Why not POKEY?

[jedinovice]

I'm curious because it seems such a feeble choice. (I'm not knocking the ST overall, neither am I complaining, I'm just trying to understand the reasoning.)

 

Atari know the Amiga was coming (that's how the ST was born, we all know.) The SID chip was famous. Atari had the POKEY chip already. The SID was out of their hands - it was owned by Commodore - but the POKEY chip was there, ready to be used, surely? If not, Yamaha must have had better chips available! After all, the DX7 was out in 1985! Why did they choose a sound chip used by 8 bitters, one inferior to the C64 and the 8 bit ataris?

 

I mean, everything else made sense - it had a decent graphics chip, decent OS (for the time) and decent disk drive. We might complain about single sided disks now but 360K compared very well with other disks at the time and cut costs which was important for the ST to succeed. I can understand every decision Atari made EXCEPT the 2149. Surely for a few pennies they could have used something better - indeed, why not POKEY?

 

Don't get me wrong. I'm not knocking the ST. I find the idea of a 68000 driving the soundchip of the Oric-1 quite amusing (in a nice way!) I just to understand the reasoning. It's almost as if the execs saud, "What's the cheapest chip we can throw in?" and that was it. It might have been the reasoning but it defies the rest of the machine and why didn't they use POKEY? They owned it, surely?!

 

I'm just trying to understand my 'new' machine better.

[Rybags]

A few possibilities:

- they wanted to distance the ST from the "games machine" image of the previous model ranges.

- the YM chip by default has a higher frequency range (discounting joining 2 voices on POKEY), although really it's only a marketing thing.

- the YM chip was cheaper (feasible since it was in pretty widespread use)

 

Agreed absolutely it is a far inferior chip. From a purist musical point of view, it is probably easier to program for than POKEY. But just compare games and demos and POKEY beats it in every department.

 

Even the noise generator was third rate - only one "style" of noise, and no filtering available (as per SID).

 

Realistically, they probably could have done a bit of extra work with the DMA controller and just hooked it up to a D/A convertor and had a "poor mans Amiga" style of sound production.

 

Or better still, they could have actually checked peoples resume before the mass-cullings and actually have had the expertise onboard to successfully have the AMY chip on all of it's new machines.

Edited December 14, 2006 by Rybags

[ijor]

Answering this kind of questions, unless you have some insider knowledge, is always speculative. But I mostly agree with what Rybags said.

 

It certainly wouldn't sound good to use the same chip of the 8-bit line. Disregarding if it is better or worse.

 

Pokey has a lot of extras besides sound generation that are not needed on the ST.

 

But probably the most important reason is that the YM chip has two 8-bit I/O ports. That's a huge saving if you would otherwise need to add something like a PIA.

 

Of course that the ideal solution would be a custom chip. But as we all know, there was no time for that.

[Retro Rogue]

Atari know the Amiga was coming (that's how the ST was born, we all know.)

 

That is *not* how the ST was born. I assume you're going by Mical's missinformation that's routinely thrown around in the public.

[doctorclu]

Hmmmm... interesting to think what the ST would have been like if the company was doing well, didn't change management, and had backwards compatibility on the mind.

 

A 68000 with 6502 in tow for some basic handling of side routines and called for older programs.

A "Pokier" chip, like the Pokey, but on steriods.

An "Antics" chip, capible of multiple functions.

 

Carry on the spirit of the 8-bit into the 16/32 era, giving all that the 8-bit users wanted at the time, more power, more speed, more ram, same computer they fell in love with.

 

As it was they got more power, kinda more speed, more ram, whole different computer. (Though ST X-Former was nice!)

 

 

Actually all this to say, I don't know how things could have been any different? What inspires a leap in technology? When one technology is not bringing in the bucks. The 8-bit Atari was getting old by 1986 (hard to believe since that was the hay day for me using my 400), and Atari needed something new.

