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
36 crawler(s) on-line.
 25 guest(s) on-line.
 0 member(s) on-line.



You are an anonymous user.
Register Now!
 zipper:  1 hr ago
 DiscreetFX:  1 hr 15 mins ago
 Kronos:  1 hr 24 mins ago
 clint:  1 hr 25 mins ago
 pixie:  1 hr 27 mins ago
 Karlos:  1 hr 33 mins ago
 Futaura:  2 hrs 5 mins ago
 ppcamiga1:  2 hrs 43 mins ago
 agami:  3 hrs 23 mins ago
 Maijestro:  3 hrs 27 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
Hammer 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 13-Oct-2022 5:53:36
#301 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@bhabbott

Quote:

This is total crap. The 68000 was much more powerful than other CPUs of the day. Compare it to the 8088 that IBM chose for the PC (they considered the 68000 but it was too expensive, not available in large quantities, and not so easy to port 8080 code to). The 68000 was chosen by manufacturers of workstations who saw its potential for a mainframe-like OS.

That's BS in the context of Amiga's existence.

The timeline,

1. 80286 with integrated MMU was available in February 1982.

Both 68000 and 80286 have 16-bit ALUs but 68000 has a 32-bit instruction set while 80286 has an integrated MMU.

2. In September 1982, Hi-Toro (later Amiga Corporation) was created.

For businesses that require multi-user requirements, any 286 PC clone can run Xenix. Microsoft's Xenix X86 was the best-selling licensed Unix.

Until the MC68451 MMU addon for the 68010 chip, many 68K Unix vendors have in-house custom MMUs.

68020's 68851 MMU addon is not backward compatible with 68451 LOL

3. 32-bit 80386 was released in October 1985.

4. 68030 was released in 1987 which is effectively 32-bit 68020 with integrated 68851 MMU and minor improvements. 68030 and 68020 have similar IPC.


Quote:

All irrelevant.

Look in the mirror, hypocrite. You made your 68000 vs 8088 arguments outside of Amiga's 1982 existence.

Reminder
8088 is a cost reduced 8086.
68008 is a cost reduced 68000.

MS-DOS X86 PC clones have the faster 8086 CPU instead of IBM's cost-reduced 8088.

Last edited by Hammer on 13-Oct-2022 at 05:57 AM.
Last edited by Hammer on 13-Oct-2022 at 05:55 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  
cdimauro 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 13-Oct-2022 6:16:06
#302 ]
Elite Member
Joined: 29-Oct-2012
Posts: 3104
From: Germany

@Karlos

Quote:

Karlos wrote:
@cdimauro

I do understand what HAM is and how it works but for the purposes of the C2P method being used, the base palette is not something you set set up for artistic purposes. The process here is completely unlike the processes normally involved for displaying a 24 bit image on a HAM screen at the same resolution.

The graphics in the demo are rendered in 24 bit RGB and that is in turn converted to the best HAM8 representation. 4 consecutive superhires HAM8 output pixels are used per 1 lores RGB input. There's no fringing effects to worry about here, you are essentially using the superhires pixels as subpixels.

Yes, this I've already understood.

I've only highlighted that the method is fast because he doesn't calculate an optimized palette. Then graphics artifacts come. That' all.
Quote:
I downloaded the repo and tested it in UAE where it's super smooth. I suspect you'll need an 060 to get useable frame rates otherwise.

Probably.

P.S. No time now to answer to other posts. It's work time...

 Status: Offline
Profile     Report this post  
Hammer 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 13-Oct-2022 6:24:32
#303 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@bhabbott

Quote:

bhabbott wrote:

Ah yes, the S3 Trio, one of the most over-hyped and disappointing 'graphic accelerator' cards for the PC.


https://smugnplay.com/blog/gpu-grudge-match-94-ati-vs-s3

S3 Trio64 (Diamond Stealth 64 DRAM) beating ATi mach64 GX (ATi Graphics Pro Turbo)

Doom (Max Details)
S3 Trio64 : 55.7 fps
ATi mach64 GX : 52.7 fps

http://www.vgamuseum.info/images/vlask/bench/quake320.png
Quake 320x240 benchmark

1997
NVIDIA RIVA 128 scored 199. My university fleet PCs have this video card along with 3DLab Premedia 2 and Matrox Millennium II.

