Another missed opportunity by Atari?

[calimero]

1 hour ago, zzip said:

But do hardware designers even get royalties?   I read that they did not want to be under Jack, who didn't have a great reputation in the industry even then, so that helped motivate them to sell to Commodore.

Jack Tramiel many time mention that he "made more millionaire  than any other companies".

 

As proof to this claim I would add this link: http://mcurrent.name/atarihistory/tramel_technology.html that show that many Commodore employees follow Jack to new, uncertain, company: TTL (Tramel Technology, Ltd.).

 

On other end, there is TWO person from Amiga team that spread (was spreading) negative sentiment about Jack Tramiel: 

 

- Robert J. Mical

- Dave Needle

 

Only two of them, from whole Amiga team, SPREAD LIES about Jack Tramiel. 

Both of them work on handle Lynx console that eventually Atari Corp. bought so I can understand their agenda. 

 

 

btw 

Dave Morse motivation deserve complete new thread...

Edited November 13, 2020 by calimero

[leech]

3 hours ago, zzip said:

But do hardware designers even get royalties?   I read that they did not want to be under Jack, who didn't have a great reputation in the industry even then, so that helped motivate them to sell to Commodore.

This was pre-Warner Communication sale.  So before Nolan Bushnell decided he was worn out being the boss.

[zzip]

On 11/13/2020 at 4:46 PM, calimero said:

STe come out in 1989.

and 1989. was not certainly post-VGA.

VGA start to blossom after 1990. Probably in early 1991. 

VGA came out in 1987.

 

ST was roughly on par with EGA graphics, better in some ways, worse in others.   EGA came out in 1984, ST in 85.   So ST graphics were pretty current in 85.

 

When STe came out, VGA was older than EGA was when ST came.   Even if VGA-level was too expensive,  it could have used an extra bit-plane or two to bring it to 32 or 64 colors and bring it closer to Amiga.

 

[zzip]

On 11/13/2020 at 4:36 PM, calimero said:

Beside: SGI, Sun, HP... and all other companies  (just like Atari and Commodore) FAIL after Microsoft introduced Windows NT (aka VMS for PC). 

There was consolidation in that market after NT,  but really what killed them was Linux.    Why spend pay all that money on proprietary hardware and Unix flavors when Linux was out-performing them running on common PC hardware?

[calimero]

On 11/16/2020 at 3:55 PM, zzip said:

VGA came out in 1987.

 

ST was roughly on par with EGA graphics, better in some ways, worse in others.   EGA came out in 1984, ST in 85.   So ST graphics were pretty current in 85.

 

When STe came out, VGA was older than EGA was when ST came.   Even if VGA-level was too expensive,  it could have used an extra bit-plane or two to bring it to 32 or 64 colors and bring it closer to Amiga.

The STe really needed to break that limitation in 89.  I think this was the STe's Achilles-heel in an otherwise great system.  It wouldn't necessarily have needed to do all the modes VGA did, but more colors in the modes ST supported would have been much welcomed, and supported the same monitors.

Only way for ST architecture to have more than 16 colors is like Atari did it in TT.

So STe with 32 or 64 colors would need to have TT bus width and TT shifter speed. It would be a TT without 68030

Note that in TT resolution does not affect CPU speed like in ST. TT is basically ST just scaled up in chip speed and bus widths.

 

Second way to have more colors is to make something like Falcon and Videl.

 

And third way is to have separate memory for video. Like VGA or Macs with vram.

- Please check prices of VGA cards back in 1987. I believe that they become standard (cheap enough) after/at 1991.

 

 

btw Shiraz Shivji talk about EST (probably TT later) in 1987. but he left Atari in same year (iirc). Roy God came as Shiraz replacement and EST/TT project get prolong for at least two years (I believe because this shift Shiraz-Roy) before it come to market...

 

If anyhting, I would say that Atari should release Falcon earlier... it is based on STe anyway.

