@bhabbott

Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: HAM8 isn't true color AND it USE a palette of 64 colors. You don't have to use the base colors, and then you have 3 possibilities left for each pixel:- set red / hold green and blue, set green / hold red and blue, set blue / hold red and green. It takes 3 pixels to fully change from one 18 bit 'true' (not paletted) color to another. This produces color fringing on vertical edges, but in Superhires the fringe is only 1.5 hires pixels. On a standard TV tube this is about the same size as the dot pitch. In composite its invisible.

I know it, thanks.
Quote:

Here's something interesting about how the Amiga got HAM mode... Quote: Amiga project lead Jay Miner relates: "Hold and Modify came from a trip to see flight simulators in action and I had a kind of idea about a primitive type of virtual reality. NTSC on the chip meant you could hold the hue and change the luminance by only altering four bits. When we changed to RGB I said that wasn't needed any more as it wasn't useful and I asked the chip layout guy to take it off. He came back and said that this would either leave a big hole in the middle of the chip or take a three-month redesign and we couldn't do that." 3 months just to take a feature off the chip! With today's design tools it would only take a few minutes.

And this as well.
Quote:

Quote: HAM8 needs an optimized palette to give its best quality. It means that quite some time is needed to scan the original 24-bit image and calculate a set of good 64 base colors. Only if you need the absolute 'best possible' quality.

Hey, and what I've said? It's also written above!
Quote:

"Oh no!", says the Amiga fan, "The technique used didn't produce the absolute best possible quality! Not acceptable!!!". Hyper-perfectionist Amiga fans, always enemies of the good.

And here starts your delirium, inventing non-sense things that I never said neither thought...

@agami

Quote:

agami wrote: @cdimauro Quote: Then how do you justify the Ranger chipset development, since he started already on 1986? As I said, engineers are always working on something. And just because the new Amiga team in Commodore received some funding in those early years, does not mean that they received sufficient funding to deliver a new chipset at speed. The fact that Ranger never went past an early prototype is indicative of the "tinkering" project that it was.

But at least it had funds for the prototype, which is a very good step.
Quote:

You have no idea how much I wish the purchase of Amiga by CBM was a strategic one, instead of the tactical move to deny Atari the tech. Having a talented team and bunch of patents, without a clear direction and purpose meant that CBM were going to play some Spaghetti-Wall. They were not a chipset company before that, and they were certainly not magically transformed into a chipset company simply by purchasing Amiga.

Ehm, Commodore WAS a chipset company: from PET to Vic20, C64, C16/Plus, C128, etc.. Almost all chips from the chipset were internally developed.

You can see this as well with the new projects around and after the Amiga time.
Quote:

Neither were Apple, Atari, or Acorn, chipset companies.

All of them developed their chipsets as well.
Quote:

They were computer companies like Commodore, that for the most part developed computers using "graphics, audio, I/O, etc." from other companies who's business was to focus on those components.

SOMETIMES it happened but Commodore developed also several chipsets.
Quote:

In the nascent days of the personal computer revolution, it was not uncommon to discover that none of the component OEMs had a part that fits a requirement, so some customisation is commissioned. That doesn't make these companies chipset companies.

I beg to differ: see above.
Quote:

The PC certainly doesn't count as it was not a single company, and it was the original Open Computing Platform.

Well, the PC has only discrete components from third-parties.
Quote:

You can focus on chipsets if you like, but ultimately it comes down to business management. Companies that managed R&D better, released improvements at a faster pace. Companies like Commodore that were not very good at managing R&D, released improvements at a slower pace. It's really not that complicated.

Yes, we know that Commodore sucked at managing.
Quote:

I've mentioned it before: Commodore should've kept up with Moore's Law and maintained a 2-year release cadence. The engineers I'm sure could've manage it, but that would require some serious business acumen; something Commodore never had.

I agree.
Quote:

Things that make it challenging for a single person to develop and bring to market CPU/GPU innovation in 21C: - Moore's Law has increased the complexity since 1985 by 2E+19. Sure, we have much more advanced tools, and the costs of production are much more affordable, but

Unfortunately production costs increased A LOT over the time.
Quote:

- We are not in the "Wild West" of the '80s personal computer market. Things are a lot more commoditised, and beyond the academic curiosity, a single person would be going up against multi-national giants in the industry.

Not without begin backed by a lot of money.
Quote:

- The broken patent system also keeps most single persons from even trying to start a CPU/GPU innovation project. They might have the engineering skills, but they know they are clueless about IP law and have read enough ugly stories in the media to dissuade them from going out on their own, and then choose to join one of the big firms instead.

Sad, but true.
Quote:

- The definition of base infrastructure has changed. Where in the '80s and most of the '90s it was about coding directly to HW, today it's all about the APIs. Remember in the '90s when even the average consumer cared about what GPU was in their game console? 16-bit, 32-bit, 64-bit. Today, I'm not even sure that most game devs care about what GPU is in the Xbox Series X or PS5. Can it Run UE5? is the main question.

Indeed. And for very good reasons...

@bhabbott

Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: So, according to you the Amstrad CPC464 was an Amiga competitor. In which parallel universe? In the 'parallel universe' that I lived in. In 1985 I bought a CPC664. I could have had an A1000 if I tried hard enough (they weren't sold in New Zealand at the time), but the Amstrad was a better fit to my needs at the time. 2 years later I did buy an A1000, but I kept using the Amstrad for several years after that, partly because there was an active local user group and nothing for the Amiga.

I did the same with my Plus4 first and C128 after: the Amiga 1000 was too expensive.

But this does NOT mean that they were products for the same market segment. In fact, they were home computers whereas the 1000 was a personal computer.
Quote:

Quote: The Amiga competitors were Apple Macintosh, Atari ST, Acorn Archimedes and, of course, PCs. First off, I don't class the PC as a 'home computer' in the traditional sense.

In fact all of those were Personal Computers.
Quote:

In New Zealand it was used almost solely by businesses. The Acorn Archimedes was hardly competitive. It was very expensive here and had little support. No retail stores sold it. Ditto for the Apple Macintosh. These machines filled a very niche market here (schools and universities).

Nevertheless, they were all personal computers. Definitively NOT home computers.
Quote:

But let's take a look at the Archimedes:- Quote: The first models were released in June 1987, as the 300 and 400 series. The 400 series included four expansion slots and an ST-506 controller for an internal hard drive... Speculation gathered pace about new machines in the Archimedes range in early 1989, with commentators envisaging a low-cost, cut-down model... The new model sported only a single internal expansion slot, which was physically different from that of the earlier models. In late 1991, the A5000 was launched to replace the A440/1 machine in the existing product range. With the existing A400/1 series regarded as "a little tired", being largely unchanged from the A400 models introduced four years previously,... The A5000 featured the new 25 MHz ARM3 processor, 2 or 4 MB of RAM, either a 40 MB or an 80 MB hard drive and a more conventional pizza box-style two-part case... offering resolutions of 1024 x 768 in 16 or 256 colours and with 24-bit palettes... The A5000 (along with the earlier A540) supported the SVGA resolution of 800 x 600 in 16 colours, although the observation that "Archimedes machines have simply not kept pace" arguably remained... In 1992, several new models were introduced to complement the A3000 and to replace the low-end A400 series models - the A3010, A3020 and A4000... The ARM250, running at a higher 12 MHz clock frequency and used in conjunction with faster 80ns memory chips, compared to the 8 MHz of the ARM2 and the 125ns memory of the A3000, gave a potential 50% performance increase over such older systems A bit late to the party, but by 1992 the Archimedes equivalent of the A1200 and A4000 sported a mere 50% performance increase and no significant change to the graphics and sound subsystems in 5 years. 'Evolution'? Not so much. "But but", you say, "their high-end models evolved much more than the Amiga!"

They already started on '87 with a phenomenal chipset. Have you ever read its features? Here is it: https://en.wikipedia.org/wiki/Acorn_Archimedes#Hardware

The Archimedes was founded on its ARM processors for everything else, so the company then mainly focused on its development (because, as I've said, the starting hardware was very good).
But it had updates on other hardware parts. Check them here:
https://en.wikipedia.org/wiki/Acorn_Archimedes#A540
https://en.wikipedia.org/wiki/Acorn_Archimedes#A5000_and_A4_laptop
Quote:

Quote: The introduction of the 68040... put ARM3-based Archimedes models at an increasing performance disadvantage. An Amiga 4000 with 68040 CPU (or suitably upgraded Amiga 2000) could achieve a reported 18.7 - 21.6 VAX MIPS... Against such performance ratings only Acorn's Risc PC 600 (18.4 VAX MIPS[360] to 21.8 VAX MIPS[363]) fitted with an ARM610 CPU would be able to respond.

