Amiga.org
Amiga computer related discussion => General chat about Amiga topics => Topic started by: ElPolloDiabl on March 02, 2010, 05:09:17 AM
-
If you were clever enough to spot the trends how would you have designed the AGA Amigas? Turning too little, too late into just enough and just in time.
-
I would have ditched the A1200 and come out with 2 A4000's a low spec version and a high spec version. The low spec version could be 020 and 32 bit zorro. Then have a high spec version with and 030 or greater.
Get rid of the A600, but keep the CD32 as the 'wedge' system with optional keyboard & mouse.
CD-Roms were a great cost saving after the initial investment. Staying with big box systems would make it easier to add RTG, 16 bit sound and a CD-rom. It might have made the difference.
-
The 4000 would not have been AGA (RTG only).
The main problem with the 1200 was that it was released too late. If it had come out in say 1990 as the A500+ with a fast 68000 and AGA it would have sold like hot cakes.
-
I'd have pushed hard drives, as well as some fast RAM, but the trick is to still hit a spot where the amiga would be competitive price-wise.
I don't think there's any way around the fact that the 68020 just was not cutting it though
-
Improve the old 7MHz bus/chipset speed. 12.5MHz would be nice and would work with the same 80ns DRAM. There wouldn't be time for two memory reads within two clocks, behind the CPU's back, as there is on AGA, but a longer burst read could be used and bandwidth would be almost the same without it anyway. Other benefits: blitter runs faster, 12.5MHz/25MHz gives a proper 320/640 wide resolution on a VGA monitor.
Scanline buffering: so the bus speed doesn't need to divide evenly into the pixel clock, and enabling scan doubling with full backwards compatibility
Chunky 8-bit mode. Possibly with a mixed-mode where if the high bit is set then the low 7 bits are combined with the next byte to make a low-res 15-bit RGB pixel. So 128-color high-res could be mixed with 32K-color low-res.
2-4 more sound channels with high sampling rates (video scanrate independent) and the possibility to combine two 8-bit channels into one with 16-bit resolution.
Other Paula improvements to enable high serial port speeds without dropping bits and high density floppies at the normal rotational speed.
Make it so the blitter can handle 32-bit word size
-
I think on the hardware side we would need Ethernet and 3d graphics. This would have taken Commodore to the mid 90. A Quake like network game would have driven the sales.
On the software side something like today's Amicygnix would have kept many folks from leaving for Linux.
To gain developers in 98 Commodore should have had it's own Java Implementation. None of these had to be for free. Just make sure you don't mess with DRM and sueing your customer.
Finally in 2001 you had to be there with something portable and something that can burn DVDs from DV-Camcorder stuff.
Oh, and somebody would have had to create a good Microsoft Word compatible converter or wordprocessor.
-
The A1200 would have had at least a 68030 with 28Mhz or 40Mhz . Ram would be upgradable via standard PC ram allowing at least up to 32Mb.
The A1200 would also have as standard a low spec Picasso96 compatible 1Mb RTG graphics hardware built in with a slot to optionally upgrade it. A built in Scandoubler/Flickerfixer would have been a great addition too, but if that would have been too expensive then a slot/connector for an optional one would have been good :)
Another nice touch would have been an external 40pin IDE connector as standard for the cheap and easy installation of IDE CD-Rom.
-
optional 2mb fast ram socketed on board, allow added ram to be configured as chip ram, 68030@28, fpu socket on board, RTC, proper chunky modes, proper 24-bit output with flicker-fixer/scan-doubler (even if only as an optional add-on to reduce the cost of the basic model), faster/more capable AGA chipset, 16-bit Paula, PIO-5 capable IDE controller, high density floppy, 32bit pcmcia slot, vga out or -at least- an adapter for RGB->VGA, clock port at least twice as fast to be a proper expansion option, workbench with virtual mem and memory protection (since 68030 has a full MMU) even if it killed much compatibility, 12 Fkeys, pgup-pgdown keys, full parallel port.
-
I'd have got rid of the PCMCIA port. It was an expensive (though almost out of date) laptop part. Would have spent the money on A superHD floppy drive, fast RAM and a faster processor. Rejigged the case to allow a 3.5" HDD upgrade, and maybe room for a slimline CD upgrade.
