Click Here
home features news forums classifieds faqs links search
6052 members 
Amiga Q&A /  Free for All /  Emulation /  Gaming / (Latest Posts)
Login

Nickname

Password

Lost Password?

Don't have an account yet?
Register now!

Support Amigaworld.net
Your support is needed and is appreciated as Amigaworld.net is primarily dependent upon the support of its users.
Donate

Menu
Main sections
Home
Features
News
Forums
Classifieds
Links
Downloads
Extras
OS4 Zone
IRC Network
AmigaWorld Radio
Newsfeed
Top Members
Amiga Dealers
Information
About Us
FAQs
Advertise
Polls
Terms of Service
Search

IRC Channel
Server: irc.amigaworld.net
Ports: 1024,5555, 6665-6669
SSL port: 6697
Channel: #Amigaworld
Channel Policy and Guidelines

Who's Online
40 crawler(s) on-line.
 30 guest(s) on-line.
 0 member(s) on-line.



You are an anonymous user.
Register Now!
 zipper:  1 hr 16 mins ago
 DiscreetFX:  1 hr 30 mins ago
 Kronos:  1 hr 40 mins ago
 clint:  1 hr 40 mins ago
 pixie:  1 hr 43 mins ago
 Karlos:  1 hr 49 mins ago
 Futaura:  2 hrs 20 mins ago
 ppcamiga1:  2 hrs 59 mins ago
 agami:  3 hrs 38 mins ago
 Maijestro:  3 hrs 42 mins ago

/  Forum Index
   /  Amiga General Chat
      /  How good or bad was the AGA chipset in 1992/1993.
Register To Post

Goto page ( Previous Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 Next Page )
Poll : How good or bad was the AGA chipset in 1992/1993.
10p Excellent (Best at 2D/3D, colors, and resolution, frame rate etc.)
5p Good / better than most computer.
0p Barely hanging in there.
-5p Below average / slow but usable
-10p useless / horrible
 
PosterThread
bhabbott 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 21:51:58
#201 ]
Regular Member
Joined: 6-Jun-2018
Posts: 251
From: Aotearoa

@Karlos

Quote:

Karlos wrote:

One could make the claim that if additional Fast Ram were present by default on AGA systems that larger and more complex 2D games would be possible as larger, more complex level data could reside in fast memory as well as the code and all of the chip ram could be for media assets. Games fit their non graphical data and code into tens of KB because that was their budget.

More is always better for sure, except that it costs more. The A4000 came with FastRAM, so that is the AGA machine people could have bought if they thought it was needed. But the Amiga 'standard' for games was a base model without FastRAM, so it was a well established and accepted design goal.

The trend of code being only a small fraction of total memory usage is largely due to inherent design of the game, not just a 'budget' matter. Most 2D action games don't need a lot of code, but they do need as much space for graphics data as possible. Many games utilized disk space to make up for limited RAM. The most famous is probably The Faery Tale Adventure, which at the time had the largest map of any game - 32000 screens - loaded from disk in real time as the player moved through the game. Yet all the code and active data fitted in 512k, so the game worked fine on a base A500 or 512k A1000. The Faery Tale Adventure was written in C, so the code is probably a bit bloated compared to asm. But that doesn't matter as much as making efficient use of data space (which usually is mostly graphics).

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 22:12:17
#202 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@bhabbott

Quote:
More is always better for sure, except that it costs more. The A4000 came with FastRAM, so that is the AGA machine people could have bought if they thought it was needed


It's not as if there was no possible price point in-between those two extremes. Seriously, how much more would 1MB of onboard fast ram have cost back then? It literally would've been a game changer because if it were standard, all AGA games could have assumed it as a minimum and been more ambitious.

https://jcmit.net/memoryprice.htm

About 35 dollars for 1MB (consumer cost) in April 1992. One would hope they were able to source it for less.

Last edited by Karlos on 08-Oct-2022 at 10:53 PM.

