Atari v Commodore

[+CharlieChaplin]

It would have been great if a redesigned Antic/GTIA made it into the XL machines ( or even the XE machines )

 

Well,

there were redesigned Antic/GTIA chips - some of them exist only on paper, some as prototypes. Maybe you have heard something from Curt Vendel and his Atari museum ?!? Once upon along ago, he listed some chips here at atari-age forum that were planned to be in Atari computers around 1984/1985, like Keri (Antic+GTIA in one chip) and other chips. But as we all know, when the video game market (and also the home computer market) crashed in 1984, all these works were stopped. And when Tramiel took over Atari he was not interested anymore in these 8bit thingies...

 

Its not random luck, that we can have stereo with two Pokey chips, Atari already had plans for that, as well as for many other upgrades and updates of the XL computer line and last not least the Sweet 16 project... -Andreas Koch.

[supercat]

80 columns = twice the memory bandwidth needed. That's by far no simple change to a system, that's a completely new design.

 

When does what data get fetched on the A8? The A8 performs two memory cycles in the time required to display each character, but from my understanding it does not interleave memory cycles with the CPU the way the C64 and Apple II do. Are the bytes required for each line fetched in a burst, or what? It would seem the A8 should have enough memory bandwidth to at least handle 320-wide 4-color graphics or 160-wide 16-color graphics.

[+Stephen]

Well,

there were redesigned Antic/GTIA chips - some of them exist only on paper, some as prototypes. Maybe you have heard something from Curt Vendel and his Atari museum ?!? Once upon along ago, he listed some chips here at atari-age forum that were planned to be in Atari computers around 1984/1985, like Keri (Antic+GTIA in one chip) and other chips. But as we all know, when the video game market (and also the home computer market) crashed in 1984, all these works were stopped. And when Tramiel took over Atari he was not interested anymore in these 8bit thingies...

 

Its not random luck, that we can have stereo with two Pokey chips, Atari already had plans for that, as well as for many other upgrades and updates of the XL computer line and last not least the Sweet 16 project... -Andreas Koch.

 

Well, the combined ANTIC/GTIA didn't offer any new features. It simply crammed both chips on one die.

 

I don't know what happened to Atari - the 2600 progressed nicely into the 400/800 series, and then - nothing. It's a real shame Atari let Miner & Co go. The ST line was cobbled together with off the shelf parts, the 7800 was designed out of house, the Lynx was designed by Epyx, the Jaguar was designed by Flare 1. I don't understand how a company as innovative as Atari was from 1972 to 1979 managed almost nothing afterwards.

 

Stephen Anderson

[Tezz]

well, after that famous 83 crash you could kinda understand that the hey days of inovation were gone. The original Atari guys left early on and the later years of Atari were just really the same name repackaging an old favorite in the bid to make money as a business with no interest in progressing hardware. It is the sad part of Atari's history

Edited November 24, 2008 by Tezz

[Crazyace]

80 columns = twice the memory bandwidth needed. That's by far no simple change to a system, that's a completely new design.

 

When does what data get fetched on the A8? The A8 performs two memory cycles in the time required to display each character, but from my understanding it does not interleave memory cycles with the CPU the way the C64 and Apple II do. Are the bytes required for each line fetched in a burst, or what? It would seem the A8 should have enough memory bandwidth to at least handle 320-wide 4-color graphics or 160-wide 16-color graphics.

 

Antic already fetches 80 bytes in character mode lines ( not every line though ) - so it could easily handle 160x16 colour with no change to the memory system.

 

Given the Antic architecture - 16 colours pallete rather than a colour ram would be more suitable..

 

Make 16 palletes .. BK, PF0 - PF3 ... [ PF4,PF5,PF6 ]

8 players ( no missiles ) P0 - P3 .. [ P4 - P7 ]

For mono text select colour pairs, rather than colour+BK

For 4 colour text select ) BK/PF0-PF2 and PF4-PF6/PF3 - two groups of 4 colours, rather than 5

and 16 pallette entries for GTIA modes

 

It would be a cool design maybe some FPGA wizard could implement it... ( or maybe Curt knows of some hidden prototype already )

[supercat]

I don't know what happened to Atari - the 2600 progressed nicely into the 400/800 series, and then - nothing.

