Was anyone else disappointed with the A1200?

[Bruce Abbott]

Originally Posted by hammer

Big box Amigas has a very tiny install base.

And?

PC cloners focused on CD32 cost reduction style with expansion slots.

Nonsense. The CD32 integrated everything on the motherboard, with a single edge connector for future expansion. That's quite different to the average PC clone of the time.

PC cloners focused on making the standard PC architecture cheaper. This was mostly done by using far East manufacturing plants, and later on integrated chipsets incorporating the functions of standard ICs. Slots were actually more expensive than putting it all on the motherboard, but more flexible and allowed greater economy of scale because manufacturers could concentrate on just making one component (motherboard, video card etc.). But the biggest advantage PC cloners had was simply the insane demand for PCs.

A 64 bit wide sprite still has four colors. Sprites to be ganged to increase colors.

Yep, and with ganging you got 16 colors (15 plus transparent) each with its own color palette (unlike OCS which shared sprite colors with the bitmap). That extra width made a huge difference, allowing the use of sprites for large objects when previously you had to blit them. It was also commonly used for parallax backgrounds. A single sprite only has to be repeated 5 times to cover a 320 pixel wide screen.

HAM8 wasn't good for displaying hi-color with the WIMP GUI system.

HAM8 has 64 base colors, plenty enough for a 'WIMP GUI'. That could be combined with true color images if desired. Workbench doesn't do that, but the Amiga isn't limited to a single surface for everything so it's not needed.

Between packed pixels vs 24-bit color palette, I rather have packed pixels.

I'd rather have 24 bit for video applications, thanks. 256 colors is too limiting. But when I aren't doing video stuff I am quite happy with 8 colors on the Workbench and custom screens for displaying photos and playing games etc., with each application getting the resolution it wants rather than being forced to use a global desktop. This is much tidier and more convenient than every app trying to share the desktop. On the PC I am constantly shuffling windows around, which is a pain.

Packed pixels or bitplanes makes no difference to me, the results are the same - except that with bitplanes I can choose fewer colors to save memory and speed up operation.

One of the really annoying things about Windows in the old days was that 256 colors on a typical ISA bus VGA card was painfully slow, but you needed it to run any app that wanted more than 16 colors. And switching resolutions in Windows wasn't easy. You could easily get stuck with a bad video mode that crashed Windows and dumped you back into DOS. Then you scrambled for the system disks and fed them in one after another trying to repair the damage. Pathetic!

Even on more modern systems it can be a problem. In Windows 95 (and others I presume), when a game wants a different screen mode the system switches the desktop to it, with hilarious results if a game running in 320x200 crashes back to desktop. In Tomb Raider one of the key combinations is CTRL-ESC, which also happens to be the keys for the start menu in Windows XP. If you hit those keys Windows switches back to the desktop and the game loses focus. You can try clicking on the minimized window on the task bar but the game won't come back!

[hammer]

Originally Posted by oscar_ates

I found an interesting article :

https://www.appuntidigitali.it/20663...-no-new-chips/

Against https://www.appuntidigitali.it/20663...-no-new-chips/

Yes, the possibility of having 1 or 2MB of Chip memory is certainly handy for being able to use more assets in games or for graphics or audio programs. Of much rarer use are, on the other hand, the sprites that can be displayed at the edges, or the possibility of managing areas larger than 1024×1024 with the Blitter.


1989 released A500 Rev 6A has the following:

1. 2 MB Chip RAM is a reserve capability on this motherboard variant with a valid jumper setting for 2 MB Chip RAM configuration.

This is a business decision to assign the "2 MB Chip RAM" feature for the A3000 i.e. product segmentation reasons.

2. With A500 Rev 6A's 2 MB Chip RAM jumper configuration and four higher capacity DRAM chips, A3000's ECS Agnus AB and B models are fully supported on this motherboard variant.

3. A500 Rev 6A has all ECS Agnus improvements of A3000's ECS Agnus except for the 2 MB Chip RAM memory address.

A500 Rev 6A's R&D would have occurred before its 1989 release.

My point, R&D is already committed to ECS 2 MB Chip RAM design for the 1989 A500 Rev 6A revision, not just for the A3000.

Dave Haynie is correct on "read my lips, no new chips" directive since A500 Rev 6A's 2 MB Chip RAM ECS design is tied with A3000's 2 MB Chip RAM ECS.