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
BigD 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 22:48:22
#203 ]
Elite Member
Joined: 11-Aug-2005
Posts: 6915
From: UK

@Karlos

Yeah, ambition was severely lacking! Dennis AGA? Oscar? We had a platform that pushed the envelope with Shadow of the Beast etc and the AGA era is ushered in with Titus the Fox clones with as many colours and in the case of Oscar, reflections as you can throw at the screen! Only Team17, ClickBoom, the Fightin' Spirit guys, Black Magic and a handful of others really picked up the baton for really expanding on what people thought the Amiga could do and they needed extra fast ram as a minimum!

Last edited by BigD on 08-Oct-2022 at 10:49 PM.

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:10:40
#204 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@BigD

Oscar wasn't so bad, especially as a freebie.

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
kolla 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:11:35
#205 ]
Elite Member
Joined: 21-Aug-2003
Posts: 2478
From: Trondheim, Norway

@Karlos

Quote:

Karlos wrote:
@kolla

Sure, but games development is productivity, is it not?


Of course it is, but doesnt it help that the machine youre doing the development on has the same chipset that youre developing for?

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:13:16
#206 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@kolla

Of course it does. What are we arguing about, again?

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
Gunnar 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:21:32
#207 ]
Regular Member
Joined: 25-Sep-2022
Posts: 152
From: Unknown

Hello Karlos


Quote:
Seriously, how much more would 1MB of onboard fast ram have cost back then?


There are also other factors than increased price.

You surely want 32bit fast-memory, this means you need 4 memory chips each 8xbit.
Also you would need to include somewhere some glue logic for chip select and refresh logic.

I not see much free space for these extra 4 memory Chips on the A1200 mainboard.
Do you see a space for them?




 Status: Offline
Profile     Report this post  
kolla 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:25:33
#208 ]
Elite Member
Joined: 21-Aug-2003
Posts: 2478
From: Trondheim, Norway

@Karlos

Dont think we are, I just wanted to point out why a "big box" machine existed with AGA in the first place. Its kinda striking that there never was any expansion board for A1200 from CBM themselves. None that I can recall anyways.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

 Status: Offline
Profile     Report this post  
kolla 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:28:21
#209 ]
Elite Member
Joined: 21-Aug-2003
Posts: 2478
From: Trondheim, Norway

@Gunnar

At least the A1200 motherboard has space for a real FPU

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:30:03
#210 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@Gunnar

That's a better argument than most.

They found space for a 6888x FPU coprocessor that was never fitted. At least that's what I assume it was. However I think it's a bit moot: if the system were designed to have that additional memory, the layout of the whole thing could have been different.

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 8-Oct-2022 23:32:33
#211 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@kolla

Right. I have no disagreement there. The only thing I will say is that really it should've kept the SCSI. How much did IDE cost reduce the A4000 by?

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
bhabbott 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 0:11:10
#212 ]
Regular Member
Joined: 6-Jun-2018
Posts: 251
From: Aotearoa

@Hypex

Quote:

Hypex wrote:

While useful, especially when I added a HDD to mine, that was because they used a cheap IDE interface that was even cheaper with no DMA modes and only base PIO mode. This was a real step backwards really. The Amiga had DMA on SCSI controllers AS STANDARD.

Not really. Only the A3000 had SCSI with DMA built in, and it had problems. The A2090 and A2091 cards for the A2000 and the A590 expansion for A500 had DMA, but with only 24 bit addressing. Furthermore the MFM and XT-IDE drives supplied with them were not fast, so they were the bottleneck. The CDTV also had DMA with an optional SCSI interface, but was even more limiting with no Fast RAM. DMA certainly helped it perform better, but that is not really relevant because the machine is so limiting in other ways.

Quote:
But then IDE comes along, because the drives are cheap, along with Amiga IDE cards later and DMA is lost as a feature. There are no Amiga IDE cards that even support DMA I'm aware of making them all obsolete to a basic Amiga SCSI controller with DMA.

IDE was always about being cheap, since the drives were generally cheaper than SCSI drives. However even before that a lot of non-DMA ST506 and SCSI drive interfaces were produced by 3rd parties. The reason is that a DMA controller is expensive and hard to get working right, while not improving basic loading times significantly. Most people were happy just getting the extra speed and capacity of a hard drive at all, and DMA was only sought after by those who wanted the best possible performance and/or had specific requirements for video recording etc.