Where is the "increasing performance disadvantage" here?
Quote:

The ST was competition for sure. So how much did it 'evolve'? Quote: In late 1989, Atari released the 520STE and 1040STE (also written STE), enhanced version of the ST with improvements to the multimedia hardware and operating system. It features an increased color palette of 4,096 colors from the ST's 512 (though the maximum displayable palette without programming tricks is still limited to 16 in the lowest 320 × 200 resolution, and even fewer in higher resolutions), Genlock support, and a blitter co-processor (stylized as "BLiTTER") which can quickly move large blocks of data (particularly, graphics data) around in RAM. The STE is the first Atari with PCM audio; using a new chip, it added the ability to play back 8-bit (signed) samples at 6258 Hz, 12517 Hz, 25033 Hz, and even 50066 Hz, via direct memory access (DMA). The channels are arranged as either a mono track or a track of LRLRLRLR... bytes. RAM is now much more simply upgradable via SIMMs. 4 years of 'evolution', copying features the Amiga already had and not even getting to OCS level. But... Quote: The final model of ST computer is the Falcon030. Like the TT, it is 68030-based, at 16 MHz, but with improved video modes and an on-board Motorola 56001 audio digital signal processor. Like the Atari STE, it supports sampling frequencies above 44.1 kHz; the sampling master clock is 98340 Hz (which can be divided by a number between 2 and 16 to get the actual sampling frequencies). It can play the STE sample frequencies (up to 50066 Hz) in 8 or 16 bit, mono or stereo, all by using the same DMA interface as the STE, with a few additions. It can both play back and record samples, with 8 mono channels and 4 stereo channels, allowing musicians to use it for recording to hard drive. Although the 68030 microprocessor can use 32-bit memory, the Falcon uses a 16-bit bus, which reduces performance and cost. In another cost-reduction measure, Atari shipped the Falcon in an inexpensive case much like that of the STF and STE. Aftermarket upgrade kits allow it to be put in a desktop or rack-mount case, with the keyboard separate. Released in 1992, the Falcon was discontinued by Atari the following year. Finally some real evolution! After 7 years. But still not much competition, as it didn't even last a year. By this time AGA Amigas were out.

As usual, you report only what you like. Here's the full evolution for the Atari ST series:
https://en.wikipedia.org/wiki/Atari_ST#Mega_models
https://en.wikipedia.org/wiki/Atari_TT030#Technical_specifications
https://en.wikipedia.org/wiki/Atari_Falcon#Specifications

Take a look at the introduced new features over the time.
Quote:

But even before then Commodore had 'evolved' the Amiga by a fair bit, with ChipRAM increased to 2MB, productivity mode, flicker fixer, 020 and 030 accelerators, 7 port serial card and Ethernet card for the A2000, hard drive + FastRAM for the A500, PC in an addon box for the A1000 and 'Bridgeboard' cards for the A2000. If we consider 3rd party devices (as we would for the PC) then by 1992 the Amiga had a 40MHz 040 with buffered serial port and PC compatible parallel port, 24 bit graphics at up to 1600x1200 with 440 MB/s graphics bus bandwidth and 50 MB/s CPU bus bandwidth, 16 bit audio with 8 tracks at 44kHz and direct to disk recording, Video CD/CDI MPEG 1 video decoding and playback.

Sure, and it happened also on other platforms.

You missed the Apple Macintosh here. Guess why... And here are the models and their evolution:
https://en.wikipedia.org/wiki/Macintosh_II
https://en.wikipedia.org/wiki/Macintosh_IIx
https://web.archive.org/web/20170119115038/http://tech-insider.org/mac/research/acrobat/9003.pdf
https://en.wikipedia.org/wiki/Macintosh_Quadra
Take a look at the time and their features.

Apple also introduced the II GS: https://en.wikipedia.org/wiki/Apple_IIGS
With an outstanding audio section...

So, you cherry-pick only what you like.

And, as I've said, there were also the PCs.
On '87 IBM presented the VGA, yes, but also the 8514: https://en.wikipedia.org/wiki/IBM_8514 Take a look at its features, which were quite advanced for the time.
And for course SVGAs came already on the same year: https://en.wikipedia.org/wiki/Super_VGA
And regarding audio, here's a small part: https://en.wikipedia.org/wiki/Sound_card#List_of_sound_card_standards
And don't tell me that all the companies which developed that stuff were so big as IBM...

So, again, you report only a small part of the history.

@Hammer

Quote:

Hammer wrote: The 1st to use 80286 was Intel's MULTIBUS-based System 286/310 microcomputer before IBM's 1984 PC AT.

Yes, I know that - but information on this systems is hard to find. It was apparently released in 1983. It was not intended to be used as a home or personal computer.

Quote:

It's not the 1st time IBM was late to the party when IBM's 386-based PS/2 was released later than Compaq's 386 PC.

Compaq certainly pulled out all stops to produce the Deskpro 386. But the result was very expensive, and the 32 bit bus proprietary. Very few PC users needed it or could justify the price. If you think the existence of this machine means the Amiga should have used a 386 CPU, you are wrong. First off it was much too 'late to the party', and it was in a completely different class. 6 years later the A3000 was a match to the Deskpro 386 selling at that time (I considered buying one instead of the A3000, but luckily I decided against it - the A3000 was much more upgradable).

Quote:

1985 Amiga 1000's low sales count acted like dev kits or "founder edition" for the 1987 Amiga 500.

True, but nevertheless it was also a practical machine. Understandably Commodore was reticent to promote it too heavily when the OS wasn't finished, however that didn't stop Atari from doing it with the ST. This was a mistake on Commodore's part. Early developers such as Electronic Arts were very keen to produce software for it, but the low sales were disappointing.

Quote:

Motorola's 68K CPU technology leadership stalled about the mid-1980s and Motorola was focusing on RISC based 88000 development and which was released in 1988.

Yes. Motorola struggled to keep up with Intel both in CPU design and chip manufacturing. Part of the problem may be lower demand, another was different strategies. Intel was fully committed to the PC and stood to win big if they could beat off the competition. They concentrated on achieving the highest possible performance even if it meant overclocking the CPU by 4x the bus speed or clamping a huge heatsink and fan onto it - because the PC world demanded it. Motorola was more into the high performance embedded market where lower power was important, with desktop CPUs almost being a sideline.

Quote:

Too bad Motorola didn't glue some 1.0 IPC DSP extensions into 68K

I don't think it was too bad. The 68060 matched Pentium performance well beyond the Amiga's commercial lifespan. Had they continued to simply speed it up we would have been quite satisfied.


Quote:

Acorn's RISC-based ARM was the solution after Commodore's crap 65 series CPU R&D roadmap. ARM v2 already has a strong IPC when compared to 68000 and 68020.

Acorn's RISC PC was a failure. They were reduced into becoming just an IP vendor for the ARM architecture. Their 'roadmap' was not a good one for Commodore to emulate.

Quote:

286's MMU benefited MS Xenix and IBM/MS OS/2 development.

Both of which had minimal market penetration. The 286's MMU was widely recognized in the PC world as being flawed, and they were desperate for something better. That's why Intel developed the 386SX, to replace the 286 in low cost 16 bit motherboards.

Quote:

Windows/386 introduced a protected mode kernel that allowed several MS-DOS programs to run in parallel in "virtual 8086" CPU mode.

But only useful for well-behaved DOS programs, not games or apps that needed all the resources of the machine.

Of course the Amiga didn't have that problem because Amiga OS was fully multitasking with a flat 32 bit memory map from the start.

Quote:

Windows NT 386 needs i386's MMU and PC world was building a large install based on i386 MMU capability for a longer time duration. [snip] PC world's long build-up with TPM 2.0 install base tactics was used before pulling Windows 11 TPM trigger.

Blah blah PC blah blah. I'm sick of hearing about the PC. This site isn't called Amiga World for nothing!

@bhabbott

Quote:

Only Zorro III cards (or controllers on the CPU board) can be 32 bit. Several A1200 accelerators had optional on-board SCSI ports. I had the DKB Cobra with Ferret SCSI board. I don't know if this used DMA (based on the small size and number of 'glue' chips in the Cobra I suspect not). The Blizzard 1230IV that I have now definitely does, but I don't have a SCSI board for it (or any SCSI devices to use with it).

I also have a Cobra and Ferret. Ferret particularly annoying as I always had trouble with that stax connector.

I have a CyberSCSI in my A4000 connected to my CyberStorm 060. So a CPU card with local SCSI. Even so, without Zorro limitations, I recall it still needed particular max transfer.

