8-bit sound chips

Technically what's the agreed upon order for the 8-bit sound chips? I was sitting here wondering more specifically about Master System Vs Spectrum 128, seeing as it already seems agreed on that capability wise C64 is #1 and NES is #2

Also I thought I'd make it a poll for general preferences (after all they all sound different so any choice for best is reasonable irrespective of actual capabilities)

For me I would have to say C64 is my favourite 8-bit sound chip, though the NES has a pretty awesomely nostalgic sound to it as well

C64 = Electric Guitar
NES = Harmonica

technically speaking, the YM2612 is 8bit too in MD, aswell as all other FM chips used in a ton of arcade machines and computers :P

Quote:

---

Originally Posted by TmEE

technically speaking, the YM2612 is 8bit too in MD, aswell as all other FM chips used in a ton of arcade machines and computers :P

---

To clarify, best sound chip used in an 8-bit home console or computer that was released during the 2nd or 3rd generation of gaming :)

I'm going to say SID from that list. Too bad there's way too many phone ringing sounds simulated on it :P

The sid chip, of course! No 8-bitter out there could do electric guitar quite like the sid, just to name one. In the hands of a competent composer, or even an amateur, this chip could do anything.
Of course I'm a bit biased, I was raised on the C64, and My SX-64 is not more then a few feet away from Me as I write this :)

I love the NES sound chip.

SID! I got the high voltage SID collection. I even have my C64, but no disk drive or cartridge= no play time. :(

The SID only had 3 channels but it was far outweighed by all the other stuff it could do. Each channel was fully customizable with several waveforms to choose from in addition to filters which none of the others had. Skilled sound programmers could easily fit drums and bass into the same sounds to make up for the lack of channels.
But most of all it had the benefit of being on an open computer and the culture/exchange of ideas that came with it. Anyone with a C64 could start making their own music, and people met at these large gatherings where they would share cool tricks and techniques they learned to get more out of the chip.

No love for the beeper?!?!

It does a mean "in the hall of the mountain king" :lol:



Also, still no word on Texas Instruments Vs AY-3-8910 sound chips? are they basically the same thing capabilities wise?

NES all the way, baby. Though SID has some nice stuff too.

There are too many SID tracks are made just for music (have no care on cpu resource). Which a lot of people don't realize. NES original tracks are starting to get like that too, though. People here NSF's and think OMG! But what do you expect when you have like 16+ channels and 4-5 addon chips in a NSF. Or a tracker that's 100% dedicated to sound generation and not practical in game cpu resource. It's deceiving. I could totally do the same thing on PCE and give it 100% cpu resource to do FM or AM modulation on a bunch of soft mixed channels to a 10bit DAC and still have 4 regular channels to manipulate at high frequencies. Might sound real nice, but not practical or realistic. But a general audience wouldn't be the wiser.

Quote:

---

Originally Posted by Thenewguy

Also, still no word on Texas Instruments Vs AY-3-8910 sound chips? are they basically the same thing capabilities wise?

---

The AY has a wider pitch range and a more precise pitch selection, and a hardware volume envelope (only one envelope, but any channel or combination of channels can use it). The AY only has three channels compared to four on the SN (three square waves plus one white noise), but each channel on the AY is capable of producing square waves, white noise, or a mix of the two. The AY is rather flexible, and is actually quite a powerful sound chip when used in certain ways. The SN is probably the most boring and basic sound chip ever created.

Quote:

---

Originally Posted by roundwars

The SN is probably the most boring and basic sound chip ever created.

---

QFT ;)

What? No love for the Adlib, MT-32, or original Sound Blaster?

My top 5:

1: SID
2: Adlib
3: Master System's YM2413
4: AY
5: NES's sound system

Also Manic Miner's rendition of Halls of the Mountain King literally feels like getting humped in the ear on my headphones... it sounds much better on speakers.

Quote:

---

Originally Posted by roundwars

The SN is probably the most boring and basic sound chip ever created.

---

The Spectrum Beeper is glad to hear that, take that Master System!

