English Amiga Board


Go Back   English Amiga Board > Main > Amiga scene

 
 
Thread Tools
Old 16 July 2019, 20:07   #501
Juz400
Registered User

 
Join Date: Mar 2017
Location: London
Posts: 61
The 1200 was nice enough to look at but the specs were a little low,
RAM cards were thin on the ground and VERY expensive at the start.

I do think CBM took the wrong route, if AGA was all they could offer at the time
then they should have treated the A600 as the `entry level` and the 4000 should have
been 2 Cased models. Take a look at 8-Bit Guys video on CBM IBM Clones.
Does something look a bit familiar at time index 3.39 and 17.04

[ Show youtube player ]

THAT should have been the low end AGA/1200 and the released 4000 as the high end.
A 14-20 Mhz 68020 could have supplied on the CPU card and 1 or 2 ZORRO only slots on the riser
It would have save CBM a buttload of cash with one mainboard to develop and manufacture.
This would have enabled them to offer the machines at a lower price at the higher end and slightly
more expensive at the lower end. HDD and Fast ram would have been cheaper to the user and more likely
purchased. CD`s were not a massive thing in 1992 and doubtfully missed by the user.
Overall the machine would felt more `Grown-Up` and not just a toy `game computer`
in comparison to the IBM Clone alternative

Last edited by Juz400; 16 July 2019 at 20:11. Reason: times on video
Juz400 is offline  
Old 17 July 2019, 10:17   #502
activist
Registered User

 
Join Date: May 2017
Location: here
Posts: 39
Quote:
Originally Posted by Juz400 View Post
THAT should have been the low end AGA/1200 and the released 4000 as the high end.
You're describing the 1991 planned A1000+ which was in the works.
Yes, sound reasoning it would have made perfect sense. Logically thinking.

But there was never a plan or logic. So cancelled it was therefore
activist is offline  
Old 17 July 2019, 10:43   #503
Dunny
Registered User

Dunny's Avatar
 
Join Date: Aug 2006
Location: Scunthorpe/United Kingdom
Posts: 1,276
But wasn't the A1200 the low-end anyway? Ar am I confusing it with the A600?

If the A1200 had had better specs and not been the low-end of the range, I would not have bought it. And neither would any other significant chunk of the userbase. And there was no point to owning the A600 when I already had an A1000.
Dunny is offline  
Old 17 July 2019, 10:55   #504
sandruzzo
Registered User
 
Join Date: Feb 2011
Location: Italy/Rome
Posts: 1,463
A1200, fully 32 bit and a good Akiko able to do c2p without cpu aid, Thats' would have been a great Base machine to do a lot.
sandruzzo is offline  
Old 17 July 2019, 14:03   #505
roondar
Registered User

 
Join Date: Jul 2015
Location: The Netherlands
Posts: 1,230
Note: to keep the post a bit shorter, I cut parts of the posts I'm replying to and tried to choose only one post per subject to reply to. This isn't because the rest of the posts are not relevant. It's just to try and keep things a bit more readable.

Quote:
Originally Posted by chb View Post
While the Archimedes is indeed impressive, especially considering it gets its power from considerable simpler chips (CPU, VIDC, MEMC) than those in the A500, the specs do not tell the whole truth. The 256 color mode on the Archimedes is very much crippled - just 16 color registers, the remaining 240 colors are derived from those 16 in some quite obscure way, most games just used the standard palette and therefore look considerably worse than real 256 color modes like VGA or AGA. I would even say that Amiga EHB in general looks better than 256 colors on the Archimedes (it's however a chunky mode).
...
I do agree specs don't tell the whole story, but if I look at 2D Amiga ports on the system, I don't see "worse". I generally see "equal or better". And that's not even looking at the machine's strength: 3D games were pretty awesome on the Archimedes. This is kind of my point: the Acorn could do 2D games on a very similar level as the Amiga, often even running better despite not having as much custom hardware. Add the 3D games to it and I'd say that the system was much stronger for games than it generally gets credit for. And yet, it was released in 1987 - the year of the A500.

As for the 256 colour mode, I do agree it was pretty odd and requires some good choices for base colours to work (kind of like HAM), but it did seem they got better at choosing how to employ it down the line.

To show what I mean about ported games, here's the Archimedes versions of a few Amiga games. They look, sound and run at least as well as Amiga version:

Gods: [ Show youtube player ]
Pacmania: [ Show youtube player ] (bigger ghosts, more colours)
SWIV: [ Show youtube player ] (twice the frame rate)
Chuck Rock: [ Show youtube player ] (more colours in both of the layers)
Lotus 2: [ Show youtube player ] (higher frame rate that holds in 2 player mode)
Lemmings: [ Show youtube player ] (higher frame rate)

Some of the videos aren't great, but there's far less to choose from here as the Archimedes was not that popular as pointed out before.
Quote:
Originally Posted by Thomas Richter View Post
There are multiple reasons for crashes and hangs if you use the wrong kickstart.
...
Thanks for the information, I find it fascinating to see how little the 68020+ was used in the later versions of the OS. Was this purely for compatibility, or were there other reasons?
Quote:
Originally Posted by Puggsy View Post
The A1200 wasn't a disappointment, it still is a massive disappointment.
There aren't many reasons to own one. Few games, most times just slightly updated ones from ECS, a few demos here and there (and lots require super expensive 060s to run)...
Well, I could try to show you why I think this is not all that accurate, but if you still think this after all these pages of this thread, I don't think you'd change your mind.