Quote:

This is how it should have been done for faster IDE too. Putting a 32 bit DMA controller on the motherboard would have been expensive, limited the speed, and probably buggy. The A1200 was designed as a low cost home computer to replace the A500+, but with open-ended expansion to much higher performance via the 32 bit expansion slot. Putting a high performance DMA hard drive interface on the motherboard didn't make sense for its intended market. You could argue that the A4000 should have had a better IDE interface, but it had the slots to take SCSI cards that professionals preferred.

I expected the A4000 cards to have 32-bit max transfers. Not always mentioned but I kept getting crashes when setting up an A4000 with SCSI HDD I didn't see with my A1200. Turned out it was max transfer and it does matter what it is set at.

Quote:

No, it wasn't.

Reducing CPU power wasn't needed on low power models? That doesn't make sense. The more CPU power resources drain the slower it will be.

Quote:

Most A600s and A1200s were sold without a hard drive, and many owners never installed one because they were happy enough with floppy disk speeds. Even with 'slow' PIO the A1200 can achieve over 2MB/s with a good hard drive. This is a huge speed boost over floppies. Sure it loads the CPU, but in typical use there isn't much the CPU would be doing anyway during eg. loading a game.

It is near impossible to use it for practical use on floppies alone so any real use would have required a HDD. I know. I had an A500 like that with no extra drives. KindWords came on three disks and was not designed for floppies, asking for each one in a row and only loading a few KBs before wanting the next one, 20 times in a row or something. I managed to squeeze it down to one floppy so I could use it, without binary compression, and it was a lot of work.

Quote:

Sure, but consider the time frame. Faster PIO modes were not standardized until 1994. The A600 was released in March 1992. The specification sheet for Gayle is dated July 1991, 3 years earlier. How was Commodore expected to support faster modes without a proper standard to work from?

The original ATA spec is said to have included support for PIO modes 0 to 2 and even DMA modes 0 to 2. It may have been standardised in 1994 but Commodore still implemented ATA so they could support it with the information that had.

And with PCMCIA Commodore had already implemented a socket before it was standardised. But it turned out to be more buggy.

http://209.68.14.80/ref/hdd/if/ide/stdATA-c.html

http://www.os2museum.com/wp/historical-ata-standard-drafts/

Quote:

Coincidentally 1991 was the year I started my business selling Amigas and PCs. I remember having a problem with new IDE hard drives and certain motherboards, which according to my supplier was was caused by logic voltage level mismatch. I don't know if this was true, but I do know that we had significant compatibility issues with IDE back in those days. \

Did you see any problems with Amigas? The A4000 could take a full desktop drive. So be interesting if same drives has any conflict there.

In any case, adding an IDE port to A600 and A1200 was a good decision and solved clunky problems of the A500 with add ons like the A590. Of course, it had to be added internally, so best installed as a dealer deal to avoid warranty issues and for those who weren't tech heads.

The only major criticism levelled at IDE was using a 44 pin laptop IDE port which meant smaller and more expensive drives wee needed. However they weren't full desktop computers and space was tight. We did see A1200 hacks for full drives but I thought that was a bit over the top.

Quote:

Only one 'serial' port in the CIAs is used, for the keyboard interface. The other one is spare (I used it to remotely control a Panasonic video recorder).

It's detailed in Mapping the Amiga. But it's slightly unclear as it mentions CIA B SDR connected to internal lines but not used by OS.

How did you attach to the serial port?

Quote:

Most other home computers used a standard 'UART' chip for RS232 serial, but this was almost always an option provided on a plug-in card, not a part of the base machine. Some machines used nasty bitbanged serial, which was OK for printers but not good for bidirectional communications.

The C64 had serial, as used with the 1541, but it wasn't the best example around.

Quote:

No, it was certainly compliant enough.

It can't operate at the exact 31,250 baud.

Quote:

The only problem was that in a multitasking environment it was easily possible to overrun the serial receive buffer when reading MIDI commands from a synthesizer keyboard (imagine pressing 10 keys at once, with each one sending 3 bytes for key down). This problem could have been solved by restricting multitasking, but the companies developing MIDI apps didn't want to do that (or didn't know how). Unfortunately many Amiga developers did not understand the hardware they were working with.

That reveals a flaw in hardware or software. AmigaOS is not a realtime OS. But developers should have only needed to use the OS device driver. No restrictions should have been needed. If the serial was hardware interrupt driven it should have worked fine and override any tasks running.

Quote:

Nonsense. With a simple interface you can plug any standard ISA bus serial card into the A500. Several 16550 boards that plugged into the A1200's clock port were produced. The only reason we didn't see lot of buffered serial solutions for the A500 was lack of demand. Professional users bought a big box Amiga, for which many cards were produced.

A simple interface? Was this simple interface embedded into a product with a case that plugged neatly into an outside port? On an A500?

The clock solution needs to break the seal. And it doesn't connect to and override the external port. If the back panel is used up beside the joy ports then it ends up as an external hack.

Quote:

I remember when the 16550 became desirable as MODEMs got faster. Most older PCs only had a standard 16450. Some had a socket to install a 16550, but the chips were hard to get (at least here in New Zealand). I had ideas of making a clockport 16550 board for my A1200, but never got around to it because my 28k8 MODEM didn't need it. Then I write a driver for the CNET PCMCIA Ethernet card and it became moot.

When 56K arrived it became an issue. Finding Amiga compatible cards for modems or Ethernet wasn't so clear cut. I ended up buying a Whippet for serial and I didn't need Ethernet until years later when I had broadband. Even before then, locating a compatible card became an issue. I had a few I picked up that either were card bus or had no driver.

Quote:

External MODEMs were expensive, so the industry was moving towards internal MODEMs (which had their own serial chip so the stock serial port buffering was irrelevant). A few were produced for the Amiga, but of course it was 'proprietary' compared to PC cards.

I wasn't into the BBS scene so bought an older 28.8K as my first modem IIRC. By the time I bought a new modem a few years later it wasn't that expensive for a basic one. But by that time it was 56K and EOL was coming.

Quote:

This is where Commodore missed an opportunity. Instead of making Bridgeboards they should have produced a simple card that connected the ISA bus slots to the Zorro II bus. Then customers could use cheap ISA bus cards in their Amiga. Had they done this the early days they might have expanded it to PCI so we could use high performance graphics cards etc.

I thought the point of the ISA slots was being able to plug cards into it?

Quote:

Hand scanners designed for the Amiga could be plugged into it - I sold one in my shop. But like PC hand scanners they were not very good. A flatbed scanner was much better, but expensive. I had a SCSI flatbed scanner on my A1200.

I got one of those. Not used it much. But what I wanted to do was plug in a common parallel flatbed scanner into my A1200, that I had acquired, but I didn't know what it needed to work.

Quote:

The only thing stopping parallel drives being plugged into the Amiga was that they were designed for PCs. I developed software to use a ZIP parallel drive on the Amiga. I have often thought of getting another ZIP drive since they often go cheap on auction sites, but I can't be bothered because we have better solutions now (those ZIP drives had a nasty habit of dying!).

Was that the joystick port solution?

They also expected ECP or EPP ports. On top of this is the daisy chaining but I don''t know if it needs hardware support. I've got a Imation SuperFloppy with IDE parallel interface I was testing for a friend but don't know how it connects. The only PC parallel port I have is in my A1/XE. I ended up compiling the Linux kernel with specific parallel IDE support. But the drive goes dead when plugged into parallel. Buy itself it turns on.

It only came with a stub cable so I needed a another parallel cable to even connect it. But I don't know if the original parallel drive cable has any extra logic or particular wiring. I've spent ages trying to find info and cannot find anything. Strange, there are Linux drivers for IDE parallel chip, but how do you connect it up?

My friend wanted to plug it into his A500 for fast easy transfer of data. Well, faster than normal floppy. But it didn't come with all cables. I looked up the protocol and it does support older transfer modes so should work without ECP or EPP needed. But without the dedicated cable needed to connect it's a moot point I guess. If it can't be used at all without a hack then no point really.

Quote:

Sure it's possible. Or just throw in a 765 controller like other machines did. But why do we need it? DD disks work fine.

Unless the 765 can handle Amiga MFM it won't be that easy.

We needed it so transfers run at full speeds, be that Amiga or PC, at full capacity and so the internal drive can be easily replaced. Even DD PC drives became rare aside from logic needed to connect them. It became hard to exchange data with other computers when HD was common and you had to use a DD or ask someone to use a DD disk, where all the data nay not fit. The Amiga was a machine ahead of its time, so by being made obsolete by a simple floppy drive, it really lost the plot.

Quote:

No, 'they' didn't. The drive manufacturer made the drives. But again, why bother?