I know of at least one DMA IDE controller card, the Masoboshi Mastercard. But most other 3rd party IDE cards use PIO, even those produced before the A600 was released. So PIO was an 'industry standard' on the Amiga long before Commodore adopted it.

Quote:
Quote:
My 3rd hate was the stupid slow non-fifo serial port garbage.


They didn't update it. The CIA chips were just copies of the VIA chips used in the C64. They should have redesigned it to be faster and work with DMA. The design went against the Amiga. The Amiga had to use hardware interrupts to manage data. And each register in the CIA oddly took up 256 bytes in memory, 4KB for 16 registers per chip. It was an 8 bit chip, it wasn't design for a 16 bit computer.

Nitpick - the serial port is part of Paula, the CIA is only used for RS232 control and status lines.

The 68000 was designed to use 6800/6502 peripherals, including Commdore's VIAs - so it made perfect sense to use them in the Amiga. When the Amiga was designed the idea of high speed serial and parallel hadn't yet caught on. PCs didn't have buffered serial chips, and printers couldn't work at full bus speeds (the CIA's strobe pulse is deliberately slowed down to meet Centronics printer timing).

Next sticking point is that the Amiga didn't have a general purpose onboard DMA controller. It was expected that addon cards would provide their own if needed. The A3000 had one for SCSI, but even just doing that one job was difficult. If you put RAM on an accelerator card then the onboard DMA has to work over its bus at the same speed, which is problematic when the card is faster than the motherboard. If that is not possible then data still has to be copied into accelerator card RAM by the CPU. Another problem for serial ports is that DMA is generally block based, while serial is stream based and often needs to be monitored character by character.

The serial port in Paula was just a bonus for low speed uses that were common at the time (it did at least do MIDI, which standard PC serial ports couldn't because they didn't have the required baud rate). But the Amiga had an open hardware interface that was expected to be used for more demanding applications. There was nothing stopping from you installing a board with 16550 buffer serial chip if you needed it, or even a sophisticated comms board with local RAM and Z80 CPU.

Quote:
So the PC caught up with the Amiga in 1987. And added extra input lines. After then the Amiga could still compete as it wasn't until 1991 and 1992 that we saw EPP and ECP come to the fore.

Actually only the PS/2 line got a bidirectional printer port. Most clones stuck with the standard output-only port, and had to use 'nibble mode' via the status lines for input. Another thing they didn't have was power on the connector, so any device plugged into the port had to be separately powered.

Quote:
Once again Commodore had still not added DMA to hardware nor added extra lines. It fell behind.

It wasn't necessary. Any device that needed high speed parallel could use a plugin card. The primary use for EPP/ECP on PCs was GDI printers, which had secret proprietary protocols and needed large amounts of RAM and CPU power to operate.

Quote:
Sidenote:
The popular laplink cable is inferior to what the Amiga can do. It hacks data to printer lines and only transmits 5 bits at a time. An Amiga variant, if anyone ever tried and should work over a standard cable, would be superior and transfer 8 bits at a time with no hacks. But could be hacked to transfer more with a hybrid design. But who wants to manage 13 bits of data at once?

Indeed. Parnet on the Amiga is faster and more reliable than Laplink parallel on a PC. I used it regularly for transferring files between Amigas.


Quote:
Again, all they needed to do was engineer Paula and CIA to run at double clock. Couldn't be too hard.

Not quite. CIAs didn't need to be speeded up for HD floppies. Paula MFM decoding could easily have been speeded, but more DMA slots would have to be made available - with bigger buffers for it. Still doable, but not quite as simple and possibly breaking compatibility.

Quote:
Instead they hack HD drives to run at half speed. I'm sure it would have been more cost effective to speed up the chips

Perhaps not. One way was a clever hack using a couple of TTL logic chips in the floppy drive. The other involved critical redesign of the custom chips.

I think they should have done nothing, then Amiga fans wouldn't be disappointed by the results. 880k was plenty enough for normal uses, and CDROM was about to make floppies redundant for software distribution. Today PC don't have a floppy drive and you can't even buy HD disks new.

Quote:
and replace the old Shugart bus connector for the now standard PC one.