Amiga was obviously a 800 on steriods, and generally was the machine I mentioned above, minus the backwards compatibility. But honestly, I'm glad Atari didn't make Amiga.

 

The ST to me seemed a step backwards in graphics compared to the 8-bit Atari, but it had the basics for being a good programming and utility machine, like good 80 column text for one. To me the ST line didn't really seem to make outstanding graphics till the STe line, and didn't really feel it was doing applications without struggling till the Mega STe and TT.

[mos6507]

There was no way to mix the ST and 8-bit under one hood. They were too different. I think it would have been a good idea to finish the AMY chip, though.

[Goochman]

Did the Yamaha chip handle MIDI better than Pokey? If so I would imagine this drove the decision more so since MIDI was instrumental (no pun intended ) to the ST.

[jens]

Is midi done by the soundchip?

[ijor]

Is midi done by the soundchip?

 

No, MIDI has nothing to do at all with the sound chip.

[Goochman]

Sorry - my bad - Funny thing about the Yamaha chip is to me it sounded more like a SID chip when I first got my ST.

[supercat]

SID was out of their hands - it was owned by Commodore - but the POKEY chip was there, ready to be used, surely? If not, Yamaha must have had better chips available! After all, the DX7 was out in 1985! Why did they choose a sound chip used by 8 bitters, one inferior to the C64 and the 8 bit ataris?

 

Many sound chips designed for 6502's expect to be sitting on a bus that will run the a nice predictable rate for which they were designed, with a memory access every cycle. Interfacing such a chip to a 68000 bus would be somewhat complicated. The 68000 does offer a couple features that would ensure the task wasn't totally hopeless, but it would still probably be necessary to add two or three extra chips to interface the POKEY to the 68000.

 

Further, Atari may have decided that the supply of POKEY chips, while ample for its current offerings, was sufficiently finite that it wasn't worth using in a new design.

 

As chip makers move to more advanced processes, getting old designs fabricated becomes more and more difficult and expensive. Even if someone had all of the original rubylith films for the POKEY and they were in pristine condition, it would be cheaper to re-engineer the device for modern processes than to run a special 'old process' batch using those films. Really the only thing the old films would be good for (aside from being a piece of history) would be the generation of simulation models to try to mimic any quirks in new silicon.

[jens]

No, MIDI has nothing to do at all with the sound chip.

I already thought I'd got it wrong.

[Bryan]

Many sound chips designed for 6502's expect to be sitting on a bus that will run the a nice predictable rate for which they were designed, with a memory access every cycle. Interfacing such a chip to a 68000 bus would be somewhat complicated. The 68000 does offer a couple features that would ensure the task wasn't totally hopeless, but it would still probably be necessary to add two or three extra chips to interface the POKEY to the 68000.

 

This is the real reason. Pokey was designed for a 6502 system running at <2 MHz. You can't just stick it on an 8MHz bus. The YM chip is designed to be flexible in its interfacing. Another reason the ST uses the YM chip are the IO ports which control various things and reduce the chip count.

 

-Bry

[Rybags]

Yeah, the I/O ports probably swayed the deal.

 

With some re-engineering, POKEY probably could have had the keyscan reworked to instead provide I/O capability.

 

To interface it, just a /4 clock probably would have worked.

 

They probably could have even devised a system for 4-bit digitized sound by having the DMA controller directly stuffing the AUDCn registers.

 

The ADSR and more linear frequency control are about the only SID-like features, IMO.

 

Yamaha did have some great synthesizer chips, but I'd imagine they were probably pretty pricey around the mid-1980's.

[ijor]

Many sound chips designed for 6502's expect to be sitting on a bus that will run the a nice predictable rate for which they were designed, with a memory access every cycle. Interfacing such a chip to a 68000 bus would be somewhat complicated. The 68000 does offer a couple features that would ensure the task wasn't totally hopeless, but it would still probably be necessary to add two or three extra chips to interface the POKEY to the 68000.