Suffice to say I don't agree.
Quote:
There are easier and cheaper ways to run Amiga stuff and the 1200 is not one of them.
The A1200 with WHDLoad is pretty much the de-facto standard for running games and demos on real hardware, though. This is not because it is poorly suited for the job.

As for easier: IMHO only emulation is easier. Of all the 'real' classic Amiga's available, the A1200 is by far the easiest and convenient to get running your games & demos on: add a CF card, WHDLoad and almost everything just works.
Quote:
Originally Posted by vulture View Post
So, here's a rather bad clip of DoomAttack running Doom on my Blizzard MK2. Personally, as I said, I find it to be very playable:

[ Show youtube player ]
Thanks for the video!
And I'll eat crow - it runs a lot better than I thought it would. I remember it running much worse, guess I was wrong. Which is good news, as that does kind of show that even plain AGA could do these games better than I and others here seem to think

Is that in 256 colours/high details or did you use 64 colours and low details? Or some other combination? And because I now want to try this for myself: could you post me where I can find the executable, plus all the settings etc you used? Looks like fun!

Assuming I can find the .WAD files somewhere
Quote:
Originally Posted by Dunny View Post
But wasn't the A1200 the low-end anyway? Ar am I confusing it with the A600?
It was very much the low end, yes. The A600 was discontinued within a year of going on sale (the A500/A500+ had already been discontinued earlier), leaving only the A1200 as low end option.
roondar is offline  
Old 17 July 2019, 14:50   #506
Thomas Richter
Registered User
 
Join Date: Jan 2019
Location: Germany
Posts: 234
Quote:
Originally Posted by roondar View Post
Thanks for the information, I find it fascinating to see how little the 68020+ was used in the later versions of the OS. Was this purely for compatibility, or were there other reasons?

I do not really know since I did not prepare the kickstart and I am neither working for CBM. But if I would guess then there would be two reasons: First, it does not offer much of an improvement in terms of speed, except cutting down the size of the kickstart, which was indeed a practical problem. Thus, instead of delivering the full flexible kickstart, deliver only what was necessary. Second, it means modifying the build infrastructure for all other components just to make the machine dependent, and that is already quite a bit of work.


In the end, it rather worked against this idea as soon as CPU upgrades for the A1200 appeared, such as those equipped with the 68060. This led to all wierd workarounds such as the 68060.library having to remove mathieeesingbas because its 68020 based branch is inappropriate for the 68060.


So, I guess, the answer is that this was probably a bad idea, and it should not be repeated.
Thomas Richter is offline  
Old 17 July 2019, 16:00   #507
vulture
Registered User
 
Join Date: Oct 2007
Location: Athens , Greece
Posts: 956
@roondar

I've used DoomAttack AIO http://aminet.net/game/shoot/DoomAttack-AIO.lha and I've just copied Doom1.wad , Doom2.wad and DoomU.wad into its base directory. I've found that on my setup, the 020 optimized (non-blitter) c2p is the fastest and also the doomsndlibrary plugin for audio to be the least taxing. Using NTSC 320x200@8bit color, HUD overlay is turned off and (obviously) the .020 executable is best suited in my case.
As for the in-game settings, as you can see, I've reduced the window size just one level and set the detail to low. I've no idea how big a difference that setting does to the visual quality when it runs on a smaller screen, but on my already overblown 32'' TV, it's negligible. DoomAttack allows for full mouse control and I suggest you use it, makes it more fun to play.
Hope you enjoy it!

EDIT:
Just out of curiosity, I've tried the c2p_020_blitter routine and I think it is noticeably smoother after all. Also, I think DoomAttack has no option for less than 8bit colour. Maybe that's an ADoom-only feature?

Last edited by vulture; 17 July 2019 at 21:09.
vulture is offline  
Old 17 July 2019, 16:52   #508
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 634
Quote:
Originally Posted by roondar View Post
Thanks for the video!
And I'll eat crow - it runs a lot better than I thought it would. I remember it running much worse, guess I was wrong. Which is good news, as that does kind of show that even plain AGA could do these games better than I and others here seem to think
I hope you don't include me in that statement because:

Quote:
Originally Posted by grond View Post
Doom can run well enough on an 030.
I knew a fast 030 can run Doom just fine on an AGA Amiga. An only standard 030 with a chunky mode would have done just as well - the planar mode is just a penalty but one that would have required an expensive CPU instead of a mid-range CPU. And suddenly there wouldn't have been a reason why a reasonably priced "A1300"-type Amiga with chunky (or at least an "A1200+" with a choice between an 020 or an 030) couldn't have run Doom. However, with that c2p ordeal nobody in the industry even wanted to try implement Doom on an Amiga. Only die-hard Amigans were insistent enough to finally solve the c2p problem as well as possible (years after Doom came out).

As for our discussion about whether Doom requires processor grunt or memory bandwidth:

Of course, having too little of either will always affect the overall performance. If I have too little memory bandwidth, the CPU will be waiting for the memory while trying to get rid of the calculated data. If I have too little CPU power, I won't be able to make use of the memory bandwidth because I haven't got the data ready for the next RAM slot.

Another thing may be that we simply understand different things as "CPU grunt". Doom only requires very few simple instructions (ADD, ADDX and MOVE) and mostly reads texture data and writes to the screen buffer. CPU grunt to me would mean complex algorithms that use many complex instructions such as MUL, DIV or even floats.