It's the same interface, just with different signals on the pins. Older floppy drives had jumpers to set the pin signals, to suit other devices which used a similar configuration to the Amiga. As competition heated up in the PC market, drive manufacturers dropped the jumpers and hardwired them to save money. Then they stopped making drives altogether. If Commodore had had survived they would have continued to have correctly configured drives made for them.

Quote:
Quote:
not 16 bit sound wasn't great either.


And they didn't even double the tracks to compensate.

Boohoo.

What good have 16 bit have done? A stock A1200 doesn't have the CPU power to decode MP3's or the memory to store large samples. If you want CD quality music then use a CDROM drive. Amiga MODs were always 8 bit - and sounded great. 8 track MODs generally don't add much. If you really wanted more then cards were available, but few people bought them because - unlike PCs - every Amiga has a 'sound card' built in.



 Status: Offline
Profile     Report this post  
Gunnar 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 0:16:41
#213 ]
Regular Member
Joined: 25-Sep-2022
Posts: 152
From: Unknown

@Karlos

At the end of the day... you always dream like this ...

How much nicer game grafics and animations could have been possible all AGA machines would have had 4 MB chipmem pre default?
How much more demanding games would have been came out - if the machines all had fastmem?
And so on ..

The A1200 could easily upgrade Fastmem with up to 8MB.
Your argument was if all A1200 would have has fastmemory, better games would have developed.
Maybe you can also turn the question around:
if there would an excellent game requiring 2 MB Fastmem - would then the users all simply bought a memory expansion?

I think Commodore tried to do something real good with A1200.
- IDE drives had a great performance / price value.
- PCMCIA port was cool.
- 020 CPU was nice
- the FPU option shows some good intention
- Expansion Port was good
Fastmem expansion were available.

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 0:33:25
#214 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@Gunnar

I was thinking more along the "minimum AGA specification" lines. If it had included a minimum of 68EC020 with 1 MB of CPU only fast memory rather than just adding more and more. Defining the lowest spec, not the highest.

Everyone and his dog knew back then that if you bought a bare A1200, it was half the machine I should've been until you put at least a trapdoor memory card in.

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
bhabbott 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 0:38:56
#215 ]
Regular Member
Joined: 6-Jun-2018
Posts: 251
From: Aotearoa

@Kronos

Quote:

Kronos wrote:
@bhabbott

Quote:

bhabbott wrote:

The original Amiga chipset was designed on the principle of doing graphics operations much faster than the CPU could, so it didn't matter if those operations blocked the CPU.


The original Amiga (aka everything OCS with a 68000) was designed so that the chipset could access the RAM without blocking much of the CPU access.

But the topic is AGA,

AGA continued the philosophy of OCS, offloading graphics operations from the CPU to free it up for other things. When you put Fast RAM in an A1200 this really shows up, but even with just Chip RAM it's noticeable. Everything runs smoother because the blitter and CPU are working in parallel.

Quote:
But even in that "really-just-7MHz"-mode the 020 could still do some GFX operations faster then the Chipset.

Sure, but that would be at the expense it of doing other stuff. I ran Fastblit on a 50MHz 030 equipped A1200 and it made no noticeable difference in the programs I was running. If the CPU is not much faster than the blitter then it still pays to use the blitter because it reduces CPU loading 'for free'.

Quote:
AGA was pretty much obsolete even when the 1st prototype A3000+ were build.

Most technologies are 'obsolete' by the time they enter production. But for Amiga users AGA was new, and a nice step up on what they had before. happily not too much of a step up though, so the A1200 was still mostly the Amiga we were familiar with.

 Status: Offline
Profile     Report this post  
Hammer 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 1:36:40
#216 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@kolla

Quote:

kolla wrote:
@Gunnar

At least the A1200 motherboard has space for a real FPU

It's nearly pointless for games. I rather have AT&T DSP3210 in that space or remove it completely.

68881/68882 wasn't designed to be DSP RISC fast single cycle throughput floating point co-processors.

Motorola's DSP56000 was INT16/INT24 integer only DSP while AT&T DSP3210 supports INT16, INT24, and FP32 data formats.

