberg.

Then I hope that you can publish some other pearls from the above engineers...
Quote:

Quote: Guy, do you know that I've worked at Intel? It's that little (!) company which is producing... chips. And designing chips and masks was the same at Intel, do you know it? When I was there I've received chips with new revisions. New chips, since new masks were produced. And sometime they changed che implementation of technologies from one revision to the subsequent one: it happened to me with the SGX extension (I had to rewrite some tests for it). Don't apply Intel's culture to Commodore's internal matter.

It isn't Intel culture: it's how it goes on the silicon industry.
Quote:

Pentium FDIV fixed stepping didn't bring new capabilities e.g. Pentium 60 with FDIV bug is exchanged for newer stepping Pentium 60 with FDIV bug fixed. No free performance upgrade from warranty exchange. NVIDIA's bumpgate episode with the Geforce 8 family example, my GeForce 8600M GT (G84, NB8P-GS) 256 MB video memory was exchanged for GeForce 9500M GS (G84, NB9P-GE1-A2) 512 MB video memory. The GPU core design is the same i.e. 32 CUDA cores scale. Some owners of the GeForce 8600M GT have a GeForce 9700M GT (G96, NB9E-GE) exchange. GeForce 9700M GT has 32 CUDA cores scale. I have a free video memory upgrade with the warranty exchange. NB9P-GE1-A2 has a newer stepping of G84 that mitigates NB8P-GS bumpgate problems. Both NB9P-GE1-A2 and NB8P-GS are "G84" has different stepping.

Which doesn't mean that new things could not be implemented as well.

However, here you're forgetting the most important thing: Commodore engineers were sleeping with the platform upgrades, whereas Intel, Nvidia, etc. have continuously providing new products.

Having to create new masks for new chips could have been a nice chance to introduce new features as well.
You can clearly see it from what you've reported about how the new features were progressively added to the ECS Denise, 'til it reached the final status.
Quote:

Quote: See above: you're defending their lies... That's your assertion.

Well, that's what I see.

@Karlos

Quote:

Karlos wrote: Just going to say it again. Fully programmable, resonant lowpass filters, post DAC, per channel. Trust me, that would have slayed everything and everyone in 1985.

The idea wasn't to slay everything, but provide realistic sound for games using simple circuitry. And it did slay everything and everyone (in the home computer market) in 1985.

But other machines were using synth chips, so gamers expected to hear synth style music. When the Adlib card was released in 1987 PC gamers went gaga over its cheap synth keyboard style music, and OPL became the computer game music 'standard'. In 1989 Creative Labs added a single 8 bit PCM channel for spot sound effects, and that was all the PC gaming industry needed.

The Amiga had the equivalent of 4 Sound Blasters worth of PCM sound, but no synth chip. So instead of trying to cram more channels into Paula, Commodore should have just added an OPL chip. That would make porting PC games much easier as well as being better for music programs. And it would take almost no design effort.

So why didn't they? Because the Amiga's 'standard' was Paula. OPL sound was 'foreign' and few Amiga fans were interested in it, preferring Paula's more 'real' sound even though the fewer channels limited what music it could play. And It didn't take long for musicians to figure out how make awesome music with those 4 PCM channels. Turns out more channels weren't really needed!

Quote:

Sample based synthesis on the cheap, would have been lapped up by musicians in 1985.

Sure. But the Amiga wasn't designed to be a professional musical instrument. The chipset was designed for playing computer games, and doing it cheaply - just like other home computers. Your fully programmable filters would raised the price significantly, and for who - a few penny-pinching musicians?

@bhabbott

But Paula *is* a synth chip. Always was. Your sound blaster pcm does not have variable sample rates with microtonal precision over many octaves, nor does it have DMA looping. Paula does. Why? Doesn't that complicate the design unnecessarily?