:rofl:

#Spectrum starts beeping out Hall of the Mountain King in celebration

doodoodoodoo doodoodoo doodoodoo doodoodoo

Quote:

---

Originally Posted by roundwars

The AY has a wider pitch range and a more precise pitch selection, and a hardware volume envelope (only one envelope, but any channel or combination of channels can use it). The AY only has three channels compared to four on the SN (three square waves plus one white noise), but each channel on the AY is capable of producing square waves, white noise, or a mix of the two. The AY is rather flexible, and is actually quite a powerful sound chip when used in certain ways. The SN is probably the most boring and basic sound chip ever created.

---

I really wonder why the AY chip (and yamaha clone) weren't applied to several cases the SN76489 was.
With the CV it made some sense to get both the PSG and VDP from TI and the 2 had been part of the same original chipset.
The SG-1000 may have been a pretty close CV clone, but the MSX was in the works too so should have been a consideration: plus the 8-bit I/O ports should have been useful.

Same with the PC Jr and Tandy 1000, but maybe there was some other deal
involved with TI and IBM. (again, 8-bit parallel ports -or port in the 28 pin version could have been useful -at least if they meshed with the chipset and given applications in other system it would seem the case -unlike several chips meant for interfacing with the 680x/650x CPUs)

Hell, I'm a bit surprised it wasn't used on PC sound cards before the OPL2 got popular. (either AY or SN for that matter -if a gameport was added the I/O ports might be of utility, though the standard point only needed 4 parallel lines... unless some 3rd party decided to go with the broader DE-9 Atari/Commodore/Sega/Coleco/etc type joystick ports with the standard digital joystick pinouts -maybe routing analog lines too, like the Atari and PC gameports or added more digital lines like Sega/Amiga/ST they could have used all 8 data lines per joyport --and given the commonality it might have surpassed the IBM gameport which was barely in use until the late 80s anyway -and software polling of pots was CPU intensive)

Quote:

---

Originally Posted by tomaitheous

There are too many SID tracks are made just for music (have no care on cpu resource). Which a lot of people don't realize. NES original tracks are starting to get like that too, though. People here NSF's and think OMG! But what do you expect when you have like 16+ channels and 4-5 addon chips in a NSF. Or a tracker that's 100% dedicated to sound generation and not practical in game cpu resource. It's deceiving. I could totally do the same thing on PCE and give it 100% cpu resource to do FM or AM modulation on a bunch of soft mixed channels to a 10bit DAC and still have 4 regular channels to manipulate at high frequencies. Might sound real nice, but not practical or realistic. But a general audience wouldn't be the wiser.

---

Doesn't the SID have a relatively limited frequency range too, or at least limited maximum frequency? (4 kHz iirc) How do the others compare in that respect? (POKEY, NES, AY/YM2149 -I know SN is limited in range compared to others)

One thing about software channels and sample playback with CPU resource is how well it can be done in-game with limited resources and not just for demos/title screens:
Stormlord on the C64 has a mod player in the intro (with some analog synth too -ST version is all soft mod), but it also has a fair bit of samples in the in-game music.
http://www.youtube.com/watch?v=6WH6cYCXZBs


Now what could be a consideration is what chips allow good sample playback with minimal resources:
NES has the 7-bit DPCM support
AY/SN chips tend to be best using 3 square channels (AY with more range) both require costomized samples and a lot of resource. (perhaps less if interval timers are available)
SID had a couple hacks for playing PCM (again external timers might help, not sure if the C64 had that set-up)

POKEY had 4 4-bit DACs as well as built-in interval timer functionality and an IRQ line, so a bit like the YM2612 had potential for: timed IRQ with 4-bit linear PCM.

Quote:

---

Originally Posted by Dant

What? No love for the Adlib, MT-32, or original Sound Blaster?

My top 5:

1: SID
2: Adlib
3: Master System's YM2413
4: AY
5: NES's sound system

Also Manic Miner's rendition of Halls of the Mountain King literally feels like getting humped in the ear on my headphones... it sounds much better on speakers.