Also, i'd have had the machine boot completely from ROM, with a small amount of EEPROM (upgradable) For wall papers/files/utilities/system upgrades. Maybe 512K. It would then perform a lot like a HDD equipped machine, but at a fraction of the cost. None of that constant swapping disks.
Updated Paula to 16 bit
Maybe add Midi, the ST was on it's last legs, Commodore could have canibalised their market.
Maybe even buy the ROM off Atari and sell Amiga A1200 with ST compatability ;)
Added chunky modes, maybe 24bit modes as well.
Some sort of anti flicker hardware to allow interlace to display better, and be suitable for games.
Maybe added a DSP to help decode audio/video and perhaps help with 3D
Upgraded the blitter to blit 3D objects.
Socketed processor and ram, allowing the trapdoor to be used for other upgrades.
HAM or 24bit chunky workbench. That would have wowed people.
Created a mid range desktop/tower same as above, but with zorro's, but no ISA.
...sold off the pc division early,
...bought some top games companies with the proceeds.
-
In all honesty, the Atari Falcon was basically what the Amiga A1200 should have been. Had the A1200 been build to those specs, it would have been the bare minimum to survive... Commodore would have clung on much longer...
-
Easy answer Natami but it looking like its taking as long as commodore to come out with a public product. But still wishing
-
In the case of the A1200
030 CPU
socket for 68882
DSP SLOT or SOCKET for DSP sound chip (I guess this is what the clock port became)
CPU SLOT
AGA Amber chip and VGA cgraphics chip with CYBERGFX or RTG support instead of the software screen promotion
Simm Socket on motherboard
recofigurable CHIP/fast ram sizes
Midi built in
Desktop IDE port
VIDEO IN And GENLOCK PORT
Pizzabox case with PS2 mouse and Keyboard connectors
With the glut of CD32's showing up on EBAY it'd be interesting to make a CD32 expansion device that would extend the functionality and allow it to fit in an ATX style case.
After the demise of commodore I remember seeing ads for machines called the A2200 that were supposed to be modified CD32's with somekind of expansion board if I remember corectly.
-
I'd have got rid of the PCMCIA port. It was an expensive (though almost out of date) laptop part.
PCMCIA almost out of date in 1992? That's only slightly ridiculous, considering you can still find new PCMCIA cards today (sure, most are CardBus by this point.) I bet 600/1200 users everywhere appreciate the low cost to get their machines on the network and ease of data transfer via PCMCIA/CF adapters.
-
I'd take things back a year:
Still make the A500+, but with an EC020 and 2MB Chip onboard. Maybe add an internal IDE connector too, use this as a sensible intermediate system to run down stocks of remaining ECS chips and A500 cases. Remain compatible with existing edge-connector devices.
Scrap the A600 altogether.
A1200, but this time with a socketed '030 (Offer a choice of CPU speeds perhaps) and with 2MB Chip and 2/4/8MB Fast option. PCMCIA is probably a better option than a proprietary connector on the side.
-
More like how i'd have done commodore:
Stop developing 8-bit computers after the c64, but give it a ram slot and sell models with higher ram. Make a constant effort to miniturize the design.
Released the 500 and 2000 in 1985.
Make a C64 compatability card for amiga.
Clone the 68K archetecture. MOS tech is right there, so use it guys. motorola aren't doing anything special.
Put a ton of funding into developing a new chipset for release in 1990. chipset should be capble of chunky, and have atleast a simple 3D accelerator that could do texture mapping. Think "almost playstation", like ocs was "almost megadrive". Monitor drivers that include 800x600 and 1024x768 are needed too.
Develop the hell out of the OS.
skip 500+, 600, cdtv, and those stupid commodore PCs. Release the 1200 and 4000 in 1990 CPU should be socketed and a range of speeds avalible. use HD floppys in the 1200.
behold, a non bankrupt commodore.
-
Nice answers, seem like logical steps.
I think Commodore was overly concerned with price. They definitely skimped. That's in complete opposition to how consoles are made nowadays where features are the prime focus.
Nobody mentioned PC compatibility except for HD FDD. When Shapeshifter came out we would have had all bases covered. Unfortunately no new hardware except for expansions by that time.