Sure it could all be a pure accident of design but when you put those together you get what musicians instantly recognise as a wavetable synthesis model. Each Paula channel is an oscillator, capable of playing samples, but equally capable of reproducing any waveshape and playing it ad-infinitum with no input from the CPU. Its actually more complicated to play one shot samples than it is to configure as an oscillator.

Other clues that it's not just a simple DAC for PCM reproduction come from the interchannel modulation capabilities. Show me one reasonable standard sound replay use case for that. On the flip side, the volume modulation capability is useful in synthesis (ring modulation), and the period modulation allows a form - albeit extremely non intuitive to use - of FM.

_________________
Doing stupid things for fun...

@bhabbott

Quote:

bhabbott wrote: @Karlos Quote: Karlos wrote: Just going to say it again. Fully programmable, resonant lowpass filters, post DAC, per channel. Trust me, that would have slayed everything and everyone in 1985. The idea wasn't to slay everything, but provide realistic sound for games using simple circuitry. And it did slay everything and everyone (in the home computer market) in 1985. But other machines were using synth chips, so gamers expected to hear synth style music. When the Adlib card was released in 1987 PC gamers went gaga over its cheap synth keyboard style music, and OPL became the computer game music 'standard'. In 1989 Creative Labs added a single 8 bit PCM channel for spot sound effects, and that was all the PC gaming industry needed. The Amiga had the equivalent of 4 Sound Blasters worth of PCM sound, but no synth chip. So instead of trying to cram more channels into Paula, Commodore should have just added an OPL chip. That would make porting PC games much easier as well as being better for music programs. And it would take almost no design effort. So why didn't they? Because the Amiga's 'standard' was Paula. OPL sound was 'foreign' and few Amiga fans were interested in it, preferring Paula's more 'real' sound even though the fewer channels limited what music it could play. And It didn't take long for musicians to figure out how make awesome music with those 4 PCM channels. Turns out more channels weren't really needed!

It turns out that you have never talked with a musician and/or to a game developer, otherwise they all would have told you that you don't know of what you talk about.

In fact, it wasn't a matter if it was missing OPL or PCM channels: the Amiga lacked POLYPHONY!

This was very well known. And even obvious, if you had the chance to create a music track or to develop a game where you've to play the music track AND and also sound effects!

Why applications like OctaMED were developer according to you? Just for a pure exercise?
Quote:

Quote: Sample based synthesis on the cheap, would have been lapped up by musicians in 1985. Sure. But the Amiga wasn't designed to be a professional musical instrument.

Then why 4 DACs with independent PWRs for controlling their volume on... 1985? Weren't two enough? Maybe with a single PWR as master volume control (as it's used nowadays on sound cards)?
Quote:

The chipset was designed for playing computer games, and doing it cheaply - just like other home computers.

Sure. Then you've to tell me why it has 8... sprites: for controlling 8 mouse pointers?
Quote:

Your fully programmable filters would raised the price significantly, and for who - a few penny-pinching musicians?

That's something that a hardware guy should give a feedback.

@cdimauro

The penny-pinching musicians jibe makes it absolutely clear he's never talked to an actual musician. There's a saying in the music community: Get your kids into music hardware and they'll never have money for drugs. Music gear is genuinely expensive. Plenty of musicians did use the Amiga, despite this being the STs home turf, precisely because it was an affordable way to deal with samples and sequencing.

We may ask, why did SID have filters? All the competition had plinky plonky squarewave or if you are lucky, triangle and maybe sawtooth. SID had filters because it was designed to be musically useful. Filters are a key component of subtractive synthesis.

Paula was also designed to be musically useful. The single most obvious clue to that is the period register with enough range to recreate the chromatic scale over multiple octaves while retaining enough precision to make the microtonal adjustments necessary for fine tuning and effects like vibrato and pitch bend. You can literally play your sample, musically, rather than just reconstruct a stream at some fixed rate.