---

The point was chips used 8-bit home platforms, IMB PC was never 8-bit (unless you argue 8088), hence Atari ST not mentioned with AY.

But the YM2413 should certainly be mentioned, as should any other add-on used for the MSX (which I think included the OPL and some others).

Also, there's other 8-bit home computers to consider like the FM-7 and PC-8800 series.

The latter went from a beeper (maybe YM2149 at some point, but it's mot mentioned -Super Mario special uses SSG only so it could be), then to the YM2203 in 1985 (3 SSG of the AY/YM2149 plus 3 4-op FM like the YM2612 in the same OPN family), then it got the very impressive YM2608 with SSG+6 4-op FM channel+ 6 rhythm/percussion voices+ one ADPCM with 2-16 kHz sample rate. (not sure if it's mono or has hard panning like the YM2612 DAC)

The YM2608 is by far the winner in sheer capabilities from everything else on the list so far.


It's a bit of a shame that Atari Corp never upgraded the YM2149 to a YM2203, same 40 PIN DIP size chip and fully compatible with the YM2149, though it requires a small external DAC chip. (YM2608 has full compatibility too, but would have been rather expensive -especially since Yamanha's ADPCM implementations tended to be a bit costly to use -requiring SRAM for sample memory iirc).

SID definitely. When used right, it sounded quite awesome. Which happened pretty frequently. There are some games that are available on 16-bit systems, that have better-sounding music on C64. I don't know of any such cases for NES or SMS.

Quote:

---

Originally Posted by j_factor

SID definitely. When used right, it sounded quite awesome. Which happened pretty frequently. There are some games that are available on 16-bit systems, that have better-sounding music on C64. I don't know of any such cases for NES or SMS.

---

It's still a mystery to me as to where Turrican 1 music plays in the game. Of all the longplays I've watched, it's only sound FX in game.

I want to say the YM2413, but it isn't there so I'll go with the SID.I was listening to Last Ninja 2 this morning, that stuff is still impressive.

Quote:

---

Originally Posted by tomaitheous

It's still a mystery to me as to where Turrican 1 music plays in the game. Of all the longplays I've watched, it's only sound FX in game.

---

Music plays during the shoot-em-ep levels

Unfortunately the song posted by J-Factor is not in-game, some dumbass who did a longplay of Turrican a couple of years ago added in his own in-game music during editing, hence tons of confusion ever since and a knock-on effect.

That song appears during the scoreboard section I believe.

The only songs I can remember being in Turrican in-game are these two (shmup levels) -

Quote:

---

Originally Posted by kool kitty89

Hell, I'm a bit surprised it wasn't used on PC sound cards before the OPL2 got popular. (either AY or SN for that matter -if a gameport was added the I/O ports might be of utility, though the standard point only needed 4 parallel lines... unless some 3rd party decided to go with the broader DE-9 Atari/Commodore/Sega/Coleco/etc type joystick ports with the standard digital joystick pinouts -maybe routing analog lines too, like the Atari and PC gameports or added more digital lines like Sega/Amiga/ST they could have used all 8 data lines per joyport --and given the commonality it might have surpassed the IBM gameport which was barely in use until the late 80s anyway -and software polling of pots was CPU intensive)

---

Not any sound cards got popular before FM became the standard, I couldn't tell you why though.

Quote:

---

Originally Posted by kool kitty89

The point was chips used 8-bit home platforms, IMB PC was never 8-bit (unless you argue 8088), hence Atari ST not mentioned with AY.

But the YM2413 should certainly be mentioned, as should any other add-on used for the MSX (which I think included the OPL and some others).

---

Ok, I will argue 8088... You'd actually be surprised how common the 8088 was in home PCs before the early '90s and how many people were using (at least) an ad-lib, heck you even had the Tandy/PCjr sound chip (which is similar to the Master System's chip, don't think it's the same though) both of which the most popular models of had an 8088.