[hammer]

Originally Posted by Bruce Abbott

And?

Tiny user base wouldn't make a difference.

Originally Posted by Bruce Abbott

Nonsense. The CD32 integrated everything on the motherboard, with a single edge connector for future expansion. That's quite different to the average PC clone of the time.

Nonsense.

The 1993 PC clone has a highly integrated Northbridge and super IO Southbridge e.g. https://eessential.blogspot.com/2014...therboard.html
This OPTi chipset-based 486 motherboard has a familiar Northbridge and Southbridge design.
OPTi 82C895 is a tightly integrated chipset. This motherboard is from the 1993 era.

https://dosdays.co.uk/topics/motherboards.php
https://dosdays.co.uk/topics/images/...cket3_mobo.png
A Taiwanese PC Chips M919 Socket 3 motherboard from 1993 also has a familiar Northbridge and Southbridge design from UMC.

https://www.dosdays.co.uk/topics/Man.../chaintech.php
Yet another Taiwanese Chaintech's 486SCSL(1993) with a single SIS chip for 486, VLB, and ISA!

https://www.dosdays.co.uk/topics/Man...s/gigabyte.php
Yet another Taiwanese, Gigabyte's GA-386PS (1989) with two chip OPTi + IMP82C206P for 386DX.

https://www.dosdays.co.uk/topics/Manufacturers/msi.php
Yet another Taiwanese, MSI's MS-4125 (1993) with Contaq 82C596 / C&T 82C206 chipset. MSI's MS-4126(1993) single-chip SiS 85C471. MSI MS-4134(1993)'s two-chip ALI M1429 / M1435.

https://www.dosdays.co.uk/topics/Man...rs/biostar.php
Biostar's MB-1420 (1992) with two chips BIOTEQ and CT.

https://www.dosdays.co.uk/topics/Manufacturers/abit.php
Yet another Taiwanese Abit's AB-AG4 (1993) with two chip SiS 85C471. AB-AN4 (1993) with two chip SiS 461/407.

https://www.dosdays.co.uk/topics/Manufacturers/fic.php
FIC's VIA4386-VIO (1993) has two chips VIA VT82C481/495. 486-GAC-V (1994) has a two chip VIA VT82C486A chipset and embedded Cirrus Logic CL-GD5426 chipset.

A3000's CPU side has Ramsey (32-bit memory controller), Super Buster (expansion bus controller), Super DMAC (DMA controller and interface for SCSI controller), and Fat Gary (system address decoding and supporting functions for the floppy disk drive).

A500's Gary was one of the cost-reduction chip that integrated A1000's discrete chips.

A1200's Budgie chip is the cost-reduced Buster and Ramsey functions.

CD32's Akiko integrated Budgie, Gayle, and two CIAs as a cost-reduction method!

Baby-AT form factor motherboard is smaller than the A3000/A4000 motherboard + daughter bus board. A3000/A4000's design is expensive when compared to PC Chips M919.

[hammer]

Originally Posted by Bruce Abbott

Yep, and with ganging you got 16 colors (15 plus transparent) each with its own color palette (unlike OCS which shared sprite colors with the bitmap). That extra width made a huge difference, allowing the use of sprites for large objects when previously you had to blit them. It was also commonly used for parallax backgrounds. A single sprite only has to be repeated 5 times to cover a 320 pixel wide screen.

Two sprites needs to be ganged for 16 colors (1 being transparent).

A single sprite has four colors.

For Mortal Kombat AGA's background layer, at least 16 colors is needed.
Single playfield could be used for 7-bit (128 colors) planes which recycles SNES's art assets. It could be down to 6-bit (64 colors) planes if additional slots are needed.

8-bit planes with Blitter/Sprites compositing can be a problem for stock A1200 e.g. Turrcian 2 AGA. With Fast RAM, there is additional headroom.

With Doom-type games, the CPU has the 2.5/3D compositing job.

Originally Posted by Bruce Abbott

HAM8 has 64 base colors, plenty enough for a 'WIMP GUI'. That could be combined with true color images if desired. Workbench doesn't do that, but the Amiga isn't limited to a single surface for everything so it's not needed.

HAM8 also has 8-bit planes and is slow at Double NTSC 640x400p. It's slow for Shapeshifter. There are HAM8 artifacts.