The next really obvious clue you already point out, the four independent voices. More would definitely have helped, which as you say, is why channel mixing trackers became a thing. Where possible, I preferred to multitrack record since software mixing of channels basically loses all the special character. You are better off with a soundcard intended for stream playback then, because that's what all your digital mixing has resulted in: a fixed rate PCM stream that you want to reconstruct.

The channel modulation seems to have no non-musical application, although you can use period modulation to turn Paula into a data over FM device.

Free panning and programmable filters. That's what Paula lacked more than anything. Even additional voices can be overcome, which is where having more than one Amiga and MIDI over a serial link comes in useful. I mean you could buy a fair few nice and bright A1200 for the price of a dedicated hardware sampler in 1993.

The filters didn't need to be super expensive SSM licensed chips used in synths and samplers of the day. Something on par with the SID, or perhaps a bit better, would be fine, especially if it was completely bypassable for those cases where you don't want it.

Last edited by Karlos on 02-Aug-2024 at 04:32 PM.

_________________
Doing stupid things for fun...

@Karlos

Quote:

Music gear is genuinely expensive

I always found it quite cheap, considering what you could get for fairly low amounts of money. I still wish I had bought some classic 80s gear as a kid when it was new.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@cdimauro

Quote:

@cdimauro Even on 1990, after FIVE year that the Amiga 1000 was introduced?

New fabs and R&D need resources.

Quote:

From Commodore - The Final Years, However, those who looked closer at the annual report noted that Commodore’s expenditure in R&D was exceedingly low for a billion dollar company, spending only 3% of its revenues, or $31.4 million, on engineering. All this while the company spent $174.3 million, or 16.6%, on Sales and Marketing. This put Commodore at the bottom of computer companies, including Apple and Atari, in R&D expenditures.

R&D is shared between Commodore PC's cost reduction R&D and Amiga R&D.

Apple is fully focused on the Macintosh desktop platform.

Released in 1990 for $2995 USD, Atari TT has 256 colors mode, but it's not shared with its mass-produced ST /STe line. Atari's product segmentation is crippled by Jack Tramiel's C16 vs C64 product segmentation mindset.

PC's 256-color competition is mass-produced in large numbers.

Quote:

@cdimauro Then the question raises: was there any COMPETENT engineer left for upgrading the chipset? Or... none?

Commodore management removed the original Amiga engineers. Commodore in-house engineers and managers have #metoo mindset instead of trailblazing A1000.

Commodore in-house engineers and managers have a talent for cost reduction e.g. A500.

The two halves made the A500's successful.

In Commodore - The Final Years book, Commodore in-house engineers and sidelined Jeff Porter have claimed they can beat Bill Sydnes and Jeff Frank on A300's cost reduction.

Commodore in-house engineers have identified mistakes in A600's cost structures. Ex-IBM PC personnel doing cost reductions is a joke.

A300's purpose is to replace C64 in markets with weak currencies and low-income socio-economic demographics. The A300 wasn't supposed to replace the A500.

Commodore has two main revenue pillars i.e. C64 and A500.

Quote:

@cdimauro Having to create new masks for new chips could have been a nice chance to introduce new features as well. You can clearly see it from what you've reported about how the new features were progressively added to the ECS Denise, 'til it reached the final status.

"Read my lips, no new chips" refers to Amiga's graphics core architecture change.

Genlock is not a new feature since OCS Denise has Genlock, early ECS Denise broke the existing Genlock feature.

AGA has met VGA's 256 colors criteria and reached entry-level SVGA i.e. 640x480p 256 colors.

The sidelined Jeff Portter was moved to the CDTV division where several managers have resigned themselves due to low R&D resources. From the CDTV division's low R&D resources, CD32 was born with Akiko C2P i.e. a hardware patch on AGA's chunky pixels issue.

Jeff Frank and Bill Sydnes fuked-up the main Amiga R&D division. The main Amiga R&D division has Amiga custom chipset R&D and Jeff Porter has lost control over this division.