[zzip]

4 hours ago, calimero said:

So STe with 32 or 64 colors would need to have TT bus width and TT shifter speed. It would be a TT without 68030

Note that in TT resolution does not affect CPU speed like in ST. TT is basically ST just scaled up in chip speed and bus widths.

 

Second way to have more colors is to make something like Falcon and Videl.

 

And third way is to have separate memory for video. Like VGA or Macs with vram.

- Please check prices of VGA cards back in 1987. I believe that they become standard (cheap enough) after/at 1991.

There was also the forgotten MCGA standard that IBM released at the same time as VGA.   It was weaker and presumably cheaper than VGA.   It had capabilities similar to the ST, except one mode which was better:  a 320x200 mode in 256 colors from a palette of 262,144.

 

obviously there were going to be technical challenges doing it.  But the PC architecture had way more challenges, and they always found ways to overcome its limits.

 

4 hours ago, calimero said:

 

btw Shiraz Shivji talk about EST (probably TT later) in 1987. but he left Atari in same year (iirc). Roy God came as Shiraz replacement and EST/TT project get prolong for at least two years (I believe because this shift Shiraz-Roy) before it come to market..

I suspect this is the real reason.   Shiraz helped bring the ST from concept to market really quickly.   But then enhancements were always slow to come, and always seemed to fall short of what people wanted

 

4 hours ago, calimero said:

If anyhting, I would say that Atari should release Falcon earlier... it is based on STe anyway

And that would be fine too.  But as I said above, enhancements were too slow coming after the original ST release.  Also I can't imagine how expensive that thing would be in like 1990? 

[Xebec]

Was a 68000 - 16 MHz actually significantly more expensive than an 8 MHz 68000 by 88-89 ?

 

The 68000 8 MHz debuted many many years earlier..  

 

EDIT:  And why did Shiraz leave Atari in 87?

Edited November 22, 2020 by Xebec

[leech]

1 hour ago, Xebec said:

Was a 68000 - 16 MHz actually significantly more expensive than an 8 MHz 68000 by 88-89 ?

 

The 68000 8 MHz debuted many many years earlier..  

 

EDIT:  And why did Shiraz leave Atari in 87?

I wouldn't have thought so.  I mean I tend to think that if it weren't for the potential compatibility issues, they should have packaged the Mega STe with 68020 and TT modes, at the very least.  Making it a cost reduced version of the TT.

[zzip]

13 hours ago, Xebec said:

Was a 68000 - 16 MHz actually significantly more expensive than an 8 MHz 68000 by 88-89 ?

 

The 68000 8 MHz debuted many many years earlier..  

 

EDIT:  And why did Shiraz leave Atari in 87?

I would think that with 68020's and 030's coming to market, it would push down the cost of the 68000, even at higher clock frequencies.

 

I know comparing ST's to PC's based on mhz is like comparing apples and oranges,  but the general public didn't know better, and they were constantly bombarded with PC ads showing 10mhz, then 12mhz, 16, 20, etc.   Made it seem like the ST was being left in the dust.   A clock boosted ST with 'turbo switch' for compatibility could have helped with this impression.

[calimero]

On 11/23/2020 at 12:18 AM, Xebec said:

Was a 68000 - 16 MHz actually significantly more expensive than an 8 MHz 68000 by 88-89 ?

 

First MC68000 16MHz was manufactured at the end of 1989. http://www.cpu-collection.de/?l0=co&l1=Motorola&l2=68000

 

STe have 16MHz MC68000 but if you turn off cache there is almost no difference to ST (probably) so you can "stiff" faster clocked CPU in ST and expect that everything will run faster. This is because ST was very tightly planned, and back in 1984. it was really extremely fast thanks to architecture that allow MC68000 8MHz to work at full speed. In contrast, Macintosh hardware was example of bad planning and designing. At same clock speed, Mac is almost 1/3 slower than ST (and with 1/3 lower resolution)!