3DLab Premedia 2 scored 178.

I jumped ship toward NVIDIA TNT due to the university's NVIDIA RIVA 128 example.

1996
S3 Trio 64 UV+ scored 144.8 (my 1996 Pentium 150 was fitted with this video card)
Cirrus Logic 5446 scored 141.8

1995
S3 Trio 64 V+ scored 138.7 fps
Cirrus Logic 5434 scored 108.8
Opti 82C264 scored 108.3

1994
ET4000W32P scored 135.4

S3 Virge 3D is disappointing, not S3 Trio 64. 3DFX Voodoo 2 still needs a fast SVGA 2D card.

Your memory sucks.

_________________
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 13-Oct-2022 6:43:20
#304 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@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.

Quote:

Again: irrelevant.

Your argument is irrelevant.

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.

My A1200's stock 68EC020 with 16 Mhz marking.

Last edited by Hammer on 13-Oct-2022 at 06:44 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 13-Oct-2022 7:09:29
#305 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@matthey

Quote:

The 68EC020 had fewer address line pins (16MiB max) resulting in a cost savings which was nice. The 68EC030 removed the MMU so the major difference was the data cache, cache bursts and higher clock ratings. The 68020 had chips rated up to 33MHz but the parts with high ratings were often more expensive and/or in limited supply which may not be acceptable for high volume production. The 68030 went up to 50MHz while the 68EC030 stopped at 40MHz. Maybe a 68EC020@28MHz was ok and a 68EC030@28MHz was likely possible although it may have been more advisable to use 33MHz rated parts in cramped Amiga wedge cases (28.36MHz PAL, 28.64MHz NTSC). A 68EC030@28MHz should have competed nicely against the 386DX@33MHz at the low end without breaking the bank.

My main reason for the 28 Mhz target is due to the existing 28 Mhz crystal on A1200's motherboard.

https://bigbookofamigahardware.com/bboah/product.aspx?id=145
TRA1200 has 68020 25 Mhz rated @ 28 Mhz and 32-bit Fast RAM.



My A3000's 030 @ 25 Mhz runs Wing Commodore ECS at a pretty good frame rate but needs a higher color display.







_________________
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 13-Oct-2022 7:37:17
#306 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

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

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

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

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

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.

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


_________________
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 13-Oct-2022 8:37:50
#307 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@agami

Quote:

agami wrote:
@cdimauro

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.


Competitive hardware matters to a certain extent e.g. Xbox One's mediocre sales vs Xbox Series S/X tracking slightly above Xbox 360's sales curve.

PS4 is the superior hardware when compared to Xbox One.

Xbox Series X (12 TFLOPS GPU) is more than a match against PS5 (10 TFLOPS GPU).

Digital Foundry's pixel-counting social media storm has some influence on Xbox One vs PS4.


Quote:

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

Microsoft is the leader of the PC clone army. Microsoft-defined ACPI tables (part of the Designed For Windows initiative) effectively forced Linux to adopt it.

Microsoft told Intel to support AMD64/X86-64 since Intel's IA-64 is a failure.

Xbox generations have been defining PC GPU's feature set since Xbox 360.

PC DirectX8 ~= Xbox.

PC DirectX10 ~= Xbox 360's DX9X.

PC DirectX11 programmable tessellation ~= Xbox 360's DX9X fix function tessellation.

PC DirectX12 Hardware Feature Level 12_0 ~= XBox One (GCN 2.0)

PC DirectX12 Hardware Feature Level 12_1 ~= Xbox 360's ROV ROPS feature. Xenia Xbox 360 emulator recommended DirectX12 Feature Level 12_1's ROV ROPS for feature match and improved performance efficiency. NVIDIA Maxwell Gen 2, AMD VEGA GCN 5, and Intel SkyLake Gen 9 IGP era.