The manufacturers modified the drives for Commodore? Extra expense then. For one thing it was done and if done for Commodore then the parts just increased in price. Another, obviously, is keeping up with current standards. The Amiga became inferior and a joke because of these seemingly trivial hardware advances.

Quote:

I had no problem getting DD disks for my shop throughout the 1990s. The only problem most Amiga users had was finding enough disks to copy their latest 'acquisitions' onto.

By the 90's the Amiga had already outgrew DD disks. KindWords on three disks was ridiculous. 9 disk software or games? Ludicrous!

Quote:

Towards the end the number of disks that some games came on was ridiculous. But this was mostly because the games were developed for the PC first. The PC version was uncompressed and installed on the hard drive for playing, but the Amiga version had to play directly from disk because most Amiga owners were too stingy to buy a hard drive. Therefore it couldn't be compressed as efficiently, and so took up many more disks than it could have.

I thought it was because Amiga gamers were pirates so any HDD installer would have given the game away?

Before the A1200 a HDD wasn't as common. It was an expensive add on even when the monitor was a TV screen. The A590 was for the "Advanced" user.

Quote:

This problem soon want away with CDs. But it was largely too late for the Amiga, as not enough Amiga owners had CDROM drives. The other problem was more fundamental - CDROMs were less attractive to users because at that time they were too hard to pirate. For most Amiga users the ability to get any game for the price of a blank disk or two was its main attraction.

Adding CDROM to an Amiga was expensive. And you couldn't save to it. The A570 was available for the A500 but the A1200 had no solution and third parties had to step in. I added a CDROM to my A1200 and in the process, the dealer set it up in a SCSI tower solution, with a Ferret card. It proved useful later on as could add SCSI drives inside which I did. Though at the time my internal Quantum was doing fine.

@cdimauro

Quote:

cdimauro wrote: But this does NOT mean that they were products for the same market segment. In fact, they were home computers whereas the 1000 was a personal computer.

That's not how I and many others perceived it. "Personal Computer" in those days meant a PC. The A1000 was obviously designed for games. It had a nice low profile case that acted as a monitor stand (same as the C128D, which was designed at the same time as the C128, but its release was deliberately delayed because it looked too much like the Amiga). It didn't have bus slots. It did have composite output to connect to a TV, like other home computers. It had joystick ports and an internal floppy drive like other home computers of the time ere getting (eg. my CPC664). It didn't have a hard drive. It had a printer port and serial port, like many other home computers had either built in or as an option. In short, the Amiga was seen as the ultimate home computer, with everything built in that others should have had, and smashing performance barriers that limited games on 8 bit platforms.

Quote:

You missed the Apple Macintosh here. Guess why...

The assertion was that ALL other platforms evolved more that the Amiga. I don't have to refute every one in your list to make that claim false.

Quote:

Apple also introduced the II GS: https://en.wikipedia.org/wiki/Apple_IIGS With an outstanding audio section...

Outstanding? I don't think so. The II GS was a flawed design and the marketplace treated it accordingly.

Quote:

So, you cherry-pick only what you like.

Again, I don't have to 'cherry pick' to refute your claim. Just one counter example is enough - I provided several.

Quote:

And, as I've said, there were also the PCs.

And, as I said, the PC was in a different class.

But even if your claim was accurate with respect to the PC, how is it relevant? The PC started out with very low specs - 8 bit bus, barely 16 bits inside the CPU, low speed, tiny base RAM, a cassette interface and ROM BASIC in the 'home' model. The graphics card only did non-interlace NTSC scan rates with 2 colors in hires, and 4 colors in lores out of two horrible choices from the garish 16 color palette. It was basically the 8088 equivalent of an Apple II, but without the charm.

So the PC had plenty of opportunity to 'evolve' before it even came up to OCS Amiga standard. Starting off with a higher spec means the Amiga didn't have to evolve so much. Therefore the criteria you are using for being 'better' is invalid.

Last edited by bhabbott on 14-Oct-2022 at 07:40 AM.

@Karlos

Quote:

Regarding HAM8 speed, I dunno, this is quite impressive to me: https://youtu.be/SwnDqj8pNe4

That is interesting. It does seem to run fast enough. Could do with better textures and filming it off an old curved screen isn't the best. Was it ever turned into a game?

So looks like it is relying on colour tricks. Displaying 2x width in 16 bit and 4x width in 24 bit if I read it right. Going through the c2p routines it almost looked like it was doing a HAM chunky to HAM planar conversation.

But taking another look it first scales it down. In the case of 24 bit 8/8/8 it appears to scale it to 18 bit 6/6/6. It takes 8x 32-bit pixels with 24 bit RGB, shifts it into 22 bits, then encodes these RGB values in until it has 32 pixels to write per plane. At 4x width, so using 1280 for 320 source buffer, that gives 3 pixels to fully change RGB and 1 pixel to solidly display it. It needs a better description or I didn't see it in there.

Last edited by Hypex on 14-Oct-2022 at 08:09 AM.

@Hypex

It's a current project as far as I can tell. The code is on GitHub. I tried it out in UAE and it was very fluid. I might record a section later

I'm not sure exactly how the HAM8 is used but it does use 4 superhires pixels per "RGB".

Nothing a few careful screenshots in an emulator can't reveal.

One idea I had in the distant past with superhires was to use a fixed 2 bits per gun palette for the 64 base palette, then by simple bit shifting of RGB choose the closest one of those for a the first pixel. Then modify the green, red and blue in that order for the next three. Specifically in that order because green contributes the most luminance to an RGB pixel, followed by red, then blue.

I never tried it but maybe it's a good time.

_________________
Doing stupid things for fun...

@Hypex

So many points to address...

Quote:

Hypex wrote: I also have a Cobra and Ferret. Ferret particularly annoying as I always had trouble with that stax connector.

Yes, that conductive rubber connector wasn't the best idea. I did have to clean mine a couple of times. If I still had it today I would put a better connector system on it.

Connectors often cause problems. Every now and then I would have to re-seat the 060 board in my A3000. PCs can often be fixed by simply pulling out the the cards and RAM modules and re-inserting them. VL bus cards were particularly finicky. The ZX-81 was famous for crashing due to 'RAM-pack wobble'. Even the Blizzard 1230-IV in my A1200 needs to be moved a bit to clean the contacts every few years.

Quote:

I have a CyberSCSI in my A4000 connected to my CyberStorm 060. So a CPU card with local SCSI. Even so, without Zorro limitations, I recall it still needed particular max transfer.

This was a legacy issue, with the latest OS3 versions you don't need it. Not a big deal once you know about it.

More concerning to me was the flaky operation of external SCSI devices on my A3000. Even with carefully applied termination and settings some devices just didn't want to play nice. Having an 060 probably made it worse. Also early DMA chip revisions were known to have problems. But such is life on the 'bleeding edge'. The PIO IDE interface was generally much more forgiving, as well as being easier to reproduce. Discerning Amiga fans might have been disappointed by it, but being able to add a compatible IDE interface to retro machines with a simple circuit makes up for it.

Quote:

Reducing CPU power wasn't needed on low power models? That doesn't make sense. The more CPU power resources drain the slower it will be.

What low power models?

One day I want to create a low power Amiga by replacing chips that draw more than they could, just for the hell of it. For example the A1200 has four 74F245 chips to buffer the PCMCIA port. If they were replaced with CMOS equivalents it could save over 250mA on the 5V line! However even with this inefficiency the A500 PSU powering my A1200 only gets warm after many hours of use. I have PC laptops that get much hotter.

Quote:

It is near impossible to use it for practical use on floppies alone so any real use would have required a HDD. I know. I had an A500 like that with no extra drives. KindWords came on three disks and was not designed for floppies, asking for each one in a row and only loading a few KBs before wanting the next one, 20 times in a row or something. I managed to squeeze it down to one floppy so I could use it, without binary compression, and it was a lot of work.

Squashing stuff down to fit on one disk was lots of fun. But most 'productivity' users bought another floppy drive to cut down on disk swapping. I have an A1010 drive (originally for the A1000) which I plug into the A1200 or A500 as needed. On the A1000 I used to have a 5.25" drive for use with PC floppies.

However if you only want to play games a single floppy drive is often enough. There are probably thousands of games that only need a single drive. What's important is that customers could purchase the base model and get an external drive later if they needed it. Many A1200 owners may already have had a drive that they used on an earlier Amiga. Furthermore the A1200 could have a hard drive added later, which would get cheaper and larger capacity the longer you waited.

Quote:

The original ATA spec is said to have included support for PIO modes 0 to 2 and even DMA modes 0 to 2. It may have been standardised in 1994 but Commodore still implemented ATA so they could support it with the information that had. And with PCMCIA Commodore had already implemented a socket before it was standardised. But it turned out to be more buggy.