Purchase Commodore -The Final Years book for Jeff Porter's wish list before being sidelined by pro-PC Jeff Frank and Bill Sydnes.

Jeff Porter's wish list follows the operational AGA state which includes chunky pixels mode.

Pro-PC campers Jeff Frank and Bill Sydnes have "lost more than six months" and are the major contributors to the $366 million loss that mortally wounded Commodore.

Commodore -The Final Years book reveals the timeline and the fckups by pro-PC Jeff Frank and Bill Sydnes.

Jeff Frank's road map is an important factor for the situation in the critical 1991 to 1992 years.

For 3D, DSP3210 is the low cost method to increase compute power to beat 80386DX-33, Am386DX-40, 80486 and 68040 class CPUs. You would need costly 68060-50 or Pentium 60 to beat cheap DSP3210-50 on 3D.

DSP3210's i.e. DAU(FP) and CAU (INT) has access to 32-bit bus. DSP3210 is designed to interface with 680X0 and 80X86 bus i.e. not foreign for 68K based machines.

I view DSP3210 like IEEE-754 stream processor unit in one of AMD's Graphics Core Next (GCN).

Last edited by Hammer on 03-Aug-2024 at 02:16 AM.
Last edited by Hammer on 03-Aug-2024 at 02:10 AM.
Last edited by Hammer on 03-Aug-2024 at 02:02 AM.
Last edited by Hammer on 03-Aug-2024 at 01:59 AM.
Last edited by Hammer on 03-Aug-2024 at 01:56 AM.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@Hammer

Quote:

Apple is fully focused on the Macintosh desktop platform

Uh-huh, and monitors, printers, scanners, external storage devices, newton, modems, A/UX and tons of different Macintosh models.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

Quote:

kolla wrote: @Karlos Quote: Music gear is genuinely expensive I always found it quite cheap, considering what you could get for fairly low amounts of money. I still wish I had bought some classic 80s gear as a kid when it was new.

There's a gulf between the cheap stuff and the good stuff. The SK1 was a sampling keyboard, 4 voices, that you could buy for under 100 standard earth credits in 1985. About 1.7 seconds of sample time, 9kHz sample rate and a basic built-in sequencer. It also had additive synthesis and a few pcm voices. I had one,it was fun but not terribly useful. As a sampler, the Amiga, even in 1985 was significantly better.

The Emulator II, released a year earlier than the Amiga, was 8000 dollars. 8000 dollars got you 8 voices, 27kHz sampling, 8-bit (delta companded) and all analogue signal processing after the DAC (analogue VCA, analogue VCF). And if you wanted to do sampling or sample based synthesis in the 80s, the Emulator *was* the cheap option. The 16-bit Fairlights were 30,000 dollars for the starter unit.

Even with just 4 voices, with midi triggered playback or sequenced directly, the Amiga, especially the Amiga 1200 with it's 2MB chip memory and much higher reconstruction filter cutoff, made a great and affordable multitmbral sampler. Sure you could get a nice Akai S3000 by 1996, with 32 voices and digital filters, but that was still 3000 dollars new. And that's for the basic 2MB version. You want the version with more memory (up to 16MB, IIRC) and a SCSI HD, brace yourself for price hikes that make the equivalent computer upgrade seem inconsequential.

Someone somewhere didn't realise the full musical potential here. Which is a shame.

Last edited by Karlos on 03-Aug-2024 at 12:05 PM.

_________________
Doing stupid things for fun...

Just an addendum to that, an expanded A1200, running OctaMED 6 or higher with FastMemPlay enabled makes an even better multitmbral MIDI sampler, since you are no longer restricted by the 2MB chip memory limit. The midi event to playback of this solution has imperceptible latency, as the buffers involved are tiny, but thanks to the rock solid timing of the chipset, the audio is completely stable.

If you don't mind the limitations of 14-bit playback, MidiIn is also an excellent bit of software that I often used if I wanted to use 16-bit samples triggered over MIDI.

https://forum.amiga.org/gallery/1_midiin.png

The infamous mellotron choir of the undead...