Originally Posted by Bruce Abbott

I'd rather have 24 bit for video applications, thanks. 256 colors is too limiting.

HAM8 only displays up to 18 bits color which could be served by 18 bit palette. There are HAM8 artifacts.

Originally Posted by Bruce Abbott

But when I aren't doing video stuff I am quite happy with 8 colors on the Workbench and custom screens for displaying photos and playing games etc.,

For my A3000, MagicWB's 8 color them is used. On my A1200, 64 colors on Double NTSC 640x400p.

Originally Posted by Bruce Abbott

with each application getting the resolution it wants rather than being forced to use a global desktop. This is much tidier and more convenient than every app trying to share the desktop. On the PC I am constantly shuffling windows around, which is a pain.

Separate desktop screens have it's own downside since there's less visibility on active applications. Shuffle a deck of cards effect.

I use a shared desktop on A12000-PiStorm32-Emu68's P96 RTG e.g. web browser, mp3 player (or Deluxe Music 2), and a text editor (or ShapeShifter).

If I have stayed with my Amiga 3000 in 1996, it would be CyberGraphics 64 RTG. EGS is interesting and problem is just the price.

Separate desktop screens are workaround for hardware limitations.

------------
I have dual 4K monitor setup with OBS/social media on one monitor and a game on another monitor.

For multi-desktops on current Windows 11 or 10,
https://support.microsoft.com/en-au/...ion=Windows_11

I don't use it.

Originally Posted by Bruce Abbott

Packed pixels or bitplanes makes no difference to me, the results are the same - except that with bitplanes I can choose fewer colors to save memory and speed up operation.

Turrcian 2 AGA has the full 256-color VGA art asset.

[pandy71]

Originally Posted by Bruce Abbott

He says adding stereo panning to Paula would be 'trivial'. But as you can see from the die photo, there is no spare silicon to add this feature to the existing design. A completely new chip would have the designed, and as Paula is a mixed signal (digital and analog) IC this would not be easy.

In fact it is quite easy and it may be implemented on at least few ways - simplest is just negate (make it complementary) PWM AUDxVOL signal and use 8 bit registers in front of DAC to hold sample value and switch between them - silicone area can be reduced as instead 4 DAC's you could just made 2 DAC's and introduce time interleaving (this would made also possible to demux channels and allow in theory possibility to have 4 independent channels). Silicone area could be reduced after such changes - also if PWM replaced by binary rate multiplier like SN7497 overall audio quality can be improved (in theory no hard limit of 56kHz) and made all above way easier. This was possible at worst in ECS Paula - shame on Commodore that they not implemented new Paula features in AGA.

Originally Posted by Bruce Abbott

The suggestions for making ChipRAM access faster don't apply to the 68000, and would be useless on faster CPUs without greatly increasing ChipRAM bandwidth. This would be a lot more than just a little tweak. The caching ideas would also involve a lot of extra circuitry and tricky logic, with significant compatibility issues.

Not necessarily - adding address comparator should not be so difficult - developers will quickly adopt to align code within limits of DRAM page.
No cache needed just WAIT (DTACK) generator.
Small cache like 256 bytes or 512 bytes was OK - perfectly feasible even for CSG - single IC with all logic and SRAM inside running at 14..28MHz - overall complexity comparable to Amber/A2024 IC's

Originally Posted by Bruce Abbott

This article derides the "Read my lips – no new chips" command, but if it was intended to avoid stuff like this then it was justified. Trying to jam all these features into ECS could have set back its release by years. Commodore had to get the A3000 out while it still had a chance to be relevant - thus 'no new chips' - but in fact it had heaps of new chips in it already.

Nope - ECS didn't introduced anything - CBM should made AGA instead ECS - ECS could be reasonable if it was integration of OCS - from 3 IC's to 1 IC - so standard cost reduction approach.
A3000 with AGA half year later could be significant design in Amiga portfolio.

[Thorham]

Originally Posted by Bruce Abbott

On the PC I am constantly shuffling windows around, which is a pain.

Indeed. Even two monitors doesn't help much.

Originally Posted by hammer

Separate desktop screens are workaround for hardware limitations.

They're functionally superior, hence why Windows 11 offers multiple virtual desktops.

[pandy71]

Originally Posted by hammer

This topic's purpose.