AT&T DSP3210 has up to 33 MFLOPS FP32 from 33 Mhz.

Without an abstraction API layer, the problem with AT&T DSP3210 is the road map and its future support.

For multimedia and 3D, modern CPUs have at least integrated single-cycle throughput floating point co-processors.



Last edited by Hammer on 09-Oct-2022 at 01:39 AM.

_________________
Ryzen 9 7900X, DDR5-6000 32 GB RAM, GeForce RTX 4080
Amiga 1200 (rev 1D1, KS 3.2, TF1260, 68060 @ 63 Mhz, 128 MB)
Amiga 500 (rev 6A, KS 3.2, PiStorm/RPi3a/Emu68)

 Status: Offline
Profile     Report this post  
Hammer 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 2:00:55
#217 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@cdimauro

Quote:

cdimauro wrote:

I agree on this.

The Amiga was made for 2D games. And those games require chip ram. ONLY chip ram.

Fast ram was NOT useful neither important.


https://www.youtube.com/watch?v=GojpwZMBHz4
Arcade quality Final FIght port with stock 68020 with Fast Ram.

Fast Ram is important for CPU's software 2D BOBs when it operates concurrently with AGA's 16-bit Blitter on Chip RAM.

My Amiga 1200 with Amikit 8MB Fast RAM expansion speeds up games like Wing Commander AGA. Wing Commander PC was released in 1990.

You didn't read David Pleasance's book when big game developers are requesting Fast RAM-equipped A1200 SKUs.

The need for specialized external hardware is due to Motorola didn't include small-and-fast DSP/RISC instructions (one instruction per clock) in their 68K (not including 68060).
--------------
https://www.youtube.com/watch?v=vtZCsgC_LEs
Amiga 500 PiStorm/Emu68 with modern Fast RAM playing 68K OpenBOR 2D engine.

These examples are showing Pentium II 266/300 performance class CPUs (based on Quake demo3 benchmarks) with Fast RAM driving 2D games with superior results when compared to Fighting Spirit AGA.

To be fair for Apollo-Core, https://youtu.be/ugbuywwq_jQ?t=90
Amiga 1200 Vampire 2 playing OpenBOR 2D engine. About Pentium MMX CPU performance class with efficient 75 Mhz 64-bit FSB memory bandwidth-like results. Certain Socket 7 motherboards have support for 75 Mhz and 100 Mhz FSB, hence overclocking Pentium MMX 166 into 190 Mhz was easy with 75 Mhz FSB.

Last edited by Hammer on 09-Oct-2022 at 02:49 AM.
Last edited by Hammer on 09-Oct-2022 at 02:43 AM.
Last edited by Hammer on 09-Oct-2022 at 02:41 AM.
Last edited by Hammer on 09-Oct-2022 at 02:39 AM.
Last edited by Hammer on 09-Oct-2022 at 02:31 AM.
Last edited by Hammer on 09-Oct-2022 at 02:02 AM.
Last edited by Hammer on 09-Oct-2022 at 02:01 AM.

_________________
Ryzen 9 7900X, DDR5-6000 32 GB RAM, GeForce RTX 4080
Amiga 1200 (rev 1D1, KS 3.2, TF1260, 68060 @ 63 Mhz, 128 MB)
Amiga 500 (rev 6A, KS 3.2, PiStorm/RPi3a/Emu68)

 Status: Offline
Profile     Report this post  
Hammer 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 2:11:56
#218 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@Gunnar

Quote:

Gunnar wrote:
@Karlos

At the end of the day... you always dream like this ...

How much nicer game grafics and animations could have been possible all AGA machines would have had 4 MB chipmem pre default?
How much more demanding games would have been came out - if the machines all had fastmem?
And so on ..

The A1200 could easily upgrade Fastmem with up to 8MB.
Your argument was if all A1200 would have has fastmemory, better games would have developed.
Maybe you can also turn the question around:
if there would an excellent game requiring 2 MB Fastmem - would then the users all simply bought a memory expansion?

I think Commodore tried to do something real good with A1200.
- IDE drives had a great performance / price value.
- PCMCIA port was cool.
- 020 CPU was nice
- the FPU option shows some good intention
- Expansion Port was good
Fastmem expansion were available.