You can't have it both ways. Should they have used only specs they could rely on or not? At least the official PCMCIA spec was imminent and Commodore had a draft document. DMA IDE wasn't going to happen anyway (too expensive) and higher PIO speeds might violate Amiga bus timing so safer to stick with the lower speed (which was still pretty fast for the hard drives of the time).

Amiga PCMCIA isn't so much 'buggy' as some PC cards require specific timing and/or are tuned to certain features of PC controllers. Some even have trouble on PCs that have a 'different' controller. This is understandable as the designers only had a PC to test their card on, and if it passed then it was shipped.

The A1200 did have a bug where it didn't rest the card on power up (cards should reset themselves, but some don't). This is caused by an undocumented change to the reset procedure that wasn't included in the ROMs. When I developed the CNET network card driver I didn't know about this, so I made hardware reset circuit. This should not be necessary today as resetting can be done with software.

Quote:

Quote: Coincidentally 1991 was the year I started my business selling Amigas and PCs. I remember having a problem with new IDE hard drives and certain motherboards, which according to my supplier was was caused by logic voltage level mismatch. I don't know if this was true, but I do know that we had significant compatibility issues with IDE back in those days. \ Did you see any problems with Amigas? The A4000 could take a full desktop drive. So be interesting if same drives has any conflict there.

I didn't work with the A4000 much, but I put lots of hard drives in A600s and A1200s. Never had one that didn't work properly.

Quote:

The only major criticism levelled at IDE was using a 44 pin laptop IDE port which meant smaller and more expensive drives wee needed. However they weren't full desktop computers and space was tight. We did see A1200 hacks for full drives but I thought that was a bit over the top.[quote] The main reason for needing a 2.5" drive was simply lack of space. However they were also lower power and spun up faster, and significantly more robust. I one killed a brand new 3.5" IDE drive in a PC that I was installing Windows onto in the shop. Outside the shop a road worker was using a compacter, which caused vibration that crashed the heeds on the drive! [quote]It's detailed in Mapping the Amiga. But it's slightly unclear as it mentions CIA B SDR connected to internal lines but not used by OS. How did you attach to the serial port?

Through the parallel port. I used a standard CMOS dual 64 bit shift register IC to buffer the signals. Worked perfectly!

Quote:

The C64 had serial, as used with the 1541, but it wasn't the best example around.

IEC was bitbanged. There was also a serial port cartridge that plugged into the user port. This was also bitbanged. Some 3rd party cards used a proper UART chip such as the R6551, which only a had a 1 byte receive buffer like the Amiga.

Quote:

It can't operate at the exact 31,250 baud.

It's less than 1% out, plenty accurate enough (need 5% if the other end is perfect, 2.5% at both ends).

Quote:

That reveals a flaw in hardware or software. AmigaOS is not a realtime OS. But developers should have only needed to use the OS device driver. No restrictions should have been needed. If the serial was hardware interrupt driven it should have worked fine and override any tasks running.

Technically you are correct, Amiga OS is not a 'realtime' OS. Game designers understood this, but some application designers didn't. If you want real-time response you either have to turn off multitasking, or have a good appreciation of what could be compromising response time and deal with it.

The serial device driver is on disk for a reason. You can replace it with one that works better in your application, or use a completely custom driver, or have nothing and access the hardware directly. Only problem is if you want full control then you miss out on all those lovely things the OS gives you. MIDI app developers probably didn't want to do that, so they were stuck with trying to do 'realtime' with the OS adding unkown latency.

Quote:

A simple interface? Was this simple interface embedded into a product with a case that plugged neatly into an outside port? On an A500?

There were products that used this technique. eg. the 'ALF' hard drive which I copied. Dead simple interface. I don't know of anyone doing this with a PC serial port card, but it shouldn't be hard. Main reason you wouldn't do it is that you probably couldn't do much with high speed serial on a stock A500.

Quote:

The clock solution needs to break the seal. And it doesn't connect to and override the external port. If the back panel is used up beside the joy ports then it ends up as an external hack.

So does adding a realtime clock. Many A1200 users did far more 'hacking' than that. Everything from jamming a 3.5" hard drive in it to taking the entire motherboard and putting it in a PC case. Today some accelerator cards come with a clock port or two on board.

Many of us enjoy doing all kinds of hacks to our machines, and have been doing it ever since we got our first home computer. If you didn't want to do it yourself then a friendly dealer or other user was often at hand to do it for you. I was involved in several user groups for various machines, where we did stuff like this for each other. This was an important part of the hobby for us.

But hey, some people just want to do a job and not have to muck around 'hacking' their machines. That's that what 'big box' Amigas (or PCs) and computer technicians are for. They have the skills to install a serial card so you don't have to! I made a living out of it.

@Hypex
Quote:

Hypex wrote: I thought the point of the ISA slots was being able to plug cards into it?

The ISA slots were supposed be for a 'Bridgeboard' (PC on a card). In most machines they lay idle because we had no use for them. However they were useful when I put a PC hard drive in my friend's A2000. I made a card that bridged the buses to access the PC ST506 hard drive controller. Worked very well.

Quote:

Was that the joystick port solution?

Yes. It was only needed because I wanted to use the minimum possible number of logic chips (none).

Quote:

Unless the 765 can handle Amiga MFM it won't be that easy.

Not that hard. A simple MUX to switch data and control lines between Amiga and 765 is all you need.

Quote:

The manufacturers modified the drives for Commodore? Extra expense then. For one thing it was done and if done for Commodore then the parts just increased in price. Another, obviously, is keeping up with current standards. The Amiga became inferior and a joke because of these seemingly trivial hardware advances.

No, it didn't. You see, the Amiga was always inferior and a joke because it didn't comply with 'current standards'. You can blame Jay Miner for that. But at least, unlike the Apple Macintosh, it could read and write 'standard' PC 360k 5.25" floppies with Commodore's external 5.25" drive (I had one).

Quote:

By the 90's the Amiga had already outgrew DD disks. KindWords on three disks was ridiculous.

Kindwords 2:-
program disk - 819k used
dictionary disk: - 724k used
Superfonts disks:- 542k used

The program and dictionary disks alone total 1543k, so you would need two 1.44MB disks just for them. At most you save one disk by using HD PC format. But the dictionary and fonts disk are not needed all the time, and the Amiga will ask for them if necessary.

Quote:

9 disk software or games? Ludicrous!

Yes, I did say this.

But... Windows NT came on 22 disks! Beyond ludicrous!!! And one version of Windows 95 came on 28 disks!!! Ludicrous squared!!?!!! Can we top that? Sure. Microsoft Office 97 Professional was provided on a total of 55 diskettes !!!@#$%&*!!! (words fail me).

Quote:

I thought it was because Amiga gamers were pirates so any HDD installer would have given the game away?

Aha! You've caught on to why they used so many disks. You see, there comes a point when the scummy pirates can't justify using up that many disks on a game they probably won't play much anyway. :)

Quote:

Before the A1200 a HDD wasn't as common. It was an expensive add on even when the monitor was a TV screen. The A590 was for the "Advanced" user.

True, and not so necessary either, when games only came on one or two disks. But many users did get an external disk drive.

Quote:

Adding CDROM to an Amiga was expensive.

Adding a CDROM to any computer was expensive. But PC owners did because they wanted the software that came on CDs.

Despite the expense, a lot of Amiga owners got one. So much so that Amiga magazines had cover CDs. Before that Amiga PD collections were distributed on CD, including Aminet. This must have cost them a bit to produce, so they must have been reasonably confident of getting enough sales to justify it.

Quote:

And you couldn't save to it.

Unless you had a CD Writer. I did. It cost NZ$750 ($350 of which was the freight from the US).

Quote:

The A570 was available for the A500 but the A1200 had no solution and third parties had to step in.

Commodore was working on a solution called the CD1200, which would have included Fast RAM and an optional 030 CPU. Unfortunately they went bankrupt before they could get it out.

@matthey: finally I had time to read your link and think about your post.

Quote:

matthey wrote: cdimauro Quote: OK, so it proves that it wasn't a paper project. It's strong evidence anyway. It's interesting that the Amiga Corporation developers used the boing ball on the Ranger prototype instead of the checkmark used on the Amiga 1000.

Very strange decision, indeed.
Quote:

In pursuit of ASIC custom chips, I found none for Ranger, not even planned. https://obligement-free-fr.translate.goog/articles/chipsetamiga.php?_x_tr_sch=http&_x_tr_sl=auto&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp There were at least planned and named custom chips for AA+ and Hombre although Ranger plans may have been hidden away or kept separate from normal CBM activities and documentation. AA+ docs go back to May 1992 which shows how bad of financial shape CBM was in already that they never produced custom chips for AA+ or Hombre (there were custom chips produced for AAA). AA+ development was likely straightforward as much of the logic from AAA development could be backported to AA+.