Partially - in fact each discussion can be started and stopped by single expression "read my lips no new chips"

Originally Posted by hammer

You started it. This is about A1200 and Commodore.

nope - i didn't started anything - just few times decided to express my opinion (having back in my head that time can't be turned back) and sometimes correct misinformation (sadly you are source of this).

Originally Posted by hammer

You assumed I was not aware of the "spaghetti code" in corporate environments.

I didn't assumed anything - my point was that even reasonable management (i mean people are skilled and they poses technical knowledge) quite frequently taking bad decision leading later company to problems...

Originally Posted by hammer

I have many stories about Toyota's cost-cutting e.g. primary metal bummer's width vs Lexus counterpart. Toyota's offset crash test suffered.. badly. Toyota responded with improved offset crash test results in 2014-2015 models.

As i pointed earlier many (if not all) companies doing things like this.

Originally Posted by hammer

Did you assume I'm not aware of this?

I provided real life example - you bombarding topics with images of random things to prove your point, i've wrote based on my personal experience not someone experience.

Originally Posted by hammer

Up to a certain point. During H2 1993, Apple had the superior integer CPU under $1000 USD.

Nope, Apple was never HW company, they focused (probably right decision) mostly on software side - as such absent of HW graphic acceleration for very long time (i think even PC market was way more advanced)

Originally Posted by hammer

I'm aware of Executor. My old retro PC has a copy of Executor.

Running MacOS on non-Apple platforms will only benefit the MacOS market ecosystem.

Perhaps but if CBM put some effort (like they made with PC and famous Transformer) on recreation of Mac OS API and as such extend Amiga SW portfolio also for Mac software then it could be right decision.
Having ROM less clean room Mac OS API implementation ready for Amiga would be smart decision.

Originally Posted by hammer

Do you expect mainstream gamers to port open-source Doom on niche boxes?

Not at all - just saying that for skilled developers this is not so difficult as it may look on first. That's all.

Originally Posted by hammer

Newtek VT remained as a product niche. The same for DarCo.

This is obvious and i didn't claimed anything else.

Originally Posted by hammer

When I needed an NLE (MPEG2, Firewire) solution for a non-profit organization in 2002-2006, it was not Newtek. MPEG 2 NLE transition for Amiga wasn't good. Apple made FireWire standard. My nForce 2 motherboard had Firewire and hardware (GeForce 4 Ti's hardware MPEG2)/software support. The gaming PC's use case entry point leads to other use cases.

MPEG2 is NOT NLE friendly (many reasons). Firewire could be added to Amiga but Amiga was no longer developed and produced - EOT.

Originally Posted by hammer

Microsoft recognized the "gaming PC" and Xbox are grassroots campaigns.

Apple also recognized "grassroots" gaming and M2/M3 IGPs are pretty good.

When the Amiga lost its "powerful gaming for without the price" image, the Amiga platform rapidly declined along with Amiga's non-gaming use cases.

How conveniently you omitted information about Apple failed game console (infamous Pippin)... Compare its preliminary spec with CBM and all you already wrote about Apple advantages over CBM.

Originally Posted by hammer

I did state the platform must have strong 1st/3rd party game developers or strong gaming hardware for the money.

Sony's PS1 had both factors and it dominated.

All true and i agree with you but even PS1 had very difficult development and it was serious risk it will never happen yest it happened and created new market where also Microsoft was beneficial (but not Apple).

Originally Posted by hammer

I remember a 1990s review of the A3000 from mainstream computer magazines and they criticized aging graphics capabilities. In 1991, Commodore released the A2410 graphics card for AMIX (Amiga Unix) based on foreign Texas Instruments TMS34010 (TIGA).

On PC, TIGA is used as an IBM 8514 clone.

Once Again 8514 is NOT TIGA - in case of TMS340x0 - API of 8514 is usually emulated trough TIGA API however if you are searching performance this is probably NOT the best approach - TIGA was offered by TI as preferred solution for TMS340x0 - CBM however could not use TIGA but run own RTG API directly on A2410. As always there was some choices done...

[pandy71]

Originally Posted by sandruzzo

Even "only" 14mhz 32bit Blitter(with 64 bit transfer rate) with Aligned Destination to byte(to do chunky), and One more register that points to the display list would have been a great and very powerful upgrade.