The 020 and 030 need quite a few CPU cycles to perform a read operation from memory and a write operation to memory. I attribute that to the (lacking) memory bandwidth side, you could just as well consider that as forming part of the (lacking) CPU grunt aspect.
grond is offline  
Old 17 July 2019, 21:08   #509
roondar
Registered User

 
Join Date: Jul 2015
Location: The Netherlands
Posts: 1,230
Quote:
Originally Posted by vulture View Post
@roondar

I've used DoomAttack AIO http://aminet.net/game/shoot/DoomAttack-AIO.lha and I've just copied Doom1.wad , Doom2.wad and DoomU.wad into its base directory. I've found that on my setup, the 020 optimized (non-blitter) c2p is the fastest and also the doomsndlibrary plugin for audio to be the least taxing. Using NTSC 320x200@8bit color, HUD overlay is turned off and (obviously) the .020 executable is best suited in my case.
As for the in-game settings, as you can see, I've reduced the window size just one level and set the detail to low. I've no idea how big a difference that setting does to the visual quality when it runs on a smaller screen, but on my already overblown 32'' TV, it's negligible. DoomAttack allows for full mouse control and I suggest you use it, makes it more fun to play.
Hope you enjoy it!
Thanks for the info, I'll be sure to give it a spin. Perhaps even try Wolvenstein as well for an even more retro 3D-experience.
Quote:
Originally Posted by grond View Post
I hope you don't include me in that statement because:

I knew a fast 030 can run Doom just fine on an AGA Amiga. An only standard 030 with a chunky mode would have done just as well - the planar mode is just a penalty but one that would have required an expensive CPU instead of a mid-range CPU. And suddenly there wouldn't have been a reason why a reasonably priced "A1300"-type Amiga with chunky (or at least an "A1200+" with a choice between an 020 or an 030) couldn't have run Doom. However, with that c2p ordeal nobody in the industry even wanted to try implement Doom on an Amiga. Only die-hard Amigans were insistent enough to finally solve the c2p problem as well as possible (years after Doom came out).
I wasn't talking about you, no. You said several times you felt the 68030 could do it. The price of your hypothetical A1300 would've been interesting to see, though. I couldn't find a direct cost for the different 68030 speed grades at the time, but I did find two turbo cards by the same manufacturer: one with a 25MHz 68030 and one with a 40MHz 68030. The latter was 120 pounds more expensive.

Regarding the implementing of Doom on the Amiga, as I understood it John Carmack himself said at one point he felt it needed a 68040@25MHz and 4MB RAM and he didn't think anybody would buy them. Do note this is from memory though, I have no quote and may have misremembered.

I do agree that a chunky mode would've been better, but looking at Doom in action now it seems to run a good deal better on a fast 68030 than any of the 386 examples I could find. Taking this and several posts by others in this thread insisting that even a low specced 386 ran Doom well enough using a smaller windo into account... Well, I'm beginning to think Doom for Amiga with AGA would probably have been just as 'good enough' if a proper c2p solution had been available earlier and a reasonably fast 68030 was used.

Vulture's demonstration and me looking up 68060 performance for games like Doom on the A1200 have made me change my mind. Seeing how well it actually runs on a 68030 A1200, I'm simply no longer sure that AGA's low bandwidth or lack of c2p was really as much of a problem as some in this thread have stated. Not that they weren't problems (they were!), but it seriously seems to me that the negative aspects of this part of AGA have been exaggerated somewhat.
Quote:
As for our discussion about whether Doom requires processor grunt or memory bandwidth:

Of course, having too little of either will always affect the overall performance. If I have too little memory bandwidth, the CPU will be waiting for the memory while trying to get rid of the calculated data. If I have too little CPU power, I won't be able to make use of the memory bandwidth because I haven't got the data ready for the next RAM slot.

Another thing may be that we simply understand different things as "CPU grunt". Doom only requires very few simple instructions (ADD, ADDX and MOVE) and mostly reads texture data and writes to the screen buffer. CPU grunt to me would mean complex algorithms that use many complex instructions such as MUL, DIV or even floats.

The 020 and 030 need quite a few CPU cycles to perform a read operation from memory and a write operation to memory. I attribute that to the (lacking) memory bandwidth side, you could just as well consider that as forming part of the (lacking) CPU grunt aspect.
I think you hit the nail on the head here. When I talk about memory bandwidth, I basically mean the speed of the bus. When I talk about CPU grunt, I basically mean how long the CPU takes to execute instructions, including those that move data onto the bus.

I'm not actually all that clear on how fast memory access on a 486 is, but I do know that a 68030@50MHz should be able to (in theory at least) do 50MB/sec, if it has zero wait state 32 bit memory. It might even be faster if it runs a copy loop from the cache or uses movem. But I honestly don't know how many cycles a move takes when run from the cache, nor what the cycle count for movem on a 68030 is.
roondar is offline  
Old 17 July 2019, 22:54   #510
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 634
Quote:
Originally Posted by roondar View Post
The price of your hypothetical A1300 would've been interesting to see, though.

I estimate that AGA with chunky, 1 MB of chipmem, 2 MB of fastmem and a 25 MHz 030 would have sufficed to run Doom decently. This could have been a reasonably priced Amiga at the time Doom came out. Some of these features could have been easy upgrades to the hypothetical A1300. The biggest impediment certainly was the very low number of Amigas that had more than the well-known minimum specs (because Commodore chose lowering the price instead of adding hardware value).