 

On 11/23/2020 at 12:18 AM, Xebec said:

EDIT:  And why did Shiraz leave Atari in 87?

Shiraz left Atari for Momenta Corp.

 

"After leaving Atari, Shivji designed the Momenta Pen Computer in 1991. This was one of the first full- sized tablet computers with a sophisticated hardware for that time. However, its operating system was based on DOS and did not deliver sufficient applications for its graphical user interface."

 

https://www.microsoft.com/buxtoncollection/a/pdf/Shiraz_Shivji.pdf

 

http://web.archive.org/web/20070110182526/www.byte.com/art/9611/sec4/art1.htm

 

This probably latest article about Shiraz Shivji: https://www.digisaurier.de/computerhelden-2-shiraz-shivji-der-mann-der-uns-den-atari-st-schenkte/ 

[ParanoidLittleMan]

Idea of faster 68000 (than 8 MHz) constantly comes up here. But you can not simply put faster CPU in some concrete HW design, components with limited speed, and expect that then it can work 2x faster. Because everything around CPU is designed for 8 MHz 68000. Would need faster RAM in first place, and that means higher price. And of course other chips need to be faster - again higher price ... 

Finally, 16 MHz 68000 arrived in Mega STE, in late 1991, but they used less expensive solution - instead complete new chipset (2x faster) it is practically same - so same RAM, MMU, DMA, shifter. External CPU cache of 16 KB was used, and that was pretty efficient  for ST(E) SW - because most of CPU activity is RAM access, and it resulted in average speed gain about 70-80% when CPU was set to 16 MHz. Just to add here there there are diverse shallow writings about that gain - like 15-20% only - maybe they forgot to activate cache, and only set CPU to 16 MHz ? I can say that it was wise design (even without Shiraz Shivji), and surely much less additional cost than whole design for 2x CPU speed.

[leech]

46 minutes ago, ParanoidLittleMan said:

Idea of faster 68000 (than 8 MHz) constantly comes up here. But you can not simply put faster CPU in some concrete HW design, components with limited speed, and expect that then it can work 2x faster. Because everything around CPU is designed for 8 MHz 68000. Would need faster RAM in first place, and that means higher price. And of course other chips need to be faster - again higher price ... 

Finally, 16 MHz 68000 arrived in Mega STE, in late 1991, but they used less expensive solution - instead complete new chipset (2x faster) it is practically same - so same RAM, MMU, DMA, shifter. External CPU cache of 16 KB was used, and that was pretty efficient  for ST(E) SW - because most of CPU activity is RAM access, and it resulted in average speed gain about 70-80% when CPU was set to 16 MHz. Just to add here there there are diverse shallow writings about that gain - like 15-20% only - maybe they forgot to activate cache, and only set CPU to 16 MHz ? I can say that it was wise design (even without Shiraz Shivji), and surely much less additional cost than whole design for 2x CPU speed.

My first ST was the Mega STe, and I loved it.  For things that were happy with 16mhz with cache, it was really fast.  But there were a lot of things that were unhappy with cache, but fine at 16mhz, and still ran faster, but you are right, it is never really a 2x the speed. 

There is a reason the machine defaults to 8mhz, no cache though, so that mire software would work from floppy disk.  It wasn't until probably late 89, early 90s when games would even come with hard disk installers.  best was when I installed Ultima 6, then copied it into my 2mb ram disk after upgrading to 4mb.  Load times were almost gone.  It would stutter a bit between load zones, and that was it!

But having the Mega STe be the cheaper TT rather than being a more expensive STe, put it in a weird position, I think. 

 

On the note about tight timing of the other chips, the Amiga seems to have a lot of benefit from accelerators, but at the same time I think it still doesn't really make AGA faster.  Which of course makes it sort of sluggish for certain things in Workbench.