Blitter could be internally faster, if all IC's integrated into single IC (cost reduction wise approach) then internal communication can be arranged in different way (similar to A3000), blitter could be clocked faster and use something like Write Back delay mechanism to improve performance - this is not even big RAM penalty solution - like 64..256 bits in total (assuming 6..8T SRAM design this is like 2048 transistors). Single..two cycles latency in blitter also probably unimportant from developer perspective...

[Promilus]

Originally Posted by Bruce Abbott

PC cloners focused on making the standard PC architecture cheaper. This was mostly done by using far East manufacturing plants, and later on integrated chipsets incorporating the functions of standard ICs.

So like Philippines and Malaysia used by Commodore as well? And what kind of chipset you mean integrated "standard ICs" in PC? I mean like... for example Akiko vs Gary or Gayle?

[pandy71]

Originally Posted by Promilus

So like Philippines and Malaysia used by Commodore as well? And what kind of chipset you mean integrated "standard ICs" in PC? I mean like... for example Akiko vs Gary or Gayle?

Akiko, Gary, Gayle are non standard chips and mostly they are combined standard logic TTL into some form of ASIC.
In PC there is exactly opposite situation, it is made from standard IC's like 8237, 8259, 8254, MC146818, 74612 so it was obvious to create something like 82C206 where all "custom" PC logic was integrated in single IC (84 pin PLCC vs over 22 pin in various size DIL)
Seem Akiko is most closest to this concept integrating 2 CIA's inside.

[babsimov]

Originally Posted by Bruce Abbott

Which was pushing in the wrong direction IMO. The problem was that the engineers were designing what they wanted, not what the market wanted. We also see this with Hedley Davis's A2024 monitor design, which should never have gotten past the 'what if' stage. It's a real shame that valuable resources were wasted on stuff like this - then Medhi Ali got blamed for killing these unprofitable projects.

Sorry to disagree, from the start the Amiga need a 31 khz 640x512 color display at minimum. To compete with ST, MAC and PC in the business market i'ts needed. Don't have it it's one of the reason DTP company don't see the Amiga as real platform for that.

Having a 640x400 31 khz display make the ST got more office software even localized than the Amiga.

Jay Miner was right about the Ranger 31 khz display.

And, it's so true that Commodore engineers will keep the Ranger by just removing the VRAM deemed too expensive. The Ranger was renamed later ECS (in fact, internally the code name before becoming ECS ??was The HIRES chipset). Commodore engineers just milked Jay Miner's latest work for Commodore to the bone. The Ranger was finalized and Jay Miner said so publicly. The 500 and the 2000 should have been equipped by default with its light version which we know under the name ECS. The AGA chipset is just the Ranger light "ECS" with little improvement to be able to display 256 colors and an HAM8 mode.

In 1998 Microway released their AGA-2000 'flicker fixer' for the A2000, which was a more elegant solution because it displayed all native Amiga resolutions on a standard VGA monitor. Its only problem was not having enough memory to do the full 256/512 lines in PAL, but it was close enough for most purposes.

Thanx for the link. A native 31 khz chipset like Ranger/ECS is better than a third party add on. If native all software company use it or are sure the Amiga is capable of 31 khz natively.

It's ironic that the A3000 was the first machine to have the ECS chipset, when it also had an enhanced flicker fixer that made ECS productivity mode redundant. Commodore later released the A2320 flicker fixer as a separate card for the A2000.

The ECS is obsolte for 1990. To me, the 3000 was my biggest disappointment with Commodore, far ahead of the 1200 or the 4000. It brings nothing in terms of displaying more colors, while we were hoping for at least 256 colors, nothing in terms of sound ( the AGA will be the same at this level), the case is ultimately worse than that of the 2000, because not even a 5"1/4 bay and physically the 3000 was not compatible with the video toaster which nevertheless made the success of the desktop range in the USA. And even more, the 3000 was overpriced.

AGA is competitive with typical ISA bus PC VGA cards. The only thing 'missing' is 256 color chunky pixels. But AGA greatly improved other features such as sprites (widened from 16 pixels to 64, a big improvement) HAM8 (for true photographic quality images) and a 24 bit color palette (when VGA was only 18 bit). The wider bandwidth also increased blitter throughput by up to 60% in typical applications, when VGA had no blitter at all. Compared to a typical ISA bus VGA card, AGA was actually more capable.