My Amiga 1200 rev 1D1 has a timing bug, hence a simple Amikit 8MB Fast RAM expansion still halts the system, not just with TF1260.

Commodore wasn't serious about A1200's internal expansion since they shipped certain A1200 batches with a timing bug.

My older self was able to fix A1200's timing bug, but my younger self wouldn't be able to fix the timing bug.


_________________
Ryzen 9 7900X, DDR5-6000 32 GB RAM, GeForce RTX 4080
Amiga 1200 (rev 1D1, KS 3.2, TF1260, 68060 @ 63 Mhz, 128 MB)
Amiga 500 (rev 6A, KS 3.2, PiStorm/RPi3a/Emu68)

 Status: Offline
Profile     Report this post  
matthey 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 3:43:06
#219 ]
Super Member
Joined: 14-Mar-2007
Posts: 1684
From: Kansas

Chipset rating system for desktop use
high end
mid end
low end
obsolete (retro)

Amiga Corporation developed chipsets
1985 OCS - high end (what is competitive in the price range?)
1989 Ranger - high end (higher performance & cost but VRAM price drops would have kept it practical)

CBM developed chipsets
1990 ECS - low end/obsolete (competitive with VGA & EISA without VRAM)
1992 AGA - low end (competed with VGA & EISA but was slow)

The chipset does not show the overall desktop competitiveness though.

1985 68000 OCS 256kiB - mid end (68020 and more memory were possible)
1990 68000 ECS 1MiB - obsolete (386 VGA EIDE 2MiB was low end)
1990 68030 ECS 2MiB - low end (Amiga 3000 was low end with mid to high end price)
1992 68020 AGA 2MiB - obsolete (68020@14MHz was a joke/retro)

While the 68020+AGA Amiga was obsolete on the desktop, there were a few things that kept it from being completely obsolete for all markets in 1992.

1) super low price
2) modular upgrade path
3) small footprint
4) retro 2D chipset didn't need much CPU power
5) NTSC/PAL video output features were still useful
6) Amiga backward compatibility

The 1993 Amiga CD32 hardware was mostly obsolete for the game console market yet it had a successful release outselling all other console CD based systems in the UK. It was practically retro by this time and added two more benefits which were cheap CD media and cheap minimal modular embedded hardware (perfect for kiosks). The Amiga was able to do something that the C64 was unable to do successfully which was to transition to a low cost console. The problem was that it was so late as CBM had bet on 68000+ECS with the CDTV, Amiga 500+ and Amiga 600 instead of 68020+AGA Amiga 1200 and Amiga CD32. AGA was successful even though it was late and lacking. There is ever reason to believe CBM would have survived if AGA was not delayed by Bill Sydnes. Maybe we would have seen AA+ in 1994-1995 then. Surviving the Sony Playstation would have been tough as Sony came in and aggressively integrated 3D and pushed down the price like the Amiga had done for 2D in 1985 (and CBM failed to do with Hombre). The Amiga CD32 may have survived if it could have reached half the price of the $299 Playstation which I think is possible. Faster processors and 3D cores could have been licensed for the Amiga as well. CBM was trying to turn a cost reduced 68000+OCS Amiga down into a C64 from day one. The Amiga was designed to be more versatile and expandable from the start but CBM management lacked the vision of Jay Miner and didn't enhance it or continue to integrate it. After 7 years of CBM doing practically nothing with the Amiga chipset, the competition passed it up. AGA gave us a too little too late idea of what could have been done and the AA+ specs shows us what AGA should have been around 1990.


What Amiga was most desirable in 1992 for the base model...
A) Amiga 600 68000@7MHz ECS 2MiB $499 (Amiga to C64 rebirth CBM thought we wanted)
B) Amiga 1200 68020@14MHz AGA 2MiB $599 (actual best seller)
C) Amiga ? 68030@21MHz AA+ 4MiB $649 (potential AA+ spec and price)
D) Amiga ? 68040@28MHz AAA 8MiB $1199 (32 bit AAA no VRAM)

I think C would have been the most competitive for low end desktop and console markets although B was likely better for embedded use. D was not practical for the low end and A was what the hell were you thinking CBM?