Quote:

---

Originally Posted by kool kitty89

Doesn't the SID have a relatively limited frequency range too, or at least limited maximum frequency? (4 kHz iirc) How do the others compare in that respect? (POKEY, NES, AY/YM2149 -I know SN is limited in range compared to others)

---

How would it be limited? I've never come across a Synth of any nature that wouldn't play a note it was receiving... whether it was on the bottom or top of the scale. Not sure I'm following your question then... ;)



I voted for the SID btw... it's a monster. The guys that make the Machinedrum made a really badass synth with SID hardware (hell, I think it was two chips). They didn't make that many and consequently sold out of them in a hurry. I'd love to have one but they sell for way too much money.

NES comes 2nd imo... it makes Mono Music and Sound enjoyable for me on a level that nothing else could.

I can't choose one. The SID, NES soundchip and that SN764896798475438717523423 all make lovely music to me :cool:.

Quote:

---

Doesn't the SID have a relatively limited frequency range too, or at least limited maximum frequency? (4 kHz iirc)

---

Internally, it uses a phase accumulator. I think it's 24bit. With the top 4 bits used directly to build a waveform (instead of indexing into waveform memory). Saw, triangle(the MSbit is used as an inverter to ramp back down), square. I'm not sure what the driving clock is the for the phase accumulator, but if it's high like 0.896mhz or something - then a 16bit accumulation fixed/float point seems too small IMO. That would be 13hz steps. Dunno, maybe they have another step/divide down of the clock before the accumulator or such.

Unless you mean frequency response range (4khz would be 8khz sample output). I dunno.

Quote:

---

How do the others compare in that respect? (POKEY, NES, AY/YM2149 -I know SN is limited in range compared to others)

---

Pokey's single channel 8bit divider (period system) seems too low. I forget what the driving clock is. Maybe it's low enough for you to get at least two octaves out of it before you go out of range. Otherwise, sacrificing a channel just so another can have a combined 16bit divider... just doesn't feel right. Dual Pokey sounds nice though (heard some stuff for that, that didn't sound limited in range).

Quote:

---

POKEY had 4 4-bit DACs as well as built-in interval timer functionality and an IRQ line, so a bit like the YM2612 had potential for: timed IRQ with 4-bit linear PCM.

---

Even with A8's clock speed, I can imagine it really eating into cpu resource. Unless you did something extremely limiting. Even just playing static frequency samples with no soft volume.

I did a 4 channel mod driver on PCE. No soft mixing, no soft volume (hardware volume, hardware dac's) - which frees up cpu resource and it still took like 27% cpu resource to do it (though about 5.4% is because I made it dual interrupt friendly and buffered the samples). 24bit phase accumulator for each channel: http://www.pcedev.net/audio/pce/4chan_mod_driver.wav . 27% on the PCE cpu speed is not the same as 27% on the A8 65x speed processor. In relative perspective, Air Zonk on the PCE/TG uses 20% cpu resource for the audio driver (sample+music engine). A mod parser would added another 1-3% to the mod driver.

BTW does the NES sound chip actually have a name? LOL

I tried to find a name for a while but to no avail :(

APU. But it's also listed as the 2A03 (the cpu name).

Quote:

---

Originally Posted by Thenewguy

To clarify, best sound chip used in an 8-bit home console or computer that was released during the 2nd or 3rd generation of gaming :)

---

Does the Mark III or MSX fm chip count then?

Either way, I'm in no position to judge if anything is best, but my favorites are the SMS and NES sounds. I prefer some of the best SMS chip music over anything on NES, even though NES has a greater amount of quality music.

I don't like to count computers in with consoles in these types of discussions, but I still have never really liked the C64 sound. I don't hate it and some of the music I've heard sound cool. Too much of the music is in styles I'm not crazy about that are popular in europe but.

This is all preference of course. :p

Quote:

---

Originally Posted by Black_Tiger

I don't like to count computers in with consoles in these types of discussions, but I still have never really liked the C64 sound. I don't hate it and some of the music I've heard sound cool. Too much of the music is in styles I'm not crazy about that are popular in europe but.

---

Arpeggio overload? Hehe.