Quote:
I'm simply no longer sure that AGA's low bandwidth or lack of c2p was really as much of a problem as some in this thread have stated. Not that they weren't problems (they were!), but it seriously seems to me that the negative aspects of this part of AGA have been exaggerated somewhat.

The negative aspects of planar graphics are (just summing up):


They make the porting of the game harder and may even have made it seem impossible at the time.


They require a higher CPU grade to achieve the same fps that would have been possible with a lower CPU grade if a chunky mode had been available. This mainly raises the price of the Amiga hardware capable of running the game.




Quote:
I'm not actually all that clear on how fast memory access on a 486 is, but I do know that a 68030@50MHz should be able to (in theory at least) do 50MB/sec, if it has zero wait state 32 bit memory. It might even be faster if it runs a copy loop from the cache or uses movem. But I honestly don't know how many cycles a move takes when run from the cache, nor what the cycle count for movem on a 68030 is.

I think the best the 030 can do is a memory operation every four clock cycles. Cache does not make this number smaller.
grond is offline  
Old 17 July 2019, 23:43   #511
roondar
Registered User

 
Join Date: Jul 2015
Location: The Netherlands
Posts: 1,230
Quote:
Originally Posted by grond View Post
I estimate that AGA with chunky, 1 MB of chipmem, 2 MB of fastmem and a 25 MHz 030 would have sufficed to run Doom decently. This could have been a reasonably priced Amiga at the time Doom came out. Some of these features could have been easy upgrades to the hypothetical A1300. The biggest impediment certainly was the very low number of Amigas that had more than the well-known minimum specs (because Commodore chose lowering the price instead of adding hardware value).
I see where you're coming from, but you do realise that anything over 2MB of total RAM was considered 'high end' in 1992, right? Putting 3MB of RAM in an entry level machine would've been epic but IMHO rather unlikely considering the market at the time. Even many 486 PCs were sold with only 2MB at the time.

Make it 1+1MB and I'd personally consider it a much more likely scenario. Perhaps upgrade the CPU to 33MHz instead. And while you're at it, include a SIMM slot or two. Retro-active speculation is fun
Quote:

The negative aspects of planar graphics are (just summing up):

They make the porting of the game harder and may even have made it seem impossible at the time.

They require a higher CPU grade to achieve the same fps that would have been possible with a lower CPU grade if a chunky mode had been available. This mainly raises the price of the Amiga hardware capable of running the game.
I know and understand each of these. Like I've said many times in this thread before: I'm not saying a chunky mode would not be an improvement.

However, after seeing examples of the game in action on AGA, I don't find the disadvantages as relevant as before. This is due to the result we eventually ended up with: the slow, planar based AGA chipset eventually got a version of Doom that ran better on a 68030@50MHz using c2p than on any of the 386 machines I saw - even the 386DX40 ones. Well, as far as I can tell from the video posted by Vulture and my own YouTube searches anyway. I watched something like 30 or 40 Doom videos on 386 and 486 PC's over the past few days. Plus some more of it running on various Amiga configurations

Perhaps the difference is solely due to the higher clock speed (which I'd expect to have been 'eaten' by the c2p+lower speed bus), but I found it to run much better than I expected.

Or put another way: the practical demonstration I saw today made me rethink the implications of the more theoretical discussion we've been having. It really surprised me just how well it did.

This isn't to say that this end result would've been what we'd have gotten back in 1993 should a port have been made at the time. But to me it shows that - despite AGA's clear disadvantages - the (eventual) real world differences in performance between it and chunky graphics are not as big as I originally thought.
Quote:
I think the best the 030 can do is a memory operation every four clock cycles. Cache does not make this number smaller.
Yeah, that's what I figured too. Hence the 50MB/sec theoretical limit on the 68030@50MHz I posted.
roondar is offline  
Old 18 July 2019, 00:03   #512
Galahad/FLT
Going nowhere

Galahad/FLT's Avatar
 
Join Date: Oct 2001
Location: United Kingdom
Age: 45
Posts: 7,280
Those Archimedes conversions are a misnomer to say they were better or equal than the Amiga versions.

#1 The Amiga versions in a lot of those cases weren't even bespoke Amiga versions, they were mostly ST conversions with slight improvements, but certainly not enough to say the Amiga got the very best it was capable of.

#2 The Archimedes versions usually came out quite some time after the Amiga version, so it stands to reason that there was less of a rush/deadline to get the games converted.

Chuck Rock on both machines is the same, its not that the Arch has more colours, it seems they've changed some of them for different shades.

Pacmania on Amiga was also PAL full overscan and a full TWO years before the Archimedes version.

Lemmings also was a year after the Amiga version, and frankly, looks as jerky as the Amiga version when it scrolls, its certainly not smooth, but then the Amiga version suffers from having to consider the Atari ST which unusually for DMA Design, had an influence on the Amiga version, but then Lemmings was never about the flashy visuals and scrolling in the first place.

SWIV again has legacy issues with the Atari ST and its stupid Dynamic Loading Interrupt system which affects the scrolling in parts if it can't load the required graphics in quickly enough, and again, Amiga SWIV was a year before Archimedes version.

I'm not saying the Arc is crap, its clearly not, but the Amiga whilst got a lot of games the Arc didn't or got them before, wasn't always given full consideration to making it the very best it could be, whilst Krisalis had time and inclination to spend more time on their conversions.
Galahad/FLT is offline  
Old 18 July 2019, 01:07   #513
roondar
Registered User

 
Join Date: Jul 2015
Location: The Netherlands
Posts: 1,230
Quote:
Originally Posted by Galahad/FLT View Post
Those Archimedes conversions are a misnomer to say they were better or equal than the Amiga versions.