Last edited by matthey on 09-Oct-2022 at 06:38 AM.
Last edited by matthey on 09-Oct-2022 at 06:38 AM.

 Status: Offline
Profile     Report this post  
cdimauro 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 9-Oct-2022 7:40:16
#220 ]
Elite Member
Joined: 29-Oct-2012
Posts: 3104
From: Germany

@Karlos

Quote:

Karlos wrote:
@cdimauro

One could make the claim that if additional Fast Ram were present by default on AGA systems that larger and more complex 2D games would be possible as larger, more complex level data could reside in fast memory as well as the code and all of the chip ram could be for media assets. Games fit their non graphical data and code into tens of KB because that was their budget.

You cannot arbitrarily add code or data to games only because you've available space.

2D games are bounded from the assets (graphic, primarily). And this required chip mem.

Fast mem doesn't change much the situation.

@Hammer

Quote:

Hammer wrote:
@cdimauro

Quote:

cdimauro wrote:

I agree on this.

The Amiga was made for 2D games. And those games require chip ram. ONLY chip ram.

Fast ram was NOT useful neither important.


https://www.youtube.com/watch?v=GojpwZMBHz4
Arcade quality Final FIght port with stock 68020 with Fast Ram.

Again? You already provided that link and I've already replied you. Why continue to bring complete non-sense?

Here's what the video reports:
Machine: Amiga 1200, 68020, 64 MB Fast RAM (emulated)

And looking at two videos with the real machine:
Machine: Amiga 4000T, 68040/25 MHz, 12 MB Fast RAM

I've no problem producing something similar with such massive specs. Even by using MAME on such machines and getting the original Final Fight running...
Quote:
Fast Ram is important for CPU's software 2D BOBs when it operates concurrently with AGA's 16-bit Blitter on Chip RAM.

What's not clear to you that the CPU was NOT fast enough for rending BOBs? The Blitter was much faster for this tasks.

Or were you talking about high-end systems with 68040 or even a 68060?

Do you understand that you cannot increase the hardware specs because you increase (A LOT) the price of the final computer?
Quote:
My Amiga 1200 with Amikit 8MB Fast RAM expansion speeds up games like Wing Commander AGA. Wing Commander PC was released in 1990.

Sure: it's a fake 3D game, so got some boost.
Quote:
You didn't read David Pleasance's book when big game developers are requesting Fast RAM-equipped A1200 SKUs.

For which types of games?

Because developers required more RAM for sure, but chip mem was the most important one: the only accessible by the chipset.
Quote:
The need for specialized external hardware is due to Motorola didn't include small-and-fast DSP/RISC instructions (one instruction per clock) in their 68K (not including 68060).

Are you kidding?!? No 68k CPU instruction was executed in one cycle. Except for 68040 and 68060.

How much expensive should have been such computers?
Quote:
--------------
https://www.youtube.com/watch?v=vtZCsgC_LEs
Amiga 500 PiStorm/Emu68 with modern Fast RAM playing 68K OpenBOR 2D engine.

These examples are showing Pentium II 266/300 performance class CPUs (based on Quake demo3 benchmarks) with Fast RAM driving 2D games with superior results when compared to Fighting Spirit AGA.

To be fair for Apollo-Core, https://youtu.be/ugbuywwq_jQ?t=90
Amiga 1200 Vampire 2 playing OpenBOR 2D engine. About Pentium MMX CPU performance class with efficient 75 Mhz 64-bit FSB memory bandwidth-like results. Certain Socket 7 motherboards have support for 75 Mhz and 100 Mhz FSB, hence overclocking Pentium MMX 166 into 190 Mhz was easy with 75 Mhz FSB.

And here's the usual padding...

 Status: Offline
Profile     Report this post  
Goto page ( Previous Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 Next Page )

[ home ][ about us ][ privacy ] [ forums ][ classifieds ] [ links ][ news archive ] [ link to us ][ user account ]
Copyright (C) 2000 - 2019 Amigaworld.net.
Amigaworld.net was originally founded by David Doyle