Nevertheless, it was just another paper project. Yes, Commodore already wasn't doing well.
Quote:

Transistor estimates for AA+ seem suspect to me though. ECS - 3 chips 60,000 transistors AGA - 3 chips ? transistors (Lisa 80,000 transistors, Alice 80,000 transistors) AA+ - 2 chips 200,000 transistors AAA 32 bit - 4 chips 750,000 transistors AAA 64 bit - 6 chips 1,000,000 transistors The link above has a 512kiB Agnus using 21.000 transistors while the 2MiB AGA upgrade Alice uses 80,000 transistors. The AA+ Ariel chip which combines the Alice and upgraded Paula (likely taken from AAA), uses only 100,000 transistors. This suggests that a 16 bit Paula with 8 voices used at most 20,000 transistors. While the memory access logic of Ariel was likely similar to AGA Alice, this suggests the address range upgrade from 2MiB to 4MiB had a very cheap transistor cost. This would also suggest that the Ranger upgrade from 512kiB to 2MiB had a cheap transistor cost. If looks like Paula (10,000-15,000 transistors?) could have been combined with Agnus/Alice much earlier or better, upgraded and combined seeing how 16 bit 8 voice audio likely used fewer than 20,000 transistors which is cheap already in the late '80s and cheaper yet in the '90s. With the Paula transistor estimate, we can estimate AGA used ~175,000 transistors. AA+ using 200,000 transistors likely could have been done in 1992 if CBM had it ready and may have cost less being 2 chips instead of 1.

I agree. I think that the transistors count is accurate, as well as the reconstruction.

AA+ could have been done in 1992, sure, but according to the obligement link it was an AGA upgrade. So, not an AAA "cut-down" version. In fact, it was registers-compatible with ECS and AGA. This is also why the transistor count estimate is realistic.

However it was another paper project...
Quote:

AA+ using ~200,000 transistors would have been low cost and CBM could have surely moved to a better process and made it one chip,

Indeed.
Quote:

even with a 68k CPU included if they wanted a single chip Amiga SoC. 68020 200,000 + 200,000 = 400,000 transistors 68030 273,000 + 200,000 = 473,000 68040 1,200,000 + 200,000 = 1,400,000 68060 2,500,000 + 200,000 = 2,700,000 An Amiga SoC with at least 68030 and AA+ specs should have been feasible in 1992 (AIM Agreement was in 1991 so 68k licensing should have been reasonable).

While on paper it was possible, Motorola wasn't ready to sell a 68k softcore to partners to create a SoC. I don't recall anything about that in that period of time.
Quote:

You would think Dave Haynie would know about Ranger considering he was designing AAA. At the very least, part of the Ranger design could have been reused like for VRAM access.

Actually it looks like that the discussions (not even papers) on AAA were just started at the same time that Ranger was abandoned.

Using part of Ranger could have been possible, since AAA used it.
Quote:

Jay Miner said he finished the Ranger design and gave it to CBM before he left. Maybe CBM never authorized Ranger development and Jay started developing anyway

Unlikely, because Ranger required funds.
Quote:

However, it sounds like Los Gatos and CBM Germany were in a competition for the high end Amiga so I don't know. The Ranger specs appear to have been deeply buried at CBM and at least some of the prototypes appear to have stayed with Los Gatos developers (Dale Luck). There was a lot of animosity with the Los Gatos developers. Remember the "We made Amiga, They ####ed it up" which really pissed off CBM managment? We made Amiga, they ####ed it up (ENGLISH) https://www.youtube.com/watch?v=omMOuyTLmyg

Sure. Who does NOT know it?
Quote:

Ranger would have been better than VGA in 1988-1989 and would have stood up against SVGA until the next generation of Amiga chipset was available. The chipset performance was like AGA but not as colorful. Still, 128 colors looks almost as good as 256. It would have been better if 8 bit plane EHB with 128 color registers and HAM8 could have been included (shouldn't have needed major changes so maybe could have been added). Higher resolutions would have allowed better SVGA compatibility and monitor use while 640x400/480 would likely have been usable for some games. I think fewer customers would have been lost to the competition compared to ECS in the late '80s and early '90s.

Unfortunately Ranger was too late: on 1989 was still only a prototype. Whereas on this year there were already several vendors selling SVGA video cards with much better specs (and sound cards as well: the PC market literally exploded).
Quote:

There was better and more expensive hardware but this would have moved the base Amiga up quite a bit like AGA did too late. It may have given CBM time to see that customers wanted upgraded Amigas which they saw too late with AGA while they were still producing 68000+ECS Amigas (a major factor in their downfall). Ranger would have likely made the 68020 standard as well. Ranger was designed to work with the 68010 also but by 1988-1989 the more powerful 68020 had dropped considerably in price and the 68010 was clearly on the way out.

Indeed. As for the A1200, a cost-reduced 68020 should have been the minimum base for a low-entry Amiga. On 1989 at most (which is already 4 years from the A1000 introduction).

@bhabbott

Quote:

bhabbott wrote: @cdimauro Quote: cdimauro wrote: But this does NOT mean that they were products for the same market segment. In fact, they were home computers whereas the 1000 was a personal computer. That's not how I and many others perceived it. "Personal Computer" in those days meant a PC. The A1000 was obviously designed for games. It had a nice low profile case that acted as a monitor stand (same as the C128D, which was designed at the same time as the C128, but its release was deliberately delayed because it looked too much like the Amiga). It didn't have bus slots. It did have composite output to connect to a TV, like other home computers. It had joystick ports and an internal floppy drive like other home computers of the time ere getting (eg. my CPC664). It didn't have a hard drive. It had a printer port and serial port, like many other home computers had either built in or as an option. In short, the Amiga was seen as the ultimate home computer, with everything built in that others should have had, and smashing performance barriers that limited games on 8 bit platforms.

What YOU and SOME other people could have thought about it is irrelevant. What is relevant are the FACTs and what was given to history.

https://en.wikipedia.org/wiki/Amiga
Amiga is a family of personal computers introduced by Commodore in 1985. [...]compared to previous 8-bit systems. This includes the Atari ST—released earlier the same year—as well as the Macintosh and Acorn Archimedes.

https://en.wikipedia.org/wiki/Commodore_128
The Commodore 128, also known as the C128, C-128, C= 128,[n 1] is the last 8-bit home computer that was commercially released by Commodore Business Machines (CBM).

https://en.wikipedia.org/wiki/Amstrad_CPC
The Amstrad CPC (short for Colour Personal Computer) is a series of 8-bit home computers produced by Amstrad between 1984 and 1990. It was designed to compete in the mid-1980s home computer market dominated by the Commodore 64 and the Sinclair ZX Spectrum

If you don't like you're free to change the respective entries.
Quote:

Quote: You missed the Apple Macintosh here. Guess why... The assertion was that ALL other platforms evolved more that the Amiga. I don't have to refute every one in your list to make that claim false.

True. And you didn't it.
Quote:

Quote: Apple also introduced the II GS: https://en.wikipedia.org/wiki/Apple_IIGS With an outstanding audio section... Outstanding? I don't think so.

Sure, I "believe" you: https://en.wikipedia.org/wiki/Apple_IIGS#Audio
The Apple IIGS's sound is provided by an Ensoniq 5503 Digital Oscillator Chip (DOC) wavetable synthesis chip designed by Bob Yannes, creator of the SID synthesizer chip used in the Commodore 64. The ES5503 DOC is the same chip used in Ensoniq Mirage and Ensoniq ESQ-1 professional-grade synthesizers. The chip has 32 oscillators, which allows for a maximum of 32 voices (with limited capabilities when all used independently), though Apple's firmware pairs them for 16 voices, to produce fuller and more flexible sound, as do most of the standard tools of the operating system (the Apple MIDISynth toolset goes even a step further for richer sound, grouping four oscillators per voice, for a limit of seven-voice audio). The IIGS is often referred to as a 15-voice system, because one voice, or "sound generator" consisting of two oscillators, is always reserved as a dedicated clock for the sound chip's timing interrupt generator. Software that does not use the system firmware, or uses custom-programmed tools (certain games, demos and music software), can access the chip directly and take advantage of all 32 voices.

The fox and the grape...
Quote:

The II GS was a flawed design and the marketplace treated it accordingly.

This is a Red Herring = logical fallacy.

In fact, that's ANOTHER question.
Quote:

Quote: So, you cherry-pick only what you like. Again, I don't have to 'cherry pick' to refute your claim. Just one counter example is enough - I provided several.