#1 The Amiga versions in a lot of those cases weren't even bespoke Amiga versions, they were mostly ST conversions with slight improvements, but certainly not enough to say the Amiga got the very best it was capable of.

#2 The Archimedes versions usually came out quite some time after the Amiga version, so it stands to reason that there was less of a rush/deadline to get the games converted.

Chuck Rock on both machines is the same, its not that the Arch has more colours, it seems they've changed some of them for different shades.

Pacmania on Amiga was also PAL full overscan and a full TWO years before the Archimedes version.

Lemmings also was a year after the Amiga version, and frankly, looks as jerky as the Amiga version when it scrolls, its certainly not smooth, but then the Amiga version suffers from having to consider the Atari ST which unusually for DMA Design, had an influence on the Amiga version, but then Lemmings was never about the flashy visuals and scrolling in the first place.

SWIV again has legacy issues with the Atari ST and its stupid Dynamic Loading Interrupt system which affects the scrolling in parts if it can't load the required graphics in quickly enough, and again, Amiga SWIV was a year before Archimedes version.

I'm not saying the Arc is crap, its clearly not, but the Amiga whilst got a lot of games the Arc didn't or got them before, wasn't always given full consideration to making it the very best it could be, whilst Krisalis had time and inclination to spend more time on their conversions.
Before I continue, I want to re-iterate that my whole reason for mentioning the Archimedes in the first place was a post wherein the writer essentially said that the A500 was better at 'everything' than any of the competition and that the A1200 was not. All I tried to do was show that there were systems in 1987 that were comparable or better in abilities (how much better is, naturally, a matter of debate and what you think is important in a machine) and that therefore the situation when they both launched was more similar than he thought.

That is all, it's not meant to say the A500 was a poor machine. It rocked.

With that out of the way:

#1 Not all the ports I show are from poor ST conversions. These are just games I remembered from ads I saw or interviews with programmers I saw/read. And please note: this list is not meant to be some sort of 'Archimedes rules, Amiga sucks' thing. Both are great. And for all the power of the Archimedes, I'd still much rather have and use my A500, A600 & A1200.

#2 I can't say I agree with that position. The Archimedes market was much, much smaller than the Amiga one. Game sales were therefore very likely to much lower, which in turn limits the amount of money (and hence time) that can be invested in these ports. As a result, I don't consider a game having a later release to be all that relevant and won't comment on it in any of the other points as these posts of mine are long enough as is

#3 Chuck Rock is not actually the same, it has more colours in the foreground. You can clearly see this immediately, right at the first screen of the game proper: the tree on the Amiga uses the same colours for the bark as are used for the ground texture, on the Archimedes the bark has it's own set of colours. The same goes for the first dinosaur you spot: on the Amiga it's blue and adds a green shade from the tree leaves. On the Archimedes, it has several blue shades instead that are not repeated elsewhere on screen. Similar things keep happening throughout the 1st level: the rocks have shades in them that are not repeated elsewhere in the foreground or background. Etc.

The issue is quite obviously that the Amiga version is running in Dual Playfield mode, which limits it's colour palette per layer and requires doubling of colours in the foreground and background palette if they need to be used in both. The Archimedes probably runs in it's 16 colour mode, which allows it to have seemingly more colours on the screen.

#4 Pacmania is a good conversion on both. I mostly put it in as I saw an interview with the man who converted it to the Amiga himself were he said that he considers the Acorn Archimedes version to be better. But we could call this a tie: the Acorn has bigger ghosts and more colours, the Amiga has a bigger screen. Pick which you prefer.

#5 Lemmings is not scrolling smoother, true - but it still has a higher frame rate on the Archimedes regardless (just try exploding the lemmings).

As it happens, I've been to an Amiga event where the man who designed the Lemmings did a talk. This talk gave me the distinct impression the game was designed for the Amiga first and all other platforms followed later. It certainly was the lead platform and AFAIK was only released on the Atari ST several months after the Amiga release

#6 SWIV may have had legacy issues. But the reason it doesn't run at full frame rate on the Amiga have nothing to do with it being a shared ST/Amiga port. SWIV on the Amiga clearly uses the hardware quite well. It intelligently uses the sprite hardware, makes use of undocumented tricks and heavily uses the Blitter.

For instance, if you read the interview with Ronald Pieket Weeserik over at codetapper.com (http://codetapper.com/amiga/intervie...eket-weeserik/) you can get some interesting info about SWIV:
Quote:
Originally Posted by codetapper interview
...
Was the Amiga version of SWIV the lead version?

Yes, very much so.
...
Were 16 colours chosen so it could be easily ported to the Atari ST? Or was that all Ned wanted/needed to create an awesome atmosphere?

No, the choice to use only four bitplanes was made for memory and performance reasons.
...
The game runs at 25fps. Was that a deliberate design from the start or there was too much action happening to make it 50fps?
We wanted mayhem. We chose gameplay over frame rate.
...
Anyway, this isn't meant to become an A500 vs Archimedes thread. The only reason I started about it was to show the abilities of non-Amiga systems in 1987.
roondar is offline  
Old 18 July 2019, 02:51   #514
Bruce Abbott
Registered User

Bruce Abbott's Avatar
 
Join Date: Mar 2018
Location: Hastings, New Zealand
Posts: 233
Quote:
Originally Posted by grond View Post
The negative aspects of planar graphics are (just summing up):


They make the porting of the game harder and may even have made it seem impossible at the time.
The game wasn't even released when the A1200 was designed, so porting it was already impossible.