The SID also has a harsh/gritty sound to it (but filtered at the same time, if that makes sense. It's not 'sharp'). Sometimes it just works against it IMO.

Quote:

---

Originally Posted by tomaitheous

Arpeggio overload? Hehe.

---

Arpeggio overload indeed. That's actually the reason why I don't like a lot of European chiptunes, as they go way overboard on the rapid arpeggios.

Chose the NES sound chip. The Gameboy's chip would be my second pick if it was up there (Similar to the NES except for wavetable based third channel and no DPCM).

The SID sounds good as a synthesizer (The bass sounds pretty good), but most C64 games don't seem to take advantage of it.

Quote:

---

Originally Posted by tomaitheous

It's still a mystery to me as to where Turrican 1 music plays in the game. Of all the longplays I've watched, it's only sound FX in game.

---

Is it one of the many cases where you could chose sfx OR music perhaps? (and all those longplayers opted for music)

Quote:

---

Originally Posted by tomaitheous

I did a 4 channel mod driver on PCE. No soft mixing, no soft volume (hardware volume, hardware dac's) - which frees up cpu resource and it still took like 27% cpu resource to do it (though about 5.4% is because I made it dual interrupt friendly and buffered the samples). 24bit phase accumulator for each channel: http://www.pcedev.net/audio/pce/4chan_mod_driver.wav . 27% on the PCE cpu speed is not the same as 27% on the A8 65x speed processor. In relative perspective, Air Zonk on the PCE/TG uses 20% cpu resource for the audio driver (sample+music engine). A mod parser would added another 1-3% to the mod driver.

---

You'd at least be a good bit better off than cases requiring tricks like PWM, the SID's bug, or multi-channel stepped volume of the SN/AY (albeit AY can ideally approximate 8-bit linear samples with enough CPU grunt and all 3 channels). Even plain linear DACs would be tougher than with timed IRQ helping out.

Couldn't you manage 1 4-bit PCM channel with no soft mixing/volume used occasionally for music/sfx without digging into a ton of CPU time? Outside of demos and games halting for samples -as with many SMS and some ST games- the C64 had a couple that used such for more than just title demos -again, Stormlord off the top of my head. (and the C64 had less than 60 % the CPU speed and had to use a hack for PCM and no help from timers iirc)

Quote:

---

Originally Posted by Dant

Not any sound cards got popular before FM became the standard, I couldn't tell you why though.

---

The only other mainstream card was Creative's Game Blaster (12 channel stereo square PSG) and that was released the same year as Adlib... (and was actually was used better in a few games than early adlib contemporaries and sometime the simple stereo was even used)
What I meant was why wasn't anything released widely before 1987 at all. (supposedly there was an early SID based sound card that was very uncommon)
There were a variety of mass market PSGs and FM chips before '87. (the original OPL for one, or the AY/YM2149 -the YM2151 would have been more expensive, the YM2149 also facilitated direct upgrade to the YM2203 -1985 at least- and YM2608 -1987 or earlier, though the latter would have been a bit expensive by comparison)

For added PCM support a simple bare 8-bit DAC could have done wonders too (far less resource than PWM via PC speaker and PCs already had a built-in timer they could take advantage of for timed IRQ facilitating things even more) That's exactly what the Covox speech thing and Disney Sound source are, except those use the parallel port rather than an ISA slot. (in many cases they had results similar or even better than 8-bit sound blasters -depending on quality of analog comonents- though it would require a bit more resource than the SB's built-in MCU which also did ADPCM decompression)
Dosbox supports it too, see games like Wolf3D or several of the King's Quest and Space Quest games. (among others)

Quote:

---

Ok, I will argue 8088... You'd actually be surprised how common the 8088 was in home PCs before the early '90s and how many people were using (at least) an ad-lib, heck you even had the Tandy/PCjr sound chip (which is similar to the Master System's chip, don't think it's the same though) both of which the most popular models of had an 8088.