SVGA is a catchall for anything beyond standard VGA, but on most cards this only meant 640x480 in 256 colors and 800x600 in 16 colors. Both modes were rarely used. Most apps assumed 640x480 in Windows, and cheap VGA monitors struggled to provide sufficient resolution in 800x600. On most cards 640x480 in 256 colors was painfully slow. Most games stuck to 320x200.

Sorry AGA 256 colors 640x512 or more at 31 khz are useless and unusable. AGA is acceptable for 320x200 au 640x256. Commodore sold us smoke and mirrors with the AGA. On paper it looked amazing, in reality it wasn't that at all. The OCS never disappointed anyone and there it was really competitive with the competition and it remained so for a while given what it was possible to get out of it. The ECS and the AGA were both obsolete when they were marketed, but we didn't know that, well for the ECS we knew it, but for the AGA it could create an illusion for a while.

For 1992/93 AGA is as obsolete as ECS was in 1990. And this is easily understandable since the AGA is based on the Ranger 86/87 technology. So of course the AGA was, finally, a step forward for the Amiga in terms of colors, that's already it.
That being said, make no mistake; i was happy with my 1200 until the end (1998) despite everything.
What really kept me on Amiga was AmigaOS above all.

It's nice to dream. But 1988 wasn't going to happen. 1990 might have been doable if the engineers didn't have their hearts set on high-end graphics. AGA in the A3000 would have been awesome - then I wouldn't have needed to buy an A1200!

Yes it's a little dream, if Commodore have made the right decision and fund research. To me Jay Miner should have push the Ranger to the AGA level and a little beyond (better blitter, better copper, more sound).
But of course in reality something like AGA in the 3000 from 1990 would have been awesome. Even more awesome with a DSP.

You should bear in mind that the A1200 was designed to be a low-end machine, with the base model selling at the lowest possible price. However Commodore did provide a hard drive as default in the A1200HD/40. Almost all the A1200s I sold in my shop were this model. If these were not so common in the UK or elsewhere it was because the vendors didn't think customers would want them.

Remember the real AGA entry level from the start is the A1000+ and it have a hardrive by default. The target price is similar to the price I paid for my 1200 in 1993 with a 2.5 inch 60 MB hard drive.

For the 1200 Commodore should have use the A500 casing (like Atari with the Falcon) and use a 3,5" hardrive by default. After 1990 the hard drive by default was something expected by publishers, since it was the rule on PC and MAC.

If Commodore want an entry level computer without hardrive, then keep the 500+ on market, there is a lot of third party hardrive expansion. Or better add to the 500+ an internal IDE connector.

Before Ali and Sydness messing with the Amiga, the planned product line was A500+ (ECS) A1000+ (AGA) and A3000+ (AGA+DSP)

Well, the 3000 keep its "questionable" case but at least this range was coherent and above all would have been marketed a year earlier !

[Promilus]

Originally Posted by pandy71

Akiko, Gary, Gayle are non standard chips and mostly they are combined standard logic TTL into some form of ASIC.
In PC there is exactly opposite situation, it is made from standard IC's like 8237, 8259, 8254, MC146818, 74612 so it was obvious to create something like 82C206 where all "custom" PC logic was integrated in single IC (84 pin PLCC vs over 22 pin in various size DIL)
Seem Akiko is most closest to this concept integrating 2 CIA's inside.

Intel 8237 is as much custom chip as CIA (and funny enough CIA itself does kind of resemble both intel and motorola IO devices). Same could be said on many PC components like floppy controller which at some point indeed started to integrate LPT, COM and so on... So what, it is just combining several custom chips into a bigger one.
"Standard" ICs I'd say it is from 74 lineup or CMOS equivalent. And sure, there were plenty of those replaced easily with CPLDs, right? Right. But either way on PC world integrating more things into single chip ain't that different than what amiga ppl did. And that was just yet another example of Bruce's bias... Same with those mythical crashes of apps on Windows ... as if apps didn't crash on AOS along with the whole system and bad instalation of e.g. RTG drivers didn't result with pretty much messed up system. I can assure you it is fairly easily to even trip linux unresponsiveness and then after forced restart you end up with unusable OS due to RO filesystem (which can be fixed as is safety precaution in case some disk errors ocurrs so running FSCK and remounting as R/W fixes the issue). Also ... all that info about ranger chipset - if true (and I'm not so sure it is) just proves they had ECS-capability in '88 and chose to postpone it till mid '90 in case of big amiga and '91 in case of small amiga. With pretty obvious result, right? Right...