Quote:
They require a higher CPU grade to achieve the same fps that would have been possible with a lower CPU grade if a chunky mode had been available. This mainly raises the price of the Amiga hardware capable of running the game.
But redesigning the AGA chipset to work in chunky mode would also have raised the cost - for a feature that at the time was not considered to be important.

You talk about the 'negative aspects of planar graphics' but this is only relevant for one game genre that hadn't even been thought of when the AGA chipset was designed. For many other applications planar mode has advantages over chunky. With bitplanes you can:-

1. fine-tune memory usage to the number of colors wanted. Only need 128 colors? bitplane mode halves the amount of RAM required, both for the display and off-screen graphics. 64 colors? 1/4. That means you can have 4 times as much graphics with bitplances compared to chunky. And 64 colors is plenty enough for most games. On a machine with limited RAM that is a huge advantage.

2. Only write the bitplane(s) you need to for rendering particular colors. So for example a text editor could run on a 256 color screen but only have to scroll a single bitplane. Planes that are all one 'color' don't need to be stored, which can speed up rendering time as well as reducing memory requirements.

3. Have dual playfields which are truly independent and require virtually no CPU time to maintain. That makes it much easier to achieve silky smooth 50/60Hz animation with no tearing or jerking, without needing a fast CPU with high bandwidth graphic RAM access.

So bitplane graphics gives you far more flexibility to make best use of limited memory and processing power. It just isn't optimal for doing 256 color texture-mapped 3D. And of course all those advantages of bitplanes evaporate when you have a fast CPU with lots of RAM.

Which is where the PC comes in. When IBM developed VGA, processors were not very fast and RAM was expensive. SVGA required a minimum of 300k (in practice 512k) just to display a 640x480 size screen. Furthermore, RAM on the VGA card was dedicated to the display and could not be used for program code or sound samples etc. That's a waste of memory chips and makes the system more expensive, but...

By 1993 the PC Juggernaut had reached full speed, and manufacturers responded with ever-increasing CPU speeds and RAM densities. Integrated motherboard chipsets made systems much cheaper, and prices tumbled. It didn't matter anymore that VGA was memory hungry and CPU intensive, because RAM and CPU power was no longer a bottleneck. And games like Doom and Quake played to those strengths.

How could the Amiga possibly have competed against that? Well...

For a start, Motorola CPUs were too expensive and were falling behind in clock speeds (and as we all know, only the clock speed matters for marketing, not what the CPU does with it). So Commodore should have switched to using an Intel CPU.

Secondly, for fastest operation the CPU should only run from dedicated main RAM using the fastest possible access mode, with graphics RAM on a separate bus where it can be optimized for chunky operation - or better yet on a card that can be upgraded as better chips become available. All RAM should be on plugin modules that customers can upgrade to when bigger ones come out.

Thirdly, sampled sounds are nice but take up too much memory, and with the graphics chipset on a card they could no longer do sound and disk DMA. So onboard 'chip' RAM should have been reduced to (say) 640k, with a general purpose DMA system that worked with plugin sound cards which had a synthesizer chip as well as 16 bit PCM sound (2 channels should be enough), and a more standard floppy disk controller that could read and write 1.44MB disks. This DMA system would also make the latest IDE hard drives much faster.

With all those changes the Amiga might have had a chance of holding its head up amongst the ever-more powerful PCs coming out - except for one thing. With PC sales booming, why would developers be bothered porting their titles to an incompatible OS on a platform with a shrinking market share? Well that wouldn't be a problem if the Amiga could run Microsoft DOS and Windows (which would be easy now that they are using an Intel CPU). So the final thing Commodore should have done to ensure the Amiga's success was to make it 100% PC compatible. Doesn't matter that it can't run Amiga titles anymore - now that Doom is out nobody wants to play that old crap anyway!

Last edited by Bruce Abbott; 18 July 2019 at 02:57.
Bruce Abbott is offline  
Old 18 July 2019, 10:05   #515
swinkamor12
Registered User

 
Join Date: May 2017
Location: Vienna/Austria
Posts: 58
Few years ago I was playing with pc retro. I buy 386SX 25 MHz with SVGA card.

This 386SX 25 MHz is in integer operations almost two times slower than 68020 14 MHz in Amiga 1200.

Suprise!!!

DOOM on 386SX 25 MHz works as good as on Amiga 1200 with Blizzard 1230 Iv 50 MHz.

Why?

Because in 25 fps on 030 60% of time is wasted for c2p even with super fast Kalms routines.

DOOM will fly on 030 50 MHz if a1200 has chunky pixels.

68k assembler is more programmer friedly than 386.

It is obvious that Amiga 1200 with 68020, chunky pixels, fast ram and some optimisation in asm will be good enough to play DOOM.

Of course some amiga fanatics never admit, that Commodore bankrupt, because Amiga 1200 has too slow graphics.
swinkamor12 is offline  
Old 18 July 2019, 10:26   #516
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 634
Quote:
Originally Posted by roondar View Post
I see where you're coming from, but you do realise that anything over 2MB of total RAM was considered 'high end' in 1992, right? Putting 3MB of RAM in an entry level machine would've been epic but IMHO rather unlikely considering the market at the time.
Yes, sure. That's why it could have come out instead of the A500+ with a CPU card and a SIMM slot on the CPU card.