-
i agree with others that aga should have had chunky modes. i think i'd go further and say it should have been on a card. the 4000 should not have had graphics on the mobo. the 1200 yes but not the high end machine. other computer companies were doing that so the trend was there.
an aga card with chunky modes that could disable the ecs on a 3000, provide all graphics for a 4000 and dare i say it even been released for pci, could have given the system more flexibility had commodore lived to release a 5000.
of course 3rd party card makers would have suffered unless they integrated an aga chip into their design or emulated it.
also they could have upgraded paula somehow. as amazing as amiga audio is by '92 pcs had mostly caught up.
-
In all honesty, the Atari Falcon was basically what the Amiga A1200 should have been. Had the A1200 been build to those specs, it would have been the bare minimum to survive... Commodore would have clung on much longer...
A agree big time. A1200 should have had an 030 at least. We're talking about a machine that practically needed to be upgraded right out of the box. Too weak too late. 31khz output should have been standard by then (ever since the A3000 actually) as well as the floppy drives being HD.
PCMCIA was a bad idea too. WTF... 16-bit bus on a 32-bit system? 4mb limit and the cardbus reset problems never should have happened either. Ditch PCMCIA and put an external IDE or SCSI connector on the side instead of the PCMCIA port.
I would have redesigned the A4000's case and made it larger. Scratch that - desktop cases were on the outs by then. Perfect solution was to only have a towerized model from the start and include all the things that should have gone in with it (it's a frickin' multi-media computer after all, right?) like a CD-ROM drive and include the software to make it CD32 compatible. Things such as CD-ROM drives should never had been an "option". The modular design of its CPU AND Zorro Daughtercards were expensive ideas too (DUH!). IF they would have designed the A4000's mobo to be singular, with an 040 chip surface mounted and the Zorro slots inline like on an A2000 - the machine's cost could have been kept to a minimum. Still have the CPU card bus of course for those that want to expand later - but the 3640 was a terrible product and idea right off the bat (backwards caps? Idiots!).
Someone else mentioned a while back that Commodore should have paid more attention to networking and I agree. While solutions may have been around, networking was kind of hush-hush in the Amiga community. I never saw a lot of software or cards for that matter touting that very basic of computing features.
Paula should have been upgraded to 16-bit by the A3000 on and whatever chip to replace her could have been made to be backwards compatible. Would have been a triple treat if the grungy 8-bit sounds were all "promoted" to smooth 16-bit. Talk about reviving old software!
CD32 is pretty well designed (besides Paula) as-is IMO, but should have been directly plug-in compatible with the A1200.
A600 should have never happened, but since it did - should have come with at least a 14mhz 68000 CPU (AdSpeed type product, or an 020), 2MB RAM, HD floppy drive and a TON of games packed in. Again, ditch PCMCIA and slap a CD32/SCSI/IDE friendly connector to the side of her. Keep the 15khz output and tout that as your low-end legacy A500/C64 type of gaming computer.
I believe if all of these concepts were implemented, would have allowed Commodore to better position themselves as a serious computing contender in the long run.
-
Clone the 68K archetecture. MOS tech is right there, so use it guys. motorola aren't doing anything special.
...
behold, a non bankrupt commodore.
Actually, that one step would've been a bad idea and might have ruined them, as MOS learned earlier after being sued by Motorola for making a CPU pin-compatible (not even a clone!) with the 6800.
-
There was already a precident for cloning cpus by then, AMD 8088/86 clones were already in the IBM PC, hell IBM used whichever was available for the next shipment. Not to mention, they wouldn't be selling it to outside companies, bar ones making licenced amiga clones (something that should have happened, provided they were identical to commodore machines), and would probably take development down a different path entirely.
-
@hell labs
actually the 8088 and 8086 weren't cloned. amd was a second-source for intel. ibm agreed to use the 8088 only on the condition that intel get a second source. it wasnt till the 486 that amd truly cloned the 486 since the lawsuit over intel 386 microcode wasnt settled.
motorola had second-sources but only true clones of the 68000. noone cloned the 020 and up cpus.
the problem with mostec using pin compatibility with the 6800 was that they claimed it and put it right in the docs and adverts for their chips.
-
Ah I see.