PC DirectX12 Hardware Feature Level 12_2 ~= Xbox Series S/X. NVIDIA RTX Turing, AMD RDNA 2, and Intel ARC era.






Last edited by Hammer on 13-Oct-2022 at 08:40 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 13-Oct-2022 8:54:35
#308 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@Karlos

Quote:

Karlos wrote:
@bhabbott

Correct, it's an RGB raycaster. The palette is completely irrelevant. He's using Kalm's HAM8 C2P routines. If you look carefully you'll also note he's doing MIP mapping too. As you approach the walls the textures change. That's what the numbers indicate.

I do wonder which CPU it's running on, but the code is all in github at least.


From https://www.youtube.com/watch?v=yVAxSXM_5pU
Mateusz Staniszew has access to Vampire V1200 and Warp1260.

This is like throwing Pentium MMX and Pentium class CPUs at the problem.

At least AGA is not bottlenecked like IBM VGA with Athlon XP 2200+.








_________________
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  
bhabbott 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 13-Oct-2022 9:37:35
#309 ]
Regular Member
Joined: 6-Jun-2018
Posts: 251
From: Aotearoa

@Hammer

Quote:

Hammer wrote:
@bhabbott

Quote:

This is total crap. The 68000 was much more powerful than other CPUs of the day. Compare it to the 8088 that IBM chose for the PC (they considered the 68000 but it was too expensive, not available in large quantities, and not so easy to port 8080 code to). The 68000 was chosen by manufacturers of workstations who saw its potential for a mainframe-like OS.

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

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

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.

Quote:
For businesses that require multi-user requirements, any 286 PC clone can run Xenix. Microsoft's Xenix X86 was the best-selling licensed Unix.

You talk about 'in the context of Amiga's existence', then post this irrelevant crap. The Amiga wasn't a multi-user computer, and the very idea of running Unix on a stock A1000 is laughable. Xenix might have been the 'best selling licensed Unix' for a while, but that's not saying much. It was still practically unheard of in the PC world.

Quote:
Until the MC68451 MMU addon for the 68010 chip, many 68K Unix vendors have in-house custom MMUs.

That's true for certain definitions of 'many'. And not a big deal. Custom MMUs were standard practice for machines that needed them (the Amiga didn't).

Quote:
68020's 68851 MMU addon is not backward compatible with 68451 LOL

So?

Quote:
3. 32-bit 80386 was released in October 1985.

Even more off topic.

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. When the Amiga's design was started no 80286 based machine existed. However the machine they used to develop the Amiga was 68000 based, so it made sense to use the same CPU.

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. It was the opposite for 68k, with few machines using the 68008.

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 13-Oct-2022 18:15:45
#310 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@Hammer

Quote:

Hammer wrote:

From https://www.youtube.com/watch?v=yVAxSXM_5pU
Mateusz Staniszew has access to Vampire V1200 and Warp1260.

This is like throwing Pentium MMX and Pentium class CPUs at the problem.


I don't see where MMX comparison makes any sense here - no 68080 AMMX SIMD operations are used in the code.

_________________
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 13-Oct-2022 18:46:23
#311 ]
Elite Member
Joined: 11-Aug-2005
Posts: 6915
From: UK

@Thread

Bit bored of this thread now!

I love the 68060 and the AGA chipset but RTG is better for productivity.

The PC world never gave AGA or Amiga RTG a 2nd glance even if they noticed it at all!

The Amiga would have had to become an "Amiga on a card" to plug into PCs by 1995. Kinda like what Mick Tinker had planned!
Maybe that was the key to survival; the Amiga being a kind of VideoToaster Card for the PC!

DISCUSSION

Quote:
I exchanged some emails with Mick Tinker. Basically he gave up as the hardware vas virtually done but there was no s/w from Paul Nolan (of Photogenics fame) to do the interface with PC hardware (the card uses the PC's VGA card as a display).

Personally I think it would have been a great piece of kit. But times move on and now with a fast PC and WinUAE you can get 040 speeds so whats the point (OK, 060 maybe).... Maybe if it had a Coldfire and patches like Oxypatcher etc it would be interesting....