[ParanoidLittleMan]

Cache for RAM self can be cause for some SW not running properly, but in most cases the reason for problems is actually the desired faster work - SW is simply not ready for it, and will not work properly if CPU runs not at regular 8 MHz . Yeah, programmers often made such code, and it was OK for 99% of users.

There is even SW what need cycle accurate execution - like Spectrum 512 or other hi-color displaying SW (Photochrome) .

[leech]

1 hour ago, ParanoidLittleMan said:

Cache for RAM self can be cause for some SW not running properly, but in most cases the reason for problems is actually the desired faster work - SW is simply not ready for it, and will not work properly if CPU runs not at regular 8 MHz . Yeah, programmers often made such code, and it was OK for 99% of users.

There is even SW what need cycle accurate execution - like Spectrum 512 or other hi-color displaying SW (Photochrome) .

Spectrum 512 on an STe ran weird even at 8mhz, as it really was meant to be ran with the ST's color palette.  Most thingsooked right, but occasionally they would be messed up.

[Xebec]

On 11/24/2020 at 1:19 PM, calimero said:

 

First MC68000 16MHz was manufactured at the end of 1989. http://www.cpu-collection.de/?l0=co&l1=Motorola&l2=68000

 

STe have 16MHz MC68000 but if you turn off cache there is almost no difference to ST (probably) so you can "stiff" faster clocked CPU in ST and expect that everything will run faster. This is because ST was very tightly planned, and back in 1984. it was really extremely fast thanks to architecture that allow MC68000 8MHz to work at full speed. In contrast, Macintosh hardware was example of bad planning and designing. At same clock speed, Mac is almost 1/3 slower than ST (and with 1/3 lower resolution)!

 

Shiraz left Atari for Momenta Corp.

 

"After leaving Atari, Shivji designed the Momenta Pen Computer in 1991. This was one of the first full- sized tablet computers with a sophisticated hardware for that time. However, its operating system was based on DOS and did not deliver sufficient applications for its graphical user interface."

 

https://www.microsoft.com/buxtoncollection/a/pdf/Shiraz_Shivji.pdf

 

http://web.archive.org/web/20070110182526/www.byte.com/art/9611/sec4/art1.htm

 

This probably latest article about Shiraz Shivji: https://www.digisaurier.de/computerhelden-2-shiraz-shivji-der-mann-der-uns-den-atari-st-schenkte/ 

I do recall reading that Motorola was afraid to introduce 'too high of clock speed' 68000 CPUs to avoid overshadowing perceived performance gains of 68020/68030, and this did effectively hurt designs based on that CPU from being competitive with the decreasing price of x86 CPUs over time.

 

According to wikipedia, Motorola did have a 12.5 MHz 68000 by June 1982, though the 16 MHz (16.67 Mhz actually) "was not available until the late 80's".   The 68010 was released at up to 25 MHz speeds and was highly compatible (though not 100%).  

 

..

 

EDIT:  Thanks for the additional links - they provide good info on Shivji's next steps after Atari.  It's unfortunate he disagreed with the Tramiels and ran into health problems - he seems like he was a great engineer and leader for Atari..    Interesting this source says Shivji stayed with Atari to 1990..   I don't know why I had 87 stuck in my head.  

 

..

 

It really seems like for the ST or Amiga to have succeeded we would have needed a lot of things.  A more successful and better Motorola, less than 3 68000 based computers competing (i.e. Amiga or ST to get a clear victory early on or one of them to never launch to effectively compete against the Mac mindshare), and then a lot of luck/good leadership on top of that.  

Edited December 2, 2020 by Xebec

[zzip]

45 minutes ago, Xebec said:

It really seems like for the ST or Amiga to have succeeded we would have needed a lot of things.  A more successful and better Motorola, less than 3 68000 based computers competing (i.e. Amiga or ST to get a clear victory early on or one of them to never launch to effectively compete against the Mac mindshare), and then a lot of luck/good leadership on top of that. 