 

This is the real reason. Pokey was designed for a 6502 system running at <2 MHz. You can't just stick it on an 8MHz bus. The YM chip is designed to be flexible in its interfacing.

 

This doesn't make much sense to me. The YM chip in the ST is clocked at 2 MHz, not at 8 MHz. The 68000 in the ST already interfaces directly with very slow chips as the ACIAs. GLUE adds the neccessary wait states.

 

The 68000 is flexible enough that can interface directly with very slow devices. Furthermore, one of the main goals of the 68000 design was to make it compatible with 6800 peripherals. Almost sure that no extra chip would be needed, at least not for speed reasons.

 

Both the 68000 and ST architecture are ready to interface directly with much slower devices. The slow devices don't even need to be clocked from the same crystal. They can run asyncronous (to the main bus) from a completely separated clock.

[jedinovice]

Wow! That woke everyone up! I look forward to discussion and (possible resolution)

 

What was the true story ofthe development of the ST then, BTW? Every source I have read indicates the ST was built in reaction the coming Amiga. There is another story?

[Rybags]

Not exactly.

 

If things had gone as planned, the Amiga would have been an Atari release, and the ST as we know it would never have existed.

 

But, it would probably have ended up with a much different OS (maybe even TOS/GEM) and less functionality to get it to the price point that the early ST was able to achieve.

[carmel_andrews]

Actually the 'ST' was already being prototyped after tramiel was turfed out of CBM, he basically hijacked some of their leading engineers (cbm's that is) and started a new company (Tramiel Technology Ltd...the very same company that a few months later turned up and bought Atari from Warners)

 

The only Atari part of the ST design was the work done with DRI (who supplied the O/S)

 

I guess the only reason why tramiel didn't go the pokey route was because the ST design was pretty much set in stone and anyway i'm not too sure if Pokey was 68k compatible, although one was used in the Atari Games 'Gauntlet' game (however, it needed a 65xx compatible proccy to interface with the main 68k proccy)

 

In regards to AMY, Atari apparently sold the design to a third party (as the 8bit/xe version worked) and the 3rd parties job was to develop a 68k/ST version, which they apparently did, unfortunately there was a bit of a falling out or dissagreement and Atari decided not to go don't the AMY route (even though the 3rd party developer had threatened to releasing both the XE and ST version of the hardware)

[Rybags]

From looking at KLOV's Gauntlet schematics, the 6502 seems to be used as a sound processor. Communication to the 68000 is via a buffer chip.

 

Using a second, lower specced CPU was pretty common practice in arcade games from the mid-1980s.

 

POKEY is a fairly straightforward chip which just uses TTL voltages, so would likely work with a whole array of different processors.

[ijor]

Many sound chips designed for 6502's expect to be sitting on a bus that will run the a nice predictable rate for which they were designed, with a memory access every cycle.

 

I was checking the Pokey data sheet. As I said, the difference in speed or rate shouldn't be a problem at all. However it could be a syncronous problem. And I suppose that is what supercat was talking about "a memory access every cycle".

 

However this is still an academic issue. There were lot of other reasons already mentioned why Pokey wasn't used.

[ijor]

The ADSR and more linear frequency control are about the only SID-like features, IMO.

 

Btw, I don't know how the SID ADSR looks like, but the 2149 doesn't really have ADSR. At least not what I, and probably most people, would call it ADSR. What is has is a crude envelope shape and frequency control. In particular you cannot set the period for the different ADSR stages, actually it doesn't have all the ADSR stages.

 

And what is perhaps more annoying, is that evenlope control is common to all the voices. Any ST musician here? I would guess the 2149 envelope is used for some special effects only, and not for instrument selection. And ADSR is probably implemented in pure software?

[Gunstar]

Hmmmm... interesting to think what the ST would have been like if the company was doing well, didn't change management, and had backwards compatibility on the mind.

 

A 68000 with 6502 in tow for some basic handling of side routines and called for older programs.

A "Pokier" chip, like the Pokey, but on steriods.