And I've already replied. Guess what: you silently skipped everything...
Quote:

Quote: And, as I've said, there were also the PCs. And, as I said, the PC was in a different class.

Oh, yes: let's make exceptions and pulling out inconvenient things...
Quote:

But even if your claim was accurate with respect to the PC, how is it relevant? The PC started out with very low specs - 8 bit bus, barely 16 bits inside the CPU, low speed, tiny base RAM, a cassette interface and ROM BASIC in the 'home' model. The graphics card only did non-interlace NTSC scan rates with 2 colors in hires, and 4 colors in lores out of two horrible choices from the garish 16 color palette. It was basically the 8088 equivalent of an Apple II, but without the charm.

Sure, and it was in... 1981: when the PC first appeared!
Quote:

So the PC had plenty of opportunity to 'evolve' before it even came up to OCS Amiga standard.

And it did, in fact, at all levels: more powerful processors, graphics cards, sound cards, and other peripherals.

It's just a matter to sit and seek their evolution.
Quote:

Starting off with a higher spec means the Amiga didn't have to evolve so much.

Sure. Whereas the competitors did it and outclassed the Amiga: but that was nice, right? You liked it...
Quote:

Therefore the criteria you are using for being 'better' is invalid.

Well, the point was that Amiga didn't evolved whereas competitors did it. So, I'm still right.

@Hypex

Quote:

Hypex wrote: @Karlos Quote: Regarding HAM8 speed, I dunno, this is quite impressive to me: https://youtu.be/SwnDqj8pNe4 That is interesting. It does seem to run fast enough. Could do with better textures and filming it off an old curved screen isn't the best. Was it ever turned into a game? So looks like it is relying on colour tricks. Displaying 2x width in 16 bit and 4x width in 24 bit if I read it right. Going through the c2p routines it almost looked like it was doing a HAM chunky to HAM planar conversation. But taking another look it first scales it down. In the case of 24 bit 8/8/8 it appears to scale it to 18 bit 6/6/6. It takes 8x 32-bit pixels with 24 bit RGB, shifts it into 22 bits, then encodes these RGB values in until it has 32 pixels to write per plane. At 4x width, so using 1280 for 320 source buffer, that gives 3 pixels to fully change RGB and 1 pixel to solidly display it. It needs a better description or I didn't see it in there.

Indeed. But the main problem is that SuperHires using 8 bitplanes is utterly slow: it leaves only a few memory slots free for doing something other than displaying the screen.

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: Which topic? I was talking about what developers should care of: the most common / sold Amiga platforms. This is the only thing which matters. This topic title: How good or bad was the AGA chipset in 1992/1993.

Yes, and the market / audience was relevant, as Carmack proved, to evaluate whats needed and determine how to evolve the hardware.

For example, IBM released the PGA video card on 1984: but it was A LITTLE BIT expensive and tailored to a specific market. That's the reason why people don't remember it, but clearly recall the EGA which was introduced on the same year.
Quote:

Quote: Again: irrelevant. Your argument is irrelevant.

Then prove it! A double negation does NOT prove that the original statement is correct.
Quote:

Quote: Are you kidding? ACA had FASTER processors AND memories! So this is OBVIOUS!!! http://amiga.resource.cx/perf/aibb.pl?amiga=1200&testgroup=int&order=mem&ref=a1200 ACA 1220 (020/16)'s MemTest has 2.53 score. This SKU has 68020 at 16 Mhz which is a minor clock speed increase over A1200's 68EC020 @ 14 Mhz. A1200's stock 68020 is 16 Mhz rated and Commodore reduced the clock speed to 14 Mhz due to the existing 28 Mhz crystal clock being divided by 2.

But YOU reported the numbers for OTHERS, MORE POWERFUL, ACA cards!

Here's it YOUR statement:
https://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=44362&forum=2&start=220&viewmode=flat&order=0#855874
ACA 1220 with 68020 @ 20 Mhz with ACA's Fast RAM design scored 3.04
ACA 1220 with 68020 @ 33 Mhz with ACA's Fast RAM design scored 4.22

Care to tell me where is the 16Mhz reported there? It was NOT there!

So, I've replied to what YOU reported: about the 20 and 33Mhz versions! OBVIOUSLY...
Quote:

My A1200's stock 68EC020 with 16 Mhz marking. [BIG IMAGE REMOVED]

Who cares? You talked about OTHER cards!

BTW, you're still not able to use a decent image host service and flood the threads with BIG images which break the layout and make the posts a hell to read.

When do you plan to learn something that even an average Joe is able to do?

@Hammer

Quote:

Hammer wrote: @cdimauro Quote: Gunnar already replied and I agree with him. Now I ask you: do you think that we, developers, didn't used the slow-mem (and fast-mem)? Yes, we did it! How do you think that Fighin' Spirit (and USA Racing) provided so much graphics and sound effects? The 512kB where absolutely NOT enough for keeping all that stuff! But the problem is that such ram was NOT accessible by the chipset. So, we were forced to use the CPU to transfer memory from slow to chip mem (because the Amiga has no DMA for doing copy operations, like PCs), wasting A LOT of CPU time. That's the point. Having had MORE chip ram would have allowed us to have better effects or more fluid games. FYI, You can have some kind of 1 MB Chip RAM even with 0.5M Chip RAM with 0.5MB Slow RAM configuration as long as Agnus is an ECS model. 1M Agnus ECS with 0.5M chip+0.5M slow aliases upper 1M chip ram area to slow ram area e.g. copper pointer set to 0x090000 sees memory at 0xC10000.

This is a trick / non-standard thing. Hence, something which shouldn't be done.

When I collect the information about the available non-chip mem I scan the o.s. memory list and I only rely on the information reported there. So, if the o.s. tells me that the extra memory is mapped on 0xC00000 then I use it at this address.
Quote:

This topic is about AGA Amigas, not OCS/ECS.

Exactly and the same applies. You haven't understood the discussion: read it again.
Quote:

Assume Commodore didn't have manufacturing F_ukup with HK to PH move.

?!?
Quote:

Turrican AGA requires Fast RAM. http://eab.abime.net/showthread.php?t=106735

It's an amateur project which requires not only Fast mem but also a monster processor.
Quote:

As for Fighin' Spirit, in terms of a graphics tech demo game, I prefer Elf Maina. Elf Maina's control system needs to change.

I don't question about personal tastes, but in terms of graphic Fightin' Spirit moves much more: bigger screens, much bigger characters, way more animations (just take a look at how characters "move" back and forth on both games, for example. But there's more, of course), special moves ("energy bubbles") and more much more colorful.
Quote:

Quote: Also, but it then becomes more expensive... From USA's Amiga World Magazine (November 1993), page 58 of 100, Price listed in USD in November 1993 A1200/020, 2MB, price $379 USD. UK and dollar zone countries may have different perspectives when compared to mainland Europe.

Yes, but the point was about adding more which would have made it more costly.
Quote:

My Dad has no problems spending $1500 AUD for a computer. Both my Dad and Mum have full-time work.

Classy (cit. MEGA_RJ_MICAL).

@bhabbott

Quote:

bhabbott wrote: @Hammer Quote: Hammer wrote: @bhabbott That's BS in the context of Amiga's existence. The timeline, 1. 80286 with integrated MMU was available in February 1982. 'Available', but the PC/AT didn't get it until 1984, running at 6MHz with 1 wait state (equivalent to 4MHz).

Do you understand that besides IBM there were several OTHER vendors which produced PC-compatibile systems? And with faster processors.

You also completely ignore even the x86-compatible processors, like the ones from NEC, which equipped those PC clones.
Quote:

If the Amiga had used an 80286 it would have had to add wait states to skip the display DMA slots, making it no faster than the 68000

Why?
Quote:

(actually slower since the 80286 has fewer registers and they are only 16 bit).

Which is plainly false: it was faster, even with less registers and being 16-bit. Take a look at the benchmarks of the time.
Quote:

Quote: Both 68000 and 80286 have 16-bit ALUs but 68000 has a 32-bit instruction set while 80286 has an integrated MMU. An MMU that only worked in protected mode, which wasn't compatible with DOS. Fail!

In fact, it was used on other o.ses.
Quote:

(but then Intel never intended it to be used in the PC).

False: it was usable even for DOS applications. In fact, it was possible to enter the 286 protected mode and go back to real mode using a trick.
Quote:

Oh which one to choose? An unnecessary MMU that doesn't work with your OS, or lots of lovely 32 bit registers? I know which one I would rather have.

Right and I agree: the latter was a much better choice for the consumer market. No doubt.
Quote:

Quote: 3. 32-bit 80386 was released in October 1985. Even more off topic.