But 8080/ Z80, right? That counts. Infact i'd say the Z80 was even more blatant because they gave away the design for free. Probably the worlds first recorded instance of Pirate copied hardware outside the soviet union.
-
i think the best thing atari and commodore should have done was work together. commodore should have made the high-end machines and atari the low-end and game machines. there was way too much animosity between them for that though.
-
Ah I see.
But 8080/ Z80, right? That counts. Infact i'd say the Z80 was even more blatant because they gave away the design for free. Probably the worlds fist recorded instance of Pirate copied hardware outside the soviet union.
yeah 8080 and z80 counts. that a clone and upgrade in one. the z80 was way better, and not really a clone
the z80 is just a better chip single power voltage more and better instructions, double the registers for quicker interrupt response. kindof like if amd came out with a core 5 cpu today
-
i think the best thing atari and commodore should have done was work together. commodore should have made the high-end machines and atari the low-end and game machines. there was way too much animosity between them for that though.
No kidding! To this day, I'm still stunned Tramiel left and sided with Atari!? What a goofy decision. I really don't understand it at all. Brought C= up in the 60's with typewriters and calculators, produced so many great computers and then by the time the Amiga comes around, he jumps ship and swims over to the enemy? Guess I need to read that new-ish Commodore book here soon. lol
-
yeah 8080 and z80 counts. that a clone and upgrade in one. the z80 was way better, and not really a clone
the z80 is just a better chip single power voltage more and better instructions, double the registers for quicker interrupt response. kindof like if amd came out with a core 5 cpu today
yeah, but it was still compatable. I think commodore cranking out the 68k equivalent to the z80 would make sense.
-
yeah, but it was still compatable. I think commodore cranking out the 68k equivalent to the z80 would make sense.
by the time aga came around mostecs chip fab was still in the 80's they had to outsource aga (denise?) because they couldnt make it. they could produce 16bit cpus if they tried as well as ocs, ecs, dumb frame buffers, and memory controllers but anything else was beyond their capabilities.
the vertical-integration of chip-fab and computer manufacturer was good in the 80's but commodore held mostec back. by 85 they should have had a 16-bit 6502 by 90 a 32bit etc. instead for years they spit out varients of the 6501, 6502, 6510, 8510 etc. etc.
-
actually by 85 they should have been able to make the 68000 themselves
oh wait maybe that is what you mean. ok i agree
-
actually by 85 they should have been able to make the 68000 themselves
oh wait maybe that is what you mean. ok i agree
Yeah, mos tech could have been commodores greatest strength if they'd have looked after it properly.
so outdated factories they had to outsource the chipset? goddamn commodore , every time I learn more about how they operated, they seem even more STUPID.
-
yeah they were loving up on the cash cow c64 and letting the techs do something or other with the amiga. no focus no plan for the future.
-
I would have designed AGA just as it was........
....... and release the A3000+ in summer 91 ........
-
yeah they were loving up on the cash cow c64 and letting the techs do something or other with the amiga. No focus no plan for the future.
to the time machine!
-
Going back a little earlier...
The A500+ I think was a good machine for the price. I would have left it as it is.
The A600 had cuteness going for it. It should have had a 14MHz 68000 however, just for a little more pep, especially for 3D games.
The A1200, much as I loved it, was made to a price point and it showed.
Firstly, I would have had a 1.76MB floppy disc drive. The 880KB drives were horribly limited by this time.
The PCMCIA slot should have been an A500 style sidecar slot with the full 32-bit bus available. Drop the internal trapdoor expansion apart from making a SIMM slot (or two - 1 chip, one fast) available for memory expansion.
AGA was a bodge. The blitter should have been a full 32-bit blitter for a start. I would have had more audio channels in Paula, and the capability to run the HD floppy at full speed.
A native, fast, 8-bit chunky mode was needed. Also sprites should have been arbitrarily sized (even in steps of 16 pixels) and (effectively) unlimited in number, 4, 16 or 256 colours without having to "layer" the sprites for more colours. A scanline buffer that someone mentioned would have been good. Ability to do 800x600 (and 1024x768) resolution non-interlaced, albeit in fewer colours.
3D capability isn't a consideration for me. That would have been in a follow up to AGA. Same with 16-bit audio (although hardware support for that clever 14-bit channel mixing that required loads of calibration would have been nice). Same with FPU sockets - leave them for the expansion boards.