 

Same thing happened, IMHO, with the Amiga. Commodore did finally release a 32-bit chipset, but only after the Amiga had long since been surpassed from every side.

[TMR]

And i believe i can claim the first hand-drawn logo in DFLI (above the scroller in this screenshot):

Nice one cool stuff. Good to see that a technique like FLI is posible on the plus4. Was it used outside of the demoscene?

 

Not to date, it's a relatively recent discovery and that first hand drawn logo of mine was actually drawn before there were any editors for it - i drew it with DPaint and then converted the cells to two colours for conversion before adding the colours back in through the source code! A game that uses DFLI, something like a picture shuffler or possibly a Tetris is workable but, since it only leaves free time in the borders, running software sprites over the top is less viable for action games.

[frogstar_robot]

. I don't understand how a company as innovative as Atari was from 1972 to 1979 managed almost nothing afterwards.

 

Stephen Anderson

 

They were innovative under Bushnell. It isn't that Bushnell was a particularly inspired leader but he did seem to know how to let the engineers get on with designing something. The Video Music was not the product of an anal-retentive company. And while it wasn't a huge success in and of itself, the tech used seemed to segue into the 2600 pretty nicely. In that era, marketing and management centric computer companies weren't going to do well making consumer devices because the market hadn't been defined yet so letting the engineers throw things at the wall and watching carefully what stuck was the best way to go about it.

 

Warner Bros was a media company pure and simple and obviously wanted to market games like they marketing music and movies. Games at that time hadn't fused with movies and music and couldn't be marketed or developed that way. It was during the Warner tenure that so many of Atari's best and brightest left to start their own companies. Jay Miner's engineering team and the programmers who went on to form companies like Imagic and Activision were Atari (and I find Commodore partisans who praise the Amiga and C-64 while despising the A8 to be ESPECIALLY amusing given that the A8 and the Amiga shared so many core designers. If Warner Atari had a bit more foresight, the Amiga would have been an Atari as well as they had the opportunity to buy that tech from Miner's company. I'll also note that the core of the ST engineering team worked on Commodore projects going clear back to the calculator days and yes the PET, Vic-20, C-64 as well.) At least they were the aesthetic soul of Atari and the ones left had to be content to do what they were told by PHBs.

 

Tramiel seemed to want to make Atari the IKEA of computers. It worked in the 70s and early 80s but at this point the business uses of micros were beginning to predominate. The only innovations the Tramiels were interested in were cost cutting ones.

[Oswald]

ANTIC has a lot more modes than just 160*200

 

? so does the c64.

 

Your mental concoction that higher bus width of the VIC2 means superior chip is not true.

 

why not ? it does display more information. more information makes better pictures. and better sprites. a8 can display without cpu intervention 4 color bitmaps and 5 colors in multi textmode, 1 color in hires textmode. c64: 3 own colors in ech char cell in bitmap mode, 3 overall colors + 1 own + 2x the chars in multicolor charmode, and unique color for each char in hires textmode. each mode is better than the comparable a8 ones _because_ they use up more information.

 

The C64 is slower.

 

start coding a 2d game. the features of the VIC&SID will support the cpu so much that actually you will get more performance out of it than a8. in other scenarios (3d, isometric games, etc) ofcourse the a8 wins.

 

You claiming Jay Miner did not meant to use tricks is clearly false as well. Just look at the Amiga with its HAM mode.

 

I dont remember ever claiming that. qoute ? anyhow forget the amiga, its a8 vs c64.

 

You can get higher repaint/refresh rates and lower memory useage for applications that don't need 16 colors per scanline for their applications.

 

I can get that on the c64 aswell. if I dont need 16cols/line then I dont use it. the c64 modes have as a subset more primitive modes obviously. if I have a 320x200x256 screen I can make it look like 80x100x4, 160x200x2, or whatever. 320x200x256 is better than a bunch of dumber modes. the same way antic's many modes doesnt make it better for gfx than VICII.