Maybe if IBM had chosen 68000 instead of Intel for the PC then the Motorola ecosystem would have seen the rapid advancement and falling costs we saw in the Intel world, and that would have boosted ST and Amiga.   PC compatibility would have been easier to implement too. 

The 68000 was in the running for PC, but I think IBM was concerned about supply issues.

[calimero]

Shiraz left Atari in 1989. 
 

https://mcurrent.name/atarihistory/tramel_technology.html

 

search this text for: shiraz, roy good ...

[AtariNostalgia]

On 12/3/2020 at 12:13 AM, zzip said:

Maybe if IBM had chosen 68000 instead of Intel for the PC then the Motorola ecosystem would have seen the rapid advancement and falling costs we saw in the Intel world, and that would have boosted ST and Amiga.   PC compatibility would have been easier to implement too. 

The 68000 was in the running for PC, but I think IBM was concerned about supply issues.

Well, Motorola had Apple. Granted it was no IBM+clones, but it did allow for innovation by injecting cash to them. I don't think its Motorola's fault as much as Ataris'

[ParanoidLittleMan]

IBM did selection of CPU probably in 1980, since PC was launched in 1981. Most common explanation is that MC68000 was still with some bugs then.

And I guess that 8086/88 option was cheaper. Actually, considering prices of 16-bit CPUs then, that was pretty much diff.

[AtariNostalgia]

Well, they could have used the Motorola 6809 instead of the 8088, it was produced in 78.

[ParanoidLittleMan]

9 hours ago, AtariNostalgia said:

Well, they could have used the Motorola 6809 instead of the 8088, it was produced in 78.

6809 is 8-bit CPU, with little 16-bitness. Definitely not for some not cheap IBM PC project.  While 68000 was pretty much advanced 16-bit one, with 32 bit registers, linear addressing in whole address space, surely better than 8086, but as said, it costed much more too.

[AtariNostalgia]

I was comparing the 6809 to the 8080.

[leech]

1 hour ago, AtariNostalgia said:

I was comparing the 6809 to the 8080.

Speaking of processors, I found it interesting in my 65816 Assembly language book, that they were planning on releasing a 32bit 65832 that would have been compatible with the 6502.  The lack of widespread use of the 65816 kind of killed the plans for it.

Would have loved a 32bit 800xl.  ?

[Xebec]

On 12/2/2020 at 5:13 PM, zzip said:

Maybe if IBM had chosen 68000 instead of Intel for the PC then the Motorola ecosystem would have seen the rapid advancement and falling costs we saw in the Intel world, and that would have boosted ST and Amiga.   PC compatibility would have been easier to implement too. 

The 68000 was in the running for PC, but I think IBM was concerned about supply issues.

That's an interesting change I hadn't considered.  If the PC had gone 68000 - yes, it's possible x86 would have died a slow death in the 1990s...  software emulation of a PC on the ST or Amiga would have been much faster/easier too ..    

 

I do also remember reading IBM considered the 68000 (as well as re-purposing the Atari 400/800 supposedly), but I think it was both supply of the CPU and also 'companion chips' not being ready in time for the launch of the IBM PC.  I also suspect there was a cost element there since the minimum 16-bit bus and related peripherals would impact that too.  

[Xebec]

12 hours ago, leech said:

Speaking of processors, I found it interesting in my 65816 Assembly language book, that they were planning on releasing a 32bit 65832 that would have been compatible with the 6502.  The lack of widespread use of the 65816 kind of killed the plans for it.

Would have loved a 32bit 800xl.  ?

There was a brief comment from Bill Mensch during a recent VCF video where he said he thinks he might still have the starting paperwork he (and someone else) did for the 32-bit 65816 variant.  It sounded like they may have only gotten as far as the instruction set ..  but it's something that can hopefully be preserved... 

 

Sometime towards the end of this video IIRC   (3rd video in the playlist - ignore the Dave Haynie /Amiga face )