Still, if someone made one I'd rather have real hardware rather than emulation inside my PC...

Last edited by BigD on 13-Oct-2022 at 06:52 PM.
Last edited by BigD on 13-Oct-2022 at 06:47 PM.

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

 Status: Offline
Profile     Report this post  
bhabbott 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 13-Oct-2022 19:49:24
#312 ]
Regular Member
Joined: 6-Jun-2018
Posts: 251
From: Aotearoa

@BigD

Quote:

BigD wrote:

The PC world never gave AGA or Amiga RTG a 2nd glance even if they noticed it at all!

The PC World never gave anything that wasn't a PC a second glance.

Quote:
The Amiga would have had to become an "Amiga on a card" to plug into PCs by 1995.

Why?

I helped my friend install an entire A1200 motherboard and PC motherboard inside a tower case. The two machines were linked together via the Siamese System so he could use the same keyboard and mouse for both machines. Worked brilliantly!

Quote:
Quote:
I'd rather have real hardware rather than emulation inside my PC...

I'd rather have real hardware inside my Amiga.

 Status: Offline
Profile     Report this post  
matthey 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 13-Oct-2022 22:51:37
#313 ]
Super Member
Joined: 14-Mar-2007
Posts: 1684
From: Kansas

bhabbott 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.


The AmigaOS didn't use the 68k MMU because it didn't exist. Yes, the 68k 32 bit flat memory model, 16 GP registers, support for many integer datatype sizes and good code density were more important. Jay Miner wanted the MMU and AmigaOS Exec modified to support it as I recall. I believe it is mentioned in the following video.

The Acquisition of Amiga From Commodore's Perspective & Custom Chips w/ Dave Haynie and Andy Finkel
https://www.youtube.com/watch?v=LSAD31nD3og

CBM was interested in a 68k MMU although they likely wanted to run Unix like OSs on higher end Amigas. CBM started designing a custom MMU for the 68020 and came out with a 68020 accelerator with separate MMU chip. The AmigaOS used the MMU for debugging tools and hardware bug fixes.

bhabbott Quote:

The Amiga wasn't a multi-user computer, and the very idea of running Unix on a stock A1000 is laughable. Xenix might have been the 'best selling licensed Unix' for a while, but that's not saying much. It was still practically unheard of in the PC world.


CBM was developing the Commodore 900 with Z-8000 at the same time as the Amiga which used the Unix like Coherent OS. Coherent was available on the 68k and had a small footprint. I would be surprised if it would run on an Amiga 1000 without an expensive memory upgrade though. CBM was developing Amix for the Amiga at least as early as the Amiga 68020 card with MMU chip which later matured into the Amiga 3000UX. As I recall, Dave Haynie mentions that at least 2MiB of memory were needed for Amix in the video above. That much memory in the mid '80s may have been more expensive than adding an upgraded 68k CPU and MMU which were also needed.

bhabbott Quote:

That's true for certain definitions of 'many'. And not a big deal. Custom MMUs were standard practice for machines that needed them (the Amiga didn't).


CBM likely didn't want a MMU for low end Amigas due to increased expense. There were also compatibility issues with adding MMU support to the AmigaOS and a performance loss.

 Status: Offline
Profile     Report this post  
Karlos 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 14-Oct-2022 0:04:31
#314 ]
Elite Member
Joined: 24-Aug-2003
Posts: 3565
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition!

@Hammer

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


Woah woah woah. Back up a bit. "Has no problem"? Both parents in full time work? Present tense?

Assuming for a moment you aren't actually a minor, aren't you independent enough to buy your own computers by now? Do they buy your clothes still too?

_________________
Doing stupid things for fun...

 Status: Offline
Profile     Report this post  
Hammer 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 14-Oct-2022 0:27:39
#315 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@bhabbott

Quote:

'Available', but the PC/AT didn't get it until 1984, running at 6MHz with 1 wait state (equivalent to 4MHz). 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 (actually slower since the 80286 has fewer registers and they are only 16 bit).