---

What I mean is the 8088 is arguably a 16-bit CPU, like the 68008, 65816, or 386SX is 32-bit (NOT like the 68000 is 32-bit as it's still only a 16-bit ALU, not just limited by the physical data lines)
Not that it matters that much, the 8086 wasn't much faster clock for clock in many cases.

8088 is technically the 8-bit variant of the 8086 (which IS 16-bit, kinda).

Though it uses the same instruction set..

Quote:

---

Originally Posted by evildragon

8088 is technically the 8-bit variant of the 8086 (which IS 16-bit, kinda).

Though it uses the same instruction set..

---

Yes, externally 8-bit (data bus), but a 16-bit CPU architecture, just like the 68008, 65816, and like the 386SX is 32-bit (but stuck on a 16-bit bus -same with the 68020s/30s in some low cost Macs and the Falcon).

Again, not like the 68000 is 16/32-bit as its still only got a 16-bit ALU internally. (and several 8-bit CPUs can do 16-bit operations, like the Z80) 6809 is an odd case too.

The 8088 should never have been invented. ;) I love the 8086.. It's technically 16-bit, but is odd in how it's segmented. I had learned to program my old 8086 computer years ago (i've long forgotten though). I should try and learn it again..

Quote:

---

You'd at least be a good bit better off than cases requiring tricks like PWM, the SID's bug, or multi-channel stepped volume of the SN/AY (albeit AY can ideally approximate 8-bit linear samples with enough CPU grunt and all 3 channels).

---

Oh, definitely. Well, unless you can get more bit precision another way than 4bits.

Quote:

---

Even plain linear DACs would be tougher than with timed IRQ helping out.

---

You mean it would be tougher than without the a TIMER int? Sure.

Quote:

---

Couldn't you manage 1 4-bit PCM channel with no soft mixing/volume used occasionally for music/sfx without digging into a ton of CPU time? Outside of demos and games halting for samples -as with many SMS and some ST games- the C64 had a couple that used such for more than just title demos -again, Stormlord off the top of my head. (and the C64 had less than 60 % the CPU speed and had to use a hack for PCM and no help from timers iirc)

---

Yeah single channel, fixed frequency sample, no volume or mixing on the A8. Couldn't tell you the resource % would be though. Probably low enough to be practical though.

Quote:

---

Originally Posted by evildragon

The 8088 should never have been invented. ;) I love the 8086.. It's technically 16-bit, but is odd in how it's segmented. I had learned to program my old 8086 computer years ago (i've long forgotten though). I should try and learn it again..

---

Why? It was to allow implementation in cheaper 8-bit memory systems. (one of the reasons it was chosen over the 68000 -the 68008 wasn't available when the PC was in development iirc)

The segmentation has nothing to do with the 8088 or being 16-bit for that matter. Both the 8088 and 8086 have similar segmentation with direct addressing of only a 16-bit address space (64k words), as with older 8-bit CPUs (6800, 6502, Z80, 8080 all had 16-bit address buses) with 4 segment registers expanding it to 1MB instead of 64 kB. (I'm not sure if the 8-bit vs 16-bit words make a different for the address range -an equal sized address bus with 16-bit words would be double that of 8-bit words)

It also multiplexes the address and data buses (as does the 65816) which managed to keep pin count down to 40 pins similar to older 8-bit CPUs. (65816 also did that, but managed a 24-bit address bus... well 24 address lines and a built-in bank switching scheme requiring an external latch iirc -not much more useful than external bankswitching logic, if at all -it requires slightly faster memory for proper access/response times with the address latch -or full internal bank-switching if they couldn't revise the address model for the 816 modes, but that would still require more pins)

Quote:

---

Originally Posted by tomaitheous

You mean it would be tougher than without the a TIMER int? Sure.