[Oswald]

Overscan not useful?

 

Then why do more than half the C-64 demos out there get the C= owners dribbling with joy over something as trivial (to an Atarian) as sprites in the border?

 

IMO, one of the biggest shortcomings by far of the C-64 was the 320x200 fixed architecture. Using sprites in the border is useful only in the context of things like status displays for games.

 

its just a technical stunt, appart from that its not useful. and why is 320x200 is a shortcoming? how many % of the a8 games use overscan ?

 

the biggest shortcoming by far of the c64 is the slow cpu believe me. but if you look at it with a perspective its the price it pays for having VICII being able to do all the sprites, and colors etc.

[Oswald]

I find Commodore partisans who praise the Amiga and C-64 while despising the A8 to be ESPECIALLY amusing given that the A8 and the Amiga shared so many core designers.

 

its even more amusing how atari fans are trying to byte of pieces of amiga's fame for their 8 bit machine. while the architecture is similar, amiga architecture doesnt really worked in the 8 bit world, there was not enough bandwidth and memory. a better compromise was to have a color attribute ram to color a bunch of pixels with the same colors, then another bunch of pixels with different colors, rather than widening the pixels for more colors, and even then why oh why there wasnt 16 color registers to set up a palette out of 128?

[emkay]

start coding a 2d game. the features of the VIC&SID will support the cpu so much that actually you will get more performance out of it than a8. in other scenarios (3d, isometric games, etc) ofcourse the a8 wins.

 

 

No counterstrike here. It's an almost correct statement.

[Crazyace]

Yes, comparing a8 to Amiga is a bit pointless in this thread

 

The separate colour ram in the c64 is a real strength - it would have been difficult to add to the a8 without completely changing the design philosophy..

[emkay]

its even more amusing how atari fans are trying to byte of pieces of amiga's fame for their 8 bit machine. while the architecture is similar, amiga architecture doesnt really worked in the 8 bit world, there was not enough bandwidth and memory.

 

Ofcourse it would work, because the copper and paula chip would have freed all cpu cycles even on a 3,6MHz clocked 6502... But a 7.2Mhz Motorola was throttled down by the memory acces of the AGNUS chip.

Copper was a a logical step ahead of the DL/DLI usage

Paula was a logical step ahead of POKEY. While Pokey had simple "Atari waves", PAULA was created to have fully customized waveforms at better volume and frequency resolutions. Thinking about the time, when the chips were in development (1980-82) togehter with an 8bit 3,6MHz 6502 and looking at the clocking ratio of the chipset, everything fits. Jay Miner learned his lessons, ATARI didn't.

 

 

a better compromise was to have a color attribute ram to color a bunch of pixels with the same colors, then another bunch of pixels with different colors, rather than widening the pixels for more colors, and even then why oh why there wasnt 16 color registers to set up a palette out of 128?

 

Actually, they intentionally reduced hires colours, seeing the limits of an NTSC TV.

Don't forget, the system was firstly released in a time when Intel and Apple didn't know what colours have to do on a Computer-Screen...

[emkay]

Yes, comparing a8 to Amiga is a bit pointless in this thread

 

The separate colour ram in the c64 is a real strength - it would have been difficult to add to the a8 without completely changing the design philosophy..

 

 

?

 

GTIA has enough free registers and adress range to bind a Radeon 4500 series chip without changing the design philosophy. All necessary was the sparated Screen RAM and some added logics to the GTIA. No changes on Antic or the System-Bus were needed.

[Crazyace]

Yes, comparing a8 to Amiga is a bit pointless in this thread

 

The separate colour ram in the c64 is a real strength - it would have been difficult to add to the a8 without completely changing the design philosophy..

 

 

?

 

GTIA has enough free registers and adress range to bind a Radeon 4500 series chip without changing the design philosophy. All necessary was the sparated Screen RAM and some added logics to the GTIA. No changes on Antic or the System-Bus were needed.

 

Now that's a stupid idea - binding a seperate video chip in via the GTIA ports would neglect the reason for Antic - You would actually end up with a design similar to the TI/MSX/Colecovision etc. For a redesign by Atari for XL or XE - changes to GTIA/Antic would be way easier.