The same thing can be said for X86-64 (AMD64, K8 was released on April 2003) since the Windows XP X64 edition was released on April 25, 2005. The R&D road map is a factor.

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

iSBC 286/10 computer was running Microsoft's Xenix.

Timeline
1984, IBM PC AT has 6 Mhz 286.
1985, COMPAQ DESKPRO 286 has 8 Mhz 286.
1986, Dell (PC's Limited)'s AT clone had 16 MHz 286 (via AMD 80L286), and Compaq DeskPro 386 had 16 Mhz 386.

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.

Timeline
80286 @ 10Mhz, July 1985.
80286 @ 12Mhz, July 1985.

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

During the early 1980s, 68000's road map scaling into 32-bits and linear memory address models was superior.

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.

Too bad Motorola didn't glue some 1.0 IPC DSP extensions into 68K i.e. divide the 68K instruction set into fast and slow paths. Over time, more 68K instructions are moved over to the fast path.

Many 68K Unix vendors were heading in different RISC instruction set directions after 68020/68851 e.g. around the Mid 1980s (1987) Sun-3/80 running on a 20 MHz Motorola 68030 delivered about 3 MIPS whereas the first SPARC-based Sun-4/260 with a 16 MHz SPARC delivered 10 MIPS.


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.


Intel was large enough to sustain X86, iAPX 432, and i860 R&D developments within a certain time frame while Motorola was late with 68040. 80486 was released in 1989 while 68040 was released in 1990.

X86 was sustained by the larger X86 clone PC market and it's not dependent on a single microcomputer platform vendor like single vendor platforms like Apple or Commodore.

NextGen 5x86 (1994) are early X86 CISC-to-RISC CPU before Intel's Pentium Pro (1995) while 68060 (1994)'s 68K future was EOL. 68K EOL from the high-performance CPU markets was announced around the same time as the AIM (PowerPC) alliance's 1991 creation.

Aided by DEC's and Intel's StrongARM (ARM v4), ARM's low cost, low power consumption, and good multi-vendor adoption rate enable ARM to survive the RISC instruction set war.

Quote:

An MMU that only worked in protected mode, which wasn't compatible with DOS. Fail! (but then Intel never intended it to be used in the PC).

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.

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

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

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.

Linux was designed on i386 with MMU.

PC world's 32-bit i386 MMU longer install-base build-up time with modular VGA addon standard also benefited Doom while the entire 32-bit 68020/68030 Amiga OCS/ECS install base was thrown in the bin when AGA arrived.

PC world was building a large install base with MMU capability for OS/2, Linux i386, BSD i386, and Windows NT i386.

PC world's long build-up with TPM 2.0 install base tactics was used before pulling Windows 11 TPM trigger.


Last edited by Hammer on 14-Oct-2022 at 05:18 AM.
Last edited by Hammer on 14-Oct-2022 at 04:04 AM.
Last edited by Hammer on 14-Oct-2022 at 01:09 AM.
Last edited by Hammer on 14-Oct-2022 at 01:06 AM.
Last edited by Hammer on 14-Oct-2022 at 01:00 AM.
Last edited by Hammer on 14-Oct-2022 at 12:57 AM.
Last edited by Hammer on 14-Oct-2022 at 12:51 AM.
Last edited by Hammer on 14-Oct-2022 at 12:43 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 14-Oct-2022 0:45:26
#316 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@Karlos

Quote:

Karlos wrote:
@Hammer

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


Woah woah woah. Back up a bit. "Has no problem"? Both parents in full time work? Present tense?

Assuming for a moment you aren't actually a minor, aren't you independent enough to buy your own computers by now? Do they buy your clothes still too?


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.

Last edited by Hammer on 14-Oct-2022 at 12:46 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  
matthey 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 14-Oct-2022 1:33:20
#317 ]
Super Member
Joined: 14-Mar-2007
Posts: 1684
From: Kansas

BigD Quote:

Bit bored of this thread now!

I love the 68060 and the AGA chipset but RTG is better for productivity.

The