An "Antics" chip, capible of multiple functions.

 

Carry on the spirit of the 8-bit into the 16/32 era, giving all that the 8-bit users wanted at the time, more power, more speed, more ram, same computer they fell in love with.

 

As it was they got more power, kinda more speed, more ram, whole different computer. (Though ST X-Former was nice!)

 

 

 

That's the way it originally was. The Amiga developer(s) was/were the Atari 800's original engineers (they had left Atari earlier to start Amiga, then Atari made a deal and supported them to develop the next generation Atari computer). And Atari Inc. had been supporting Amiga developments and the Amiga, very similiar in many ways to the 800, was to have been Atari's. But Atari Corp., in 1985, chose not to buy it from Amiga and started making the ST with some of the Commodore 64's old engineers (that left Commodore to Atari when the Tramiels bought Atari) came up with the ST in about 6 months while Amiga made a deal with Commodore. So with the Commodore&Atari upheavals in '84/85, Atari ended up with hardware that was closer to what next-gen Commodore hardware would have been if the Tramiels and their engineers had stayed at Commodore, and Commodore ended up with the next-gen Atari hardware in the Amiga. I'm not saying that the ST line IS what would have been the next Commodore since it was developed under different conditions and environment, but that it was done by some of the people that made the C64, so it would have been similair if made at Commodore. But the Amiga IS the exact technology that would have been the next-gen 16-bit Atari machine if Warner Communications had kept Atari and moved forward. This was all going on just about the same time in history as the Steve Jobs and Apple vs. Bill Gates and Microsoft feasco. A real turbulent time in the videogame AND computer industry that left everything on it's head. Of course you know that Jobs, and his partner Wozniak from Apple both started at Atari...

 

There's a lot more to all of this of course, I've over simplified. Like the Amiga was originally being designed as the next gen console, then found too powerful (they thought) for that and changed direction toward a new computer. And the while there's the culture behind the industry like the hippy engineers at Atari and later Apple(and Amiga) vs. the straight lacers at companies like IBM, Commodore and Microsoft. And who ended up with whom, what, and how much in the end. And how people like Nolan B. was among the fore runners of the whole "counter culture" electronics industry when he started Atari and electronics as entertainment and home computing vs. business.

Edited December 28, 2006 by Gunstar

[Retro Rogue]

Just some corrections...

 

That's the way it originally was. The Amiga developer(s) was/were the Atari 800's original engineers (they had left Atari earlier to start Amiga, then Atari made a deal and supported them to develop the next generation Atari computer).

 

Such support was cancelled before the Tramiels took over.

 

But Atari Corp., in 1985, chose not to buy it from Amiga and started making the ST with some of the Commodore 64's old engineers (that left Commodore to Atari when the Tramiels bought Atari) came up with the ST in about 6 months while Amiga made a deal with Commodore.

 

Again, that's more of Mical's missinformation. Jack and his Atari Corp. didn't find out about the Atari/Amiga contract material until after Commodore had already purchased Amiga. Jack bought Atari Consumer for manufacturing and distribution capabilities for his computer - which was 95% done by the time he took over and formed Atari Corp. that early July. (This further fueled the notion that the ST was based on Shiraz's designs for a similar Commodore model previous to leaving, adding credence to Commodore's position that material had been stolen). There was never any decision involved with regards to the Amiga being used or not being used. What's being confused was Jack's visit several months earlier (while still with his Tramel Technologies Ltd.) as a possible investor in Amiga, which never materialized (he wanted to replace most of the staff).

 

But the Amiga IS the exact technology that would have been the next-gen 16-bit Atari machine if Warner Communications had kept Atari and moved forward.

 

Again, not the case. The Amiga related material and a series of other computer proto developments were cancelled by James Morgan during his move to return Atari to its bread and butter (gaming) in an effort to turn the large losses around. It would be more fair to say the Amiga would have been the next generation Atari *if* Jay and Joe had stayed at Atari Inc. after developing the PCS's.