 

( It's not such a stupid idea if you're trying to patch in another board to a fixed design - like the videoboard )

[Fröhn]

When does what data get fetched on the A8? The A8 performs two memory cycles in the time required to display each character, but from my understanding it does not interleave memory cycles with the CPU the way the C64 and Apple II do. Are the bytes required for each line fetched in a burst, or what? It would seem the A8 should have enough memory bandwidth to at least handle 320-wide 4-color graphics or 160-wide 16-color graphics.

The A8 interleaves Antic bitmap DMA cycles with CPU cycles too. In display areas every 2nd cycles (or less, if you choose one of the low resolution modes) is used by the Antic, so in the display area the CPU is effectively running at half speed. Then at the beginning of a rasterline there are 9 DRAM refresh cycles interleaved with the Antic bitmap DMA. In character mode it gets even more complicated: Every 8th rasterline there are 40 additional cycles used to fetch the character index data, during that time the CPU is totally halted.

 

On C64 it's similar (40 bitmap byte fetches interleaved with CPU cycles, every 8th rasterline 40 additional fetches), with the difference that the remaining free cycles of the 1.97 MHz bus cannot be used by the CPU.

 

Antic already fetches 80 bytes in character mode lines ( not every line though ) - so it could easily handle 160x16 colour with no change to the memory system.

Yeps but on those lines there the CPU has to be stopped completely. For 80 characters you would have to do twice than that: 80 bitmap bytes each line and every 8th rasterline another 80 for the character indices.

[atariksi]

80 columns = twice the memory bandwidth needed. That's by far no simple change to a system, that's a completely new design.

 

When does what data get fetched on the A8? The A8 performs two memory cycles in the time required to display each character, but from my understanding it does not interleave memory cycles with the CPU the way the C64 and Apple II do. Are the bytes required for each line fetched in a burst, or what? It would seem the A8 should have enough memory bandwidth to at least handle 320-wide 4-color graphics or 160-wide 16-color graphics.

 

Yeah, the A8 can fetch 80 bytes of Graphics data in one scanline easily as it does graphics+char data already. Just won't be able to have a text mode then if using the graphics mode. But then again, you can just write a dlist to incorporate the text mode in any of the scan lines where it was needed. According to some tech refs, the refresh cycles are reduced when it does fetch 80 bytes so there are still free cycles remaining on the scanline so the CPU is not completely stopped for the scanline. Also, 3 extra DMA cycles would fit in even at that mode for the 8 sprite + 8 missile mode which I suggested.

[Heaven/TQA]

question... when everything is there... Why did Jeff Minter & team not unleashed the power? there must be a reason...

[atariksi]

>>ANTIC has a lot more modes than just 160*200

 

>? so does the c64.

 

There's the built in modes 0..F then the GTIA enabled modes chosen from 0..F, customized modes via DLISTs, etc. Some are mentioned here: http://www.atarimagazines.com/v3n5/allmodes.html.

 

>>Your mental concoction that higher bus width of the VIC2 means superior chip is not true.

 

>why not ?

 

You decided to answer without reading the rest of the post.

 

>>You claiming Jay Miner did not meant to use tricks is clearly false as well. Just look at the Amiga with its HAM mode.

 

>I dont remember ever claiming that. qoute ? anyhow forget the amiga, its a8 vs c64.

 

Can't understand a simple analogy that 6-bit HAM is better than 8-bit VGA and think now I'm talking about amigas. In post #111, you were talking about Jay Miner. In post #151, you stated you had nothing else to state. I wonder what I'm replying to.

[atariksi]

>>>>>>a better compromise was to have a color attribute ram to color a bunch of pixels with the same colors, then another bunch of pixels with different colors, rather than widening the pixels for more colors, and even then why oh why there wasnt 16 color registers to set up a palette out of 128?

 

>Actually, they intentionally reduced hires colours, seeing the limits of an NTSC TV.

>Don't forget, the system was firstly released in a time when Intel and Apple didn't know what colours have to do on a Computer-Screen...