It's not, considering the platforms evolution, as we discussed.

BTW, the first 386 PC arrived on 1986 by Compaq. Care to tell me when the first 68020+ Amiga (WITHOUR accelerator card) arrived?
Quote:

Quote: Look in the mirror, hypocrite. You made your 68000 vs 8088 arguments outside of Amiga's 1982 existence. No, I didn't. When the A1000 was introduced the vast majority of PCs sold were 8088 based.

Sure: in your fervid imagination...
Quote:

When the Amiga's design was started no 80286 based machine existed.

But the 80286 existed.
Quote:

However the machine they used to develop the Amiga was 68000 based, so it made sense to use the same CPU.

Absolutely. But would have been much better to use the 68010 for the final product: it would have avoided some issues (e.g.: GetCC).
Quote:

Quote: Reminder 8088 is a cost reduced 8086. 68008 is a cost reduced 68000. Difference is the PC did use the 'cost reduced' 8088, while the Amiga did not use the 'cost reduced' 68008 (the QL did though, which Clive Sinclair later admitted was a mistake). What's interesting about the PC was the 8088 became the 'standard' CPU, and the more powerful 8086 was largely forgotten.

See above: only in your mind.

@bhabbott

Quote:

bhabbott wrote: @Hammer Quote: Hammer wrote: [quote]Motorola's 68K CPU technology leadership stalled about the mid-1980s and Motorola was focusing on RISC based 88000 development and which was released in 1988. Yes. Motorola struggled to keep up with Intel both in CPU design and chip manufacturing. Part of the problem may be lower demand, another was different strategies. Intel was fully committed to the PC and stood to win big if they could beat off the competition.

That's plainly false. Intel did OTHER projects as well. A notable example: https://en.wikipedia.org/wiki/Intel_i860
Quote:

They concentrated on achieving the highest possible performance even if it meant overclocking the CPU by 4x the bus speed or clamping a huge heatsink and fan onto it - because the PC world demanded it.

Again, on your imagination...
Quote:

Motorola was more into the high performance embedded market where lower power was important, with desktop CPUs almost being a sideline.

That was in last years...
Quote:

Quote: Too bad Motorola didn't glue some 1.0 IPC DSP extensions into 68K I don't think it was too bad. The 68060 matched Pentium performance well beyond the Amiga's commercial lifespan.

According to Motorola (so, not me), 68060 matched Pentium performance for integer workloads. But never published information about FP workloads: guess why...
Quote:

Had they continued to simply speed it up we would have been quite satisfied.

But not competitive with the PC market, since Motorola was late with its processors and/or product processes.
Quote:

Quote: Acorn's RISC-based ARM was the solution after Commodore's crap 65 series CPU R&D roadmap. ARM v2 already has a strong IPC when compared to 68000 and 68020. Acorn's RISC PC was a failure. They were reduced into becoming just an IP vendor for the ARM architecture. Their 'roadmap' was not a good one for Commodore to emulate.

I agree, because... Commodore never had a roadmap for producing... PROCESSORS.

Whereas its chipset was overall inferior of Acorn's one introduced with the Archimedes.
Quote:

Quote: 286's MMU benefited MS Xenix and IBM/MS OS/2 development. Both of which had minimal market penetration. The 286's MMU was widely recognized in the PC world as being flawed, and they were desperate for something better.

Again, you don't know of what you talked about.

The 80286 was the first (consumer) processor to get military-grade thanks to its security features.
Quote:

That's why Intel developed the 386SX, to replace the 286 in low cost 16 bit motherboards.

That was ANOTHER reason.

In fact, the 80386, SX version included, still had the Protected Mode introduced with the 80286 AND improved it. In fact, the 386 added the virtual 8086 mode AND it had a 48-bit virtual address space in protected mode.

Why don't you talk only of things that you really know?
Quote:

Quote: Windows/386 introduced a protected mode kernel that allowed several MS-DOS programs to run in parallel in "virtual 8086" CPU mode. But only useful for well-behaved DOS programs, not games or apps that needed all the resources of the machine.

Why not? Many DOS applications were perfectly suitable for running on an 8086 virtual machine.
Quote:

Quote: Windows NT 386 needs i386's MMU and PC world was building a large install based on i386 MMU capability for a longer time duration. [snip] PC world's long build-up with TPM 2.0 install base tactics was used before pulling Windows 11 TPM trigger. Blah blah PC blah blah. I'm sick of hearing about the PC. This site isn't called Amiga World for nothing!

Are you invoking censorship because you don't like what people report to reply at your false statements?

@cdimauro

Quote:

cdimauro wrote: That's plainly false. Intel did OTHER projects as well. A notable example: https://en.wikipedia.org/wiki/Intel_i860

Did I say they didn't? No.

But...
Quote:

The i860 never achieved commercial success and the project was terminated in the mid-1990s.

Quote:

Again, on your imagination...

You refute Intel's 486 overdrive chips and the cooling required to get higher Pentium speeds? Later revision 060's needed less cooling, while later Pentiums needed more. That is a fact.

Quote:

According to Motorola (so, not me), 68060 matched Pentium performance for integer workloads. But never published information about FP workloads: guess why...

Irrelevant. What matters is real-world performance.

Lightwave 3D rendering test, Pentium 90 MHz vs 68060 50 MHz
Quote:

Test performed in CeBIT 1996 Pentium 90 MHz: 1:30 68060 50 MHz: 0:56 Notes: Amiga Lightwave rendered the test picture faster than the P90 even if it is not really optimized for the 68060 yet. ;)

Quote:

I agree, because... Commodore never had a roadmap for producing... PROCESSORS.

And good that they didn't. Imagine Commodore trying to produce CPUs rivaling Motorola and Intel. That would be stupid.

Quote:

Quote: That's why Intel developed the 386SX, to replace the 286 in low cost 16 bit motherboards. That was ANOTHER reason. In fact, the 80386, SX version included, still had the Protected Mode introduced with the 80286 AND improved it. In fact, the 386 added the virtual 8086 mode AND it had a 48-bit virtual address space in protected mode.

It's pointless discussing anything with someone who keeps (deliberately?) misinterpreting what I say. I was talking about the 386SX specifically, not the 386DX. That's why I wrote '386SX'.

Quote:

Many DOS applications were perfectly suitable for running on an 8086 virtual machine.

Many, but not all. Specifically, not DOS programs that accessed hardware directly. That includes games, and some apps that did direct hardware access (eg. to program a device through the printer port - a task that I often used the PC for).

But this is getting way off topic. How well Windows handled DOS programs on a 386 is not relevant to how good the AGA chipset was.

I'll say one thing though. In 1993 I was running DOS programs on my A3000 with full multitasking, the virtual DOS machine running in its own screen 'behind' the Workbench which could be pulled down to reveal it. Try that on a PC! With this I was able to develop DOS programs more conveniently than on a real PC. Recently I ran the DOS version of Tomb Raider on my A1200 in 256 colors, which looked wonderful on my TV. AGA made it possible.

Quote:

Are you invoking censorship because you don't like what people report to reply at your false statements?

My false statements? No, I'm not stopping you from saying whatever you like. The off-topic PC crap and Amiga bashing is starting to get boring though. So hey, keep it up and maybe you will censor me. That's your goal, right?

@Hammer

Quote:

My Dad has no problems spending $1500 AUD for a computer. Both my Dad and Mum have full-time work.

...

Quote:

The context was Amiga OCS into the ECS era. I already stated 1996 Pentium 150 was my transition period when I brought my own computer.

Tense! It's a statement about the past, not a hypnotherapy regression session into it.

At the risk of being called a grammar Nazi, these things matter, even something as simple as capitalisation. After all, that's the difference between helping your uncle Jack off a horse and helping your uncle jack off a horse

Last edited by Karlos on 15-Oct-2022 at 08:42 AM.
Last edited by Karlos on 15-Oct-2022 at 08:36 AM.
Last edited by Karlos on 15-Oct-2022 at 08:35 AM.

_________________
Doing stupid things for fun...

Since comparing the Amiga to a PC seems to be necessary part of deciding how good AGA was, I'll drop in another datapoint.

According to thandor.net, an AMD 386 DX40 got 8.24fps in the Doom timedemo 3 test. I tested it on my A1200 with 50MHz 030 and got 9.9fps, 1.2 x times faster. This exactly matches the difference in CPU clock speeds.

In 1992 386DX systems were considered 'mid-range'. They usually came with some unspecified ISA bus VGA card. Considering the AGA chipset's clear advantage for 2D games, the fact that with an equivalent CPU the A1200 could match a 386DX40 shows that AGA was not inferior to a 'typical' PC of the day, and in some ways was superior.