---

Yeah, they way the Genesis has to do it with the 68k/Z80, all software timing... (though it really shouldn't have had to as you've said before)
That and the SMS too, obviously, but that's working with the PSG, not a linear DAC. (so even timers wouldn't be so much help if I'm not mistaken -maybe a single PSG voice, but the stepped/logarithmic volume wouldn't make for nice sound on one voice alone)

In POKEY's case, would using multiple 4-bit channels allow for higher resolution in a similar fashion to the SN/AY chip? (with the linear DAC's I'd think the best you could get was 6-bit using all 4 channels)

Quote:

---

Originally Posted by kool kitty89

In POKEY's case, would using multiple 4-bit channels allow for higher resolution in a similar fashion to the SN/AY chip? (with the linear DAC's I'd think the best you could get was 6-bit using all 4 channels)

---

I think in the AY's case, it might be that the volume envelope allows you to achieve volume levels that you wouldn't be able to get with the ordinary 4-bit volume control.

Quote:

---

Originally Posted by kool kitty89

(I'm not sure if the 8-bit vs 16-bit words make a different for the address range -an equal sized address bus with 16-bit words would be double that of 8-bit words)

---

The bus is always in terms of bytes with all the CPUs I'm aware of. On the 68000, they don't give you A0, they start at A1 and give you a strobe for the upper or lower byte. The Intel family gives you A0 and no strobe, so you have to make the upper/lower byte strobe from A0 and other control lines.

Both have their pros and cons: strobes instead of A0 makes it less work for word-wide memory, but more work for byte-wide. A0 and no strobes is easier for byte-wide memory.

Quote:

---

In POKEY's case, would using multiple 4-bit channels allow for higher resolution in a similar fashion to the SN/AY chip? (with the linear DAC's I'd think the best you could get was 6-bit using all 4 channels)

---

You mean mixing all channels of POKEY vs mixing all channels of SN? SNxxxx or whatever (needs a shorter name) has log volume. I'd have to know the decibel drop for each step/volume drop. And what all channels combine at, to see all possible and usable combined output levels. Wiki already has an entry for combined AY channels of volume setting (though I'm guessing it produces a DC/unipolar waveform).

Though, why would you want to sacrifice all 4 channels on the Pokey for a single 6bit channel? If you plan to use it as a DAC for soft mix channels, than that's a redundant step. Doing one sample channel per DAC channel is the same thing. You still get a 6bit total output. And you don't have the overhead of soft mixing (or a LUT for 4x4 output).

Quote:

---

Originally Posted by tomaitheous

You mean mixing all channels of POKEY vs mixing all channels of SN? SNxxxx or whatever (needs a shorter name)
has log volume.

---

Well SN76489 is at least easier to remember than AY-3-8910, let alone the other variants like 8912 or 8913. (YM2149 is cleaner) ;)

Quote:

---

I'd have to know the decibel drop for each step/volume drop. And what all channels combine at, to see all possible and usable combined output levels. Wiki already has an entry for combined AY channels of volume setting (though I'm guessing it produces a DC/unipolar waveform).

---

One of the Atariage users (Crazyace) mentioned several times how the YM2149 made for superior resolution PCM playback than POKEY could do. (without the context of CPU resource, but ideal maximum performance -and indeed there is some pretty decent sounding PCM/MOD stuff on the ST)

This is one of the posts I remembered, but I know there was a much more recent and more detailed one:
http://www.atariage.com/forums/topic...0#entry1600980

Quote:

---

Originally Posted by Crazyace Fri Oct 10, 2008 5:08 AM

One nice thing about the yamaha chip was the volume being log based, rather than linear..
Quick work with a scope, and suddenly we had a table to give 8 bit precision sample playback..
( Could actually get more than 8 bits, but the first ADC in Replay was only 8 bit )
Everything else about it sucked though....

---

As roundwars mentioned, the envelope control may some into play as well.

Quote:

---

Though, why would you want to sacrifice all 4 channels on the Pokey for a single 6bit channel? If you plan to use it as a DAC for soft mix channels, than that's a redundant step. Doing one sample channel per DAC channel is the same thing. You still get a 6bit total output. And you don't have the overhead of soft mixing (or a LUT for 4x4 output).

---

I was thinking in the context of a direct comparison with the SN and AY chips. (both usually using all channels when playing samples)

I went with the texas instruments as phantasy star sounded beautiful.