Budget config: 020, 1MB chip, empty SIMM slot
Upgrade options: add fastmem, add harddisk, swap 020 CPU card for 030

Midrange config: 16 MHz 030, 1MB chipmem, 1..2 MB of fastmem
Upgrade options: FPU, swap fastmem for more fastmem, add harddisk

Top config: 25 MHz 030, 2MB chipmem, 2MB fastmem, harddisk
Upgrade option: FPU

Later upgraded configs: 40 MHz 030, 50 MHz 030, 060 (unforeseeable at the time)...

The budget config would have only added the costs for the CPU card connector, the SIMM slot and the 020 when compared to the A500+. I don't think such an AGA+ machine in 1991 wouldn't have sold for a price that may have been the same as that of the A500 in 1987 or slightly higher.

If that had been the new base config of the Amiga, Doom wouldn't have killed the Amiga (assuming it did). Doom would have caused many Amigans to choose one of the upgrade options just like a few years later Halflife made many people buy a new graphics card just to play that game.


Quote:
Perhaps upgrade the CPU to 33MHz instead.
Definitely too expensive in 1991. The A3000 cost a fortune and only came with either 16 MHz or 25 MHz 030.


Quote:
However, after seeing examples of the game in action on AGA, I don't find the disadvantages as relevant as before. This is due to the result we eventually ended up with: the slow, planar based AGA chipset eventually got a version of Doom that ran better on a 68030@50MHz using c2p than on any of the 386 machines I saw - even the 386DX40 ones.
Yes, but the 50 MHz 030s became popular in the Amiga world when PC people all had 486s. I bought my Blizzard MK3 in 1994/1995 and it was a top configuration in the Amiga world for some time.

I remember that a colleague of that time (I did civil service for a year in 1994/1995 which gave me the little income to get the A1200 setup and which delimits the timeframe) got a Pentium 90 which was the top configuration for a few weeks before the Pentium 100 came out. The guy had saved quite some time and blew it all on that computer so I guess the typical PC customer still bought DX2-66s at the time. However, a DX2-66 runs circles around any 030 (and 040). The 50 MHz 030s were too late. 020s and 030s should have become the norm in the Amiga world when a 16 to 25 MHz 030 still was able to compete because all the PC world had were 386s.
grond is offline  
Old 18 July 2019, 10:31   #517
sandruzzo
Registered User
 
Join Date: Feb 2011
Location: Italy/Rome
Posts: 1,463
We've seen that cpu by cpu, Amiga 1200wins, but the trouble here, even it's hard to says, it is Aga: too slow, to late. I don't understand why in the world commodore made cpu access to chip mem at half rate!
sandruzzo is offline  
Old 18 July 2019, 10:42   #518
Foebane
Registered User
 
Join Date: Sep 2011
Location: Cardiff, UK
Age: 46
Posts: 1,753
Quote:
Originally Posted by Bruce Abbott View Post
How could the Amiga possibly have competed against that? Well...

For a start, Motorola CPUs were too expensive and were falling behind in clock speeds (and as we all know, only the clock speed matters for marketing, not what the CPU does with it). So Commodore should have switched to using an Intel CPU.
Wow, you make that sound so simple. The Motorola and Intel CPU architectures are completely different, and don't forget, the Amiga was designed FOR a Motorola 68000. Redesigning the entire architecture for Intel would've simply made the Amiga like a third-party graphics and sound add-on card for PC, and why bother when SoundBlaster and GUS and SVGA already exist? I don't think you understand the gravity of your statement. Unless you mean, the Amiga should've been Intel from the start? Same thing. Even in 1985, the PC had a market for add-on expansion cards.

Quote:
With all those changes the Amiga might have had a chance of holding its head up amongst the ever-more powerful PCs coming out - except for one thing. With PC sales booming, why would developers be bothered porting their titles to an incompatible OS on a platform with a shrinking market share? Well that wouldn't be a problem if the Amiga could run Microsoft DOS and Windows (which would be easy now that they are using an Intel CPU). So the final thing Commodore should have done to ensure the Amiga's success was to make it 100% PC compatible. Doesn't matter that it can't run Amiga titles anymore - now that Doom is out nobody wants to play that old crap anyway!
I certainly didn't. I'd had my fill of 2D platformers on the Amiga by the time Doom rolled around, and I was particularly miffed that so many of them were cutesy, too, designed for little children. Even the adult gory games had lost their edge. I'd say it was the third reason I switched. When I'd seen 3D games on Amiga with no detail and no texture-mapping whatsoever, and then I saw Doom, I just KNEW the future was with the PC. Other games like Descent nailed it for me.
Foebane is offline  
Old 18 July 2019, 10:53   #519
grond
Registered User

 
Join Date: Jun 2015
Location: Germany
Posts: 634
Quote:
Originally Posted by Bruce Abbott View Post
The game wasn't even released when the A1200 was designed, so porting it was already impossible.
You are saying that porting Doom to the A1200 was impossible because the A1200 was already on the market when Doom came out? In what world does that statement make sense?


Quote:
But redesigning the AGA chipset to work in chunky mode would also have raised the cost - for a feature that at the time was not considered to be important.
I never said that AGA should have been redesigned to add a chunky mode, I said that a chunky mode should have been added while they redesigned OCS and turned it into ECS. For all I know the added colours and colour resolutions where a feature planned for ECS but then the A3000 was rushed out the door (probably its development had been negelected earlier) with only a half baked solution. With only a little more foresight AGA as we know it would not have existed at all and something better than AGA would have come instead of ECS:

They could have scratched the two new bitplanes altogether and only have made it 6bit planar OCS + 8bit chunky (they could have turned EHB into a full 64 colour mode easily while at it). And if you insist that the world absolutely needed a 128 colour mode in addition to the 64 and 256 colour modes, then they could also have made 7 bitplanes and 1 chunky buffer which would have fit nicely with the OS structures that always had space for 8 bitplane pointers.