Maybe I would have made the base A1200 1MB chip and 1MB fast. Especially so for the CD32.
The 68EC020 in the A1200 was a cost reduction thing. Shame that Commodore couldn't negotiate Motorola into using a faster chip - e.g., a 20MHz+ 68020/68030.
-
Wasn't Paula the one that C= lost the original designs for, which is why it was never upgraded throughout the entire lifespan of the Amiga?
-
In all honesty, the Atari Falcon was basically what the Amiga A1200 should have been. Had the A1200 been build to those specs, it would have been the bare minimum to survive... Commodore would have clung on much longer...
No. Medhi Ali and Co. were out to sack and pillage C=. A functional C= was worth less to them than it carved into pieces and sold off.
C= died exactly when they wanted it to.
-
I would hvae retained the A4000T, scrapped the desktop version. What I would have included as standard in the tower would have been a CD-Rom drive (not having one when almost all PC's did or could be easily upgraded to have one was a big reason for many to switch to the PC IMO). I would have added network capability, even if it was only dial up. Not being bale to connect my A4000D to the web easily in 1997 pushed me to build my own towerised PC with CD and US Robotics 33.k modem. I only got the A4000D hooked up to the web in 2007 when I got an x-surf card and broadband.
Jutst my two pennies.
Weed.
-
As someone else said; a low-spec and a high-spec 4000 box. No "wedge" 1200, ever.
I'd have done the low-spec 4000 with on-board Zorro slots (no riser) and a couple of ISA slots (as an a2000), with a fixed CPU. I'd have pushed for the 8mb Chip RAM but fixed it at 2 on the low spec machine. 28mhz 030, 2mb RAM (in a pair of 1mb SIMMs) as standard on the motherboard. Maybe a CPU slot (as per the A3000 and A4000 as it was) - dunno.
The high-spec machine would've been much like it was, excepting:
A 2nd video slot, occupied by the AGA display chipset. Future video cards could have gone there, thus "upgrading" at least the video chipset. As soon as the PCI spec hit, (1993, just a year afterward) I'd have opened the riser/daughter board specs and allowed (if not created outright) a PCI riser card and moved away from Zorro. Rather like the DCE G-Rex did, I'd have had a cable for "local bus" attachment to the (PowerPC) CPU card to ensure maximum speed.
Assuming this fictional C= survived, and the 4000 hadn't been outright replaced, I'd later have released a PPC card around '95 or so (whenever Apple made the transition) and ported the OS whole-hog to that.
That'd be the last AGA machine update. After that, new systems altogether - probably farm out chip design to 3dfx (remember, we're now into the early late 1990s - '96 or thereabouts) for video, embrace GLide or OpenGL, etc. etc. I'd also be investigating migration to Intel architecture, but now we're wandering far afield...
-
Wasn't Paula the one that C= lost the original designs for, which is why it was never upgraded throughout the entire lifespan of the Amiga?
I thought that was the entire c64. Probably most stuff they made as well.
-
PCMCIA almost out of date in 1992? That's only slightly ridiculous, considering you can still find new PCMCIA cards today (sure, most are CardBus by this point.) I bet 600/1200 users everywhere appreciate the low cost to get their machines on the network and ease of data transfer via PCMCIA/CF adapters.
How long did you have to wait for drivers for those cards? 10 years?
CF cards? only slightly rediculous, as we are talking about 1992...
As for betting that 600/1200 users appreciate it....they do now....but 99%
of users never got to take advantage of that expensive 16bit bit of crap hanging off the side,
they moved to the PC that had a CD rom slot, and the ability to add one cheaply.
-
So then it wasn't out of date, it was ahead of its time!
-
So then it wasn't out of date, it was ahead of its time!
lol. Like so much about the Amiga.
The only out of date thing a bout the Amiga eas Commodore itself ;)
-
In no particular order:
* Enable Chunky 8bit mode
Not difficult to do. THis would have helped games like DOOM a lot.
Low hanging fruit.
* Enhance Paula to support 16bit samples.
This would have been actually a very simple task if you still have the orignal design.
Paula fetches smaples anyhow in 16bit chunks so 16bit support would have been simple.