[pandy71]

Originally Posted by Promilus

Intel 8237 is as much custom chip as CIA (and funny enough CIA itself does kind of resemble both intel and motorola IO devices). Same could be said on many PC components like floppy controller which at some point indeed started to integrate LPT, COM and so on... So what, it is just combining several custom chips into a bigger one.
"Standard" ICs I'd say it is from 74 lineup or CMOS equivalent. And sure, there were plenty of those replaced easily with CPLDs, right? Right. But either way on PC world integrating more things into single chip ain't that different than what amiga ppl did. And that was just yet another example of Bruce's bias... Same with those mythical crashes of apps on Windows ... as if apps didn't crash on AOS along with the whole system and bad instalation of e.g. RTG drivers didn't result with pretty much messed up system. I can assure you it is fairly easily to even trip linux unresponsiveness and then after forced restart you end up with unusable OS due to RO filesystem (which can be fixed as is safety precaution in case some disk errors ocurrs so running FSCK and remounting as R/W fixes the issue). Also ... all that info about ranger chipset - if true (and I'm not so sure it is) just proves they had ECS-capability in '88 and chose to postpone it till mid '90 in case of big amiga and '91 in case of small amiga. With pretty obvious result, right? Right...

Nope - i will agree with you if 8237 (8257) wouldn't be available from various sources - even Soviet Union produced 8257 as KR580IK57 - 8237 was officially and unofficially cloned and made as standard LSI part.
CIA's was not substantially different than 8255 + 8253 (8254) however could be used out of Amiga ecosystem - Akiko, Gayle, Gary etc rather not.
So standard is more about versatility.
I will agree with you on rest - ECS was too late, CBM should focus on AGA functionality and it should be available before 1991 at worst.

[Bruce Abbott]

Originally Posted by babsimov

Sorry to disagree, from the start the Amiga need a 31 khz 640x512 color display at minimum. To compete with ST, MAC and PC in the business market i'ts needed.

Ah, so that explains why the ST became so successful and the beat the PC and Mac, since neither of them had 31 kHz!

[Bruce Abbott]

Originally Posted by pandy71

In fact it is quite easy and it may be implemented on at least few ways - simplest is just negate (make it complementary) PWM AUDxVOL signal and use 8 bit registers in front of DAC to hold sample value and switch between them - silicone area can be reduced as instead 4 DAC's you could just made 2 DAC's and introduce time interleaving (this would made also possible to demux channels and allow in theory possibility to have 4 independent channels). Silicone area could be reduced after such changes - also if PWM replaced by binary rate multiplier like SN7497 overall audio quality can be improved (in theory no hard limit of 56kHz) and made all above way easier.

So you agree that a complete redesign would be needed.

This was possible at worst in ECS Paula - shame on Commodore that they not implemented new Paula features in AGA.

They were working on it, but couldn't get it finished in time. I don't know what their plans were (rumor was more sound channels) but I would rather have an A1200 with the standard Paula than nothing. What was so bad about it? This idea that Commodore should have been ashamed for not putting stereo panning in Paula is just more BS. So sounds come out one side or the other, big deal. Most Amigas were probably being used on TVs that didn't even have stereo sound.

No cache needed just WAIT (DTACK) generator.
Small cache like 256 bytes or 512 bytes was OK - perfectly feasible even for CSG - single IC with all logic and SRAM inside running at 14..28MHz - overall complexity comparable to Amber/A2024 IC's

More armchair engineering.

Nope - ECS didn't introduced anything - CBM should made AGA instead ECS - ECS could be reasonable if it was integration of OCS - from 3 IC's to 1 IC - so standard cost reduction approach.
A3000 with AGA half year later could be significant design in Amiga portfolio.

ECS introduced 31kHz video, which according to some is what the Amiga desperately needed to make it in the business world. Yet AFAICT Amiga fans completely ignored it. ECS also introduced 2MB ChipRAM. A lot of people did want that. And It introduced PAL/NTSC switching, which people wanted too.