Quote:
You talk about the 'negative aspects of planar graphics' but this is only relevant for one game genre that hadn't even been thought of when the AGA chipset was designed. For many other applications planar mode has advantages over chunky. With bitplanes you can:-

1. fine-tune memory usage to the number of colors wanted. Only need 128 colors? bitplane mode halves the amount of RAM required, both for the display and off-screen graphics. 64 colors? 1/4.
Check your math. If you restrict yourself to 128 colours, you save 1/8th of the RAM required because you will still have 7 of the 8 bitplanes. Restricting yourself to 64 colours means you save 25% on chipmem.

Yes, that once was a big deal and the reason why planar was invented. It just didn't make ANY SENSE at all when you arrived at 8 bit palettes.


Quote:
2. Only write the bitplane(s) you need to for rendering particular colors. So for example a text editor could run on a 256 color screen but only have to scroll a single bitplane.
This is an academical argument at most. Depending on what colour that one colour is, you would have to scroll any number between 1 and 8 bitplanes because it is not like one bitplane represents one colour.

If you want to do monochrome text anyway, better do with a single bitplane but certainly not implement eight of them. Scrolling was free anyway, so there is no gain at all.

Furthermore, all this was already available in OCS. With 8 bit palettes it didn't make any sense at all. I never said they should have stopped supporting the OCS planar modes. They just shouldn't have added more planar modes.


Quote:
3. Have dual playfields which are truly independent and require virtually no CPU time to maintain. That makes it much easier to achieve silky smooth 50/60Hz animation with no tearing or jerking, without needing a fast CPU with high bandwidth graphic RAM access.
Already available in 1985. What's the point? BTW, there is nothing that would stop you from making chunky dual-playfield modes with two playfields of 16 colours each stored in two separate buffers of nibbles ("half bytes").

There is nothing that would make a chunky buffer less "scrollable" than a planar buffer.


Quote:
So bitplane graphics gives you far more flexibility to make best use of limited memory and processing power. It just isn't optimal for doing 256 color texture-mapped 3D. And of course all those advantages of bitplanes evaporate when you have a fast CPU with lots of RAM.
No. The point is that at 8 bits per pixel planar modes have ONLY disadvantages and NO advantages. In fact, they are even harder to implement than chunky modes because you need to collect data from eight different places in memory (no data streaming, eight different DMA addresses counter-adders required, temporal storage until the data is complete...) to then just reconstruct the pixels. No engineer in their right mind would implement an eight bit planar mode.


Quote:
Which is where the PC comes in. When IBM developed VGA, processors were not very fast and RAM was expensive. SVGA required a minimum of 300k (in practice 512k) just to display a 640x480 size screen.
Nonesense again. For this very reason the PC cards also supported some planar modes and also a 16 colour nibble-chunky mode. Often overlooked but nonetheless true. But this doesn't really matter to this discussion. I'm perfectly happy with the 6 bitplane planar modes of OCS.


Quote:
So the final thing Commodore should have done to ensure the Amiga's success was to make it 100% PC compatible. Doesn't matter that it can't run Amiga titles anymore - now that Doom is out nobody wants to play that old crap anyway!
Lot's of blabla cut. This line of argumentation always shows up in these discussions: if Commodore had done things right, the Amiga would have been a PC.

No, this is grossly oversimplifying things. Macs did fairly well without Intel CPUs for another decade, PC graphics cards could be incorporated into other systems without turning those systems into PCs, price drops in RAM and harddisks would have benefitted all competing systems at the same time, OSs in general got time to become abstracted from hardware and even CPUs. And so on.

There is absolutely no technical reason why the Amiga couldn't have gone a similar path without automatically becoming a PC. In fact, the PC would have become something different and better if the Amiga had had more success than it had because of Commodore's poor use of its potential.
grond is offline  
Old 18 July 2019, 11:06   #520
Foebane
Registered User
 
Join Date: Sep 2011
Location: Cardiff, UK
Age: 46
Posts: 1,753
All this talk of Chunky Mode on Amiga AGA: How FEASIBLE and EASY is it to implement? Maybe some hardcore engineers here can do such a project? Can't be that difficult.
Foebane is offline  
 


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Similar Threads
Thread Thread Starter Forum Replies Last Post
A1200 RF module removal pics + A1200 chips overview eXeler0 Hardware pics 2 08 March 2017 00:09
Sale - 2 auctions: A1200 mobo + flickerfixer & A1200 tower case w/ kit blakespot MarketPlace 0 27 August 2015 18:50
For Sale - A1200/A1000/IndiAGA MkII/A1200 Trapdoor Ram & Other Goodies! fitzsteve MarketPlace 1 11 December 2012 10:32
Trading A1200 030 acc and A1200 indivision for Amiga stuff 8bitbubsy MarketPlace 17 14 December 2009 21:50
Trade Mac g3 300/400 or A1200 for an A1200 accellerator BiL0 MarketPlace 0 07 June 2006 17:41

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +2. The time now is 16:15.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.
Page generated in 0.11752 seconds with 16 queries