Last edited by Karlos on 03-Aug-2024 at 12:19 PM.
Last edited by Karlos on 03-Aug-2024 at 12:14 PM.

_________________
Doing stupid things for fun...

@Karlos

Well, didn’t even consider samplers… a ping-pong tape recorder from who knows where that could mix 4 tracks to one, then 3, 2 and 1, casio cz synths, various guitars, violins.. and pedals were fairly cheap.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

You didn't consider it because it was out of the realms of affordability for most people. Even the top of Casio's home sampling keyboards, the SK 2100 (IIRC) was barely better than the SK1. Miniscule increase in sampling rate and duration, but more polyphony. This is the gap that the Amiga could've utterly dominated, but nevertheless was able to fill for people that knew about it. 4 voice, 27kHz sampling with independent left/right outputs that you can process and mix independently was a good thing (tm).

On the flip side of sampling being expensive, there were oddly affordable subtractive synths and by the mid-late 80s, reasonable good FM synthesis options. I still have a working PSS680 that I was still using right up to the late 90s for it's curiously lofi textures. Routed through external FX and mixing, as with the Amiga samples, it has a place.

Sampling was expensive originally because memory was expensive. It was expensive later because a bar and expectation had been set, it was an in-demand feature and so it was extremely profitable. By the mid 90s, memory prices had fallen but professional hardware samplers were still eye-wateringly expensive. Especially as some of the more expensive analogue components migrated to purely digital implementations that could be implemented in comparatively inexpensive DSP.



There has been a more recent interest in more lofi sampling with new gear arriving for 12 and even 8 bit grittiness.

_________________
Doing stupid things for fun...

@Karlos

Quote:

Karlos wrote: @cdimauro The penny-pinching musicians jibe makes it absolutely clear he's never talked to an actual musician. There's a saying in the music community: Get your kids into music hardware and they'll never have money for drugs. Music gear is genuinely expensive. Plenty of musicians did use the Amiga, despite this being the STs home turf, precisely because it was an affordable way to deal with samples and sequencing. We may ask, why did SID have filters? All the competition had plinky plonky squarewave or if you are lucky, triangle and maybe sawtooth. SID had filters because it was designed to be musically useful. Filters are a key component of subtractive synthesis.

I loved it. And that's why Machinae Supremacy is one of my favourite bands.
Quote:

Paula was also designed to be musically useful. The single most obvious clue to that is the period register with enough range to recreate the chromatic scale over multiple octaves while retaining enough precision to make the microtonal adjustments necessary for fine tuning and effects like vibrato and pitch bend. You can literally play your sample, musically, rather than just reconstruct a stream at some fixed rate.

Exactly. And this way samples required LESS space compared to systems with higher-frequencies but fixed.
Quote:

The next really obvious clue you already point out, the four independent voices. More would definitely have helped, which as you say, is why channel mixing trackers became a thing. Where possible, I preferred to multitrack record since software mixing of channels basically loses all the special character. You are better off with a soundcard intended for stream playback then, because that's what all your digital mixing has resulted in: a fixed rate PCM stream that you want to reconstruct.

Indeed. Mixing channels could be made to save resources, but only almost at the end of the sound processing: just before sending the digital output to the (higher precision) DAC.
Quote:

The channel modulation seems to have no non-musical application, although you can use period modulation to turn Paula into a data over FM device.

That's something which I've never experimented. So, I've no idea of what I can get.
Quote:

Free panning and programmable filters. That's what Paula lacked more than anything.

Free panning was very simple and fully-backward compatible to implement.

Programmable filters... I don't know. I've no clue of what was required.
Quote:

Even additional voices can be overcome, which is where having more than one Amiga and MIDI over a serial link comes in useful. I mean you could buy a fair few nice and bright A1200 for the price of a dedicated hardware sampler in 1993.