A3000 with AGA a half year later would certainly piss off all the people who already had one (or do you mean delay the A3000 until they had AGA? Might never have arrived if they tried to do that). A better idea would be a cheap Zorro-II/III graphics card and RTG, or even just a bridgeboard that allowed you to use any PC VGA card (on the A2000 too).

Still, I guess we can add "I was disappointed in the A1200 because the A3000 didn't have AGA in 1991" to the list.

[Promilus]

Originally Posted by pandy71

Nope - i will agree with you if 8237 (8257) wouldn't be available from various sources - even Soviet Union produced 8257 as KR580IK57 - 8237 was officially and unofficially cloned and made as standard LSI part.
CIA's was not substantially different than 8255 + 8253 (8254) however could be used out of Amiga ecosystem - Akiko, Gayle, Gary etc rather not.
So standard is more about versatility.
I will agree with you on rest - ECS was too late, CBM should focus on AGA functionality and it should be available before 1991 at worst.

Well Soviet Union along with many companies worldwide did produce 8086 microprocessor (and similarly same happened with 68000) - that's not something which turns 8086 into "standard part". And most of the components you did list plays same role and are made the same way as parts of Amiga chipset. So due to simple fact PC had clones those components are "standard parts" and for amiga the very same thing is "custom chip"? WTF? Akiko couldn't be used outside Amiga ecosystem? Why such claim? Possibilities exists, they are just generally unexplored due to low reward of doing such thing. And generally all those parts you did mention were designed for intel 8085 (8080 updated) but since intel was generally fairly compatible in 8086 were used also for those systems. That doesn't make them any more standard than anything Motorola made for their 68k systems along with 68881 and 68451...

[PortuguesePilot]

From what I read in this thread about the "ideal Amiga in 1991/1992" it looks very much like a Sharp X68000 or a Sharp X68030... Which is interesting.

In spite of all of its hardware might, the Sharp beast ALSO folded under the "PC compatibles" weight roughly at the same time the Amiga did.

[Cris1997XX]

Well, the FM Towns had similarly powerful hardware but with a x86 architecture instead (The first model shipped with a 386 IIRC). Maybe that would've been the ideal Amiga?

[babsimov]

Originally Posted by Bruce Abbott

Ah, so that explains why the ST became so successful and the beat the PC and Mac, since neither of them had 31 kHz!

PC and MAC have a HR resolution for office work natively

At the time of Ranger desgin PC have EGA : 640x350 (and later VGA 640x400) MAC : 512x342 without flicker, ST have the 640x400 monochrome, Amiga don't have something similar.

A native non flicker mode is needed to compete, and Ranger/ECS was designed for that, not to give Amiga more colors.

The Ranger target is office market to compete with PC/MAC/ST in this market.

As i say, you can't expect thrid party add on (work around) to be equivalent to a native non flicker display.

If I understand you correctly, you think that the Amiga should have remained at PAL/NTSC resolutions throughout its entire lifespan and hope users buy flicker fixer for more ?

ECS introduced 31kHz video, which according to some is what the Amiga desperately needed to make it in the business world. Yet AFAICT Amiga fans completely ignored it. ECS also introduced 2MB ChipRAM. A lot of people did want that. And It introduced PAL/NTSC switching, which people wanted too.

The Ranger is part of the next generation Amiga designed by Jay Miner using a 68020 an planned to release for 1987/88. I think Jay Miner well know what Amiga need to compete in the market, sorry.

They were working on it, but couldn't get it finished in time. I don't know what their plans were (rumor was more sound channels) but I would rather have an A1200 with the standard Paula than nothing. What was so bad about it? This idea that Commodore should have been ashamed for not putting stereo panning in Paula is just more BS. So sounds come out one side or the other, big deal. Most Amigas were probably being used on TVs that didn't even have stereo sound.

A "superPaula" was proposed for 1988 by Glen Keller, with 8 channel 16 bit and native SCSI controller. Proposal was rejected because, at the time, better sound was not the priority.
The only audio evolution was MARY for the AAA chipset. It was the only finished/working chip from the AAA.

[Thorham]

Originally Posted by Bruce Abbott

ECS introduced 31kHz video, which according to some is what the Amiga desperately needed to make it in the business world. Yet AFAICT Amiga fans completely ignored it.

I certainly enjoy it on my A1200 with LCD monitor