 

Actually, he's taking the msg out of context (probably did not have time to read all the posts). The suggestions of improving 8-bit machine started when supercat suggested having a bigger palette on C64. I think it's even more to the credit of the A8 that even though ANTIC could still be "finished" or optimized for more functionality, it still beats out the VICII in color content in every one of it's modes and has more modes and allows for the combinations of the modes with various screen widths (narrow, standard, wide).

[atariksi]

aehm... and please upgrade my Atari with

 

- 8 24x21 mono or 12x21 sprites multi

- colour ram per cell

- 256/512 chars font

- "shadow ram"

- hscrol per highres pixel

 

that would be enough for me...

 

btw. you don't need to have Wsync as there are ways around, DLI = Raster Interrupt

btw. more text modes... the c64 has already more text modes than we Atarians know... check out www.codebase64.org

 

I thought you were also a good Atari programmer. You can change the font at any DLI. 512 chars-- how are you selecting the char using 8-bit value? Yeah, great they have Raster interrupt, but Atari has the DLI, IRQs, and simply writing cycle exact code that stays in sync. As supercat suggested, all those color registers would be lot more useful if they had a good palette. If they kept adding sprites, they would end up with more color registers than colors!

[atariksi]

ORed color is less restrictive than shared colors among all sprites. If you think it out beforehand and are just using it to add color to an image, you can set the color registers in such a way so the ORed color is some unrelated color.

 

And if you're using them as intended (keep it simple, remember?) you can't do things like have white as a brightest colour on sprites and have limitations as to what luminances you use for the other two colours because otherwise they don't produce the right value for the OR'd bright colour. That's far more limiting than a couple of global colours, especially when there's nothing to stop two sprites being overlaid for a four colour object and it'll still leave more sprites on a scanline...

 

I have been playing around with 160*200*8, 160*200*10, and 160*200*11 modes (w/o interlace) on Atari and I don't have any problem setting up my palette using the OR player mode. For example, set register 704 to 2, 705 to 4 and the OR will be 6 and you have 3 shades of gray using 2 indices. Or try 704 to 8, 705 to 4, and OR will give you 12, etc. etc. C64 would have been better off doing the OR or some logical operation rather than having two fixed color registers. So on Atari if I have two multicolored sprites, I can get 6 unique colors. On C64, if I use two multicolored sprites I can get 4 unique colors.

[Oswald]

>>Your mental concoction that higher bus width of the VIC2 means superior chip is not true.

 

>why not ?

 

You decided to answer without reading the rest of the post.

 

does the antic have HAM mode or what?

 

>>You claiming Jay Miner did not meant to use tricks is clearly false as well. Just look at the Amiga with its HAM mode.

 

>I dont remember ever claiming that. qoute ? anyhow forget the amiga, its a8 vs c64.

 

Can't understand a simple analogy that 6-bit HAM is better than 8-bit VGA and think now I'm talking about amigas. In post #111, you were talking about Jay Miner. In post #151, you stated you had nothing else to state. I wonder what I'm replying to.

 

dont talk about amigas. talk about why is the less bandwidth antic superior compared to VICII ? its endlessly ammusing as when atarians are cornered they start to blabble about amiga.

[Fröhn]

I have been playing around with 160*200*8, 160*200*10, and 160*200*11 modes (w/o interlace) on Atari and I don't have any problem setting up my palette using the OR player mode. For example, set register 704 to 2, 705 to 4 and the OR will be 6 and you have 3 shades of gray using 2 indices. Or try 704 to 8, 705 to 4, and OR will give you 12, etc. etc.

It pretty much ends up that you will ALWAYS do 2;4;6 or 4;8;12. If that doesn't fit the graphics, the 3rd color is not used.

 

C64 would have been better off doing the OR or some logical operation rather than having two fixed color registers. So on Atari if I have two multicolored sprites, I can get 6 unique colors. On C64, if I use two multicolored sprites I can get 4 unique colors.

Two multicolored players are made from four single color players (for multicolor, you have to overlay one player with another). With 4 sprites on C64 you get 6 completely independent colors, and two of those sprites are even hires. And you still have 4 full sprites left.