Indeed. I've seen your last comment.
Quote:

The filters didn't need to be super expensive SSM licensed chips used in synths and samplers of the day. Something on par with the SID, or perhaps a bit better, would be fine, especially if it was completely bypassable for those cases where you don't want it.

These: https://www.c64-wiki.com/wiki/SID

Programmable filters (low pass, bandpass, high pass)

?
Quote:

Karlos wrote: @kolla Quote: kolla wrote: I always found it quite cheap, considering what you could get for fairly low amounts of money. I still wish I had bought some classic 80s gear as a kid when it was new. There's a gulf between the cheap stuff and the good stuff. The SK1 was a sampling keyboard, 4 voices, that you could buy for under 100 standard earth credits in 1985. About 1.7 seconds of sample time, 9kHz sample rate and a basic built-in sequencer. It also had additive synthesis and a few pcm voices. I had one,it was fun but not terribly useful. As a sampler, the Amiga, even in 1985 was significantly better. The Emulator II, released a year earlier than the Amiga, was 8000 dollars. 8000 dollars got you 8 voices, 27kHz sampling, 8-bit (delta companded) and all analogue signal processing after the DAC (analogue VCA, analogue VCF). And if you wanted to do sampling or sample based synthesis in the 80s, the Emulator *was* the cheap option. The 16-bit Fairlights were 30,000 dollars for the starter unit. Even with just 4 voices, with midi triggered playback or sequenced directly, the Amiga, especially the Amiga 1200 with it's 2MB chip memory and much higher reconstruction filter cutoff, made a great and affordable multitmbral sampler. Sure you could get a nice Akai S3000 by 1996, with 32 voices and digital filters, but that was still 3000 dollars new. And that's for the basic 2MB version. You want the version with more memory (up to 16MB, IIRC) and a SCSI HD, brace yourself for price hikes that make the equivalent computer upgrade seem inconsequential. Someone somewhere didn't realise the full musical potential here. Which is a shame.

Shocking! With more polyphony Amiga could have been THE platform of choice for professional musician! Those tools were so much expensive!!!
Quote:

Karlos wrote: Just an addendum to that, an expanded A1200, running OctaMED 6 or higher with FastMemPlay enabled makes an even better multitmbral MIDI sampler, since you are no longer restricted by the 2MB chip memory limit. The midi event to playback of this solution has imperceptible latency, as the buffers involved are tiny, but thanks to the rock solid timing of the chipset, the audio is completely stable. If you don't mind the limitations of 14-bit playback, MidiIn is also an excellent bit of software that I often used if I wanted to use 16-bit samples triggered over MIDI. https://forum.amiga.org/gallery/1_midiin.png The infamous mellotron choir of the undead...

I never heard it before, but I got it: nice software. To overcome the lack of polyphony...

@cdimauro

Yeah, even basic SID style filters would've been a good value add for musicians. Whether it's just filtering out unwanted noise from a lofi sample or doing subtractive synthesis on a waveform shape, that's all you need. The lowpass mode would definitely benefit from a resonance control too.

The key purpose of MidiIn is to make the Amiga an affordable hardware sampler capable of better-than-8-bit output without needing a soundcard. The screenshot there is from my Amiga back when I was using it for this.

It works exactly like a hardware sampler, allowing you to do all kinds of multitimbral, multisampling with key splits etc, controlled over MIDI. The replay on Paula is 14-bit at up to 56kHz if you are using DblPal or CGX modes. In that screenshot I was using a 48kHz output rare with 16-bit multisampled (one sample per 4 semitones) famous Mellotron choir sound. It was set to 8 voices there but I've definitely used up to 20 channels at 48kHz without any issues on a 25MHz 040.

It has some of the expected artefacts of 14 but replay but IIRC it supports several soundcards too.

Last edited by Karlos on 03-Aug-2024 at 04:04 PM.

_________________
Doing stupid things for fun...