Low hanging fruit.
Adding more channels would have been more work.
* Add 16bit hicolor direct mode.
Enough Bandwidth was available for this mode.
Low hanging fruit.
* Blitter
Improving Blitter is very tricky.
Increasing clockrate would have been easy at the time but would bring no benefit without a clever buffering. Going for 32bit like some proposed here is quite a challange so it was not worth it at this time. Not a low hangiung fruit.
Nevertheless is was a bit dissapointing that the blitting power did not increase.
* Added fastmem 2MB to the A1200 would had made a big performance difference.
Yes, 2MB fast mem would have made the system a little more expensive but the performance change would been huge.
* The 3640 was to say it frankly "sad".
The 68040 bus interface is actually a lot easier than the 68030 interface. Having to convert to better 040 interface down to the 030 interface of the motherboard was a pain. The 68040 lost a significant part because of this. Would have been cool of the would have designed the A4000 for the new interface of the 040 and 060 CPUs...
* A line buffer to get double scan for free would have been cool.
But at that time this was not for free. Today we can do this for litttle and are doing it on the Natami for example - but at that time it would have increased cost.
I think it was a reasonable decision to build the system without it.
* Nevertheless Double modes (DoublePAL etc) could have been implemented slightly differently. The way they were implemented required a changed copperlist.
This could have been done without also - this way the AGA chipset could had been able to promote any old game the VGA. Its somewhat of a pity that they did not do it this way as this would have made running old games on VGA monitors so much easier.
All in all the AGA chipset was good.
The most missed features were certainly 8bit chunky.
16bit chunky would have been very nice and would have been simple to do also.
The original Blitter would have been perfect to support this mode.
Quite many people in the scene were surprised that AGA came without chunky support.
Cheers
Gunnar
http://www.natami.net
-
How long did you have to wait for drivers for those cards? 10 years?
I bought PCMCIA ethernet card for A600/A1200 (http://www.amiga-resistance.info/bboahfaq/index.php?action=artikel&cat=67&id=2949&artlang=de), in early 1995 I think it was.
-
I bought PCMCIA ethernet card for A600/A1200 (http://www.amiga-resistance.info/bboahfaq/index.php?action=artikel&cat=67&id=2949&artlang=de), in early 1995 I think it was.
Why does it have a temperature dial from a radiator on it?
-
Why does it have a temperature dial from a radiator on it?
Thats where you put in the ethernet fluid...
Actually, I think that is a 50 ohm terminator. You remove that, and your whole network goes down.
-
That explains so much. Reminds me of changing the automatic transmission fluid in a chrysler, shifting demons ran off down the street and stole a cornetto from the shop across the way.
-
Thats where you put in the ethernet fluid...
One time I accidentally spilled Ethernet fluid into my keyboard. Then all my keystrokes started to randomly appear on the screens of other nearby computers. Freaky.
:laughing:
-
If you were clever enough to spot the trends how would you have designed the AGA Amigas? Turning too little, too late into just enough and just in time.
Simple: I would have released the A3000+ as engineering intended. Late 1991 launch would have given VGA a run for its money, and the onboard DSP would have done wonders in the professional market.
I think there were some deisgn notes to use battmen.resource to auto-force the system to NTSC/PAL as selected in the Early Boot Menu, so if that had come to fruition (and DblNTSC/DblPAL/VGAOnly were selectable options) the scandoubler issues we face now would have been pre-empted.
With that machine on the high end, a 1200 with DSP, hard drive, clock, FPU and some Fast RAM as standard would have been nice. Price would be kept low with subsidies from the phenomenal sales of the 3000+ (since we're dreaming anyway :))
-
I would only change some minor things,
No1 have a 1.78mb High Dens Disk supported and standard in the A1200 & A4000
No2 have a FPU chip at least 68882 - 20Mhz on all AGA range
No3 have a A4000CD edition launch mid 1993
No4 have Team17 or another developer make a game to show off the AGA, FPU and general power of the Amiga.
No5 port Doom to the Amiga, even if it only could run on the A4000 I think we needed it back then to prove the Amiga could do a game like it as many on the PC didn't believe it could be done, I know Alien Breed and Gloom kind of proved them wrong, but the Name alone would of added a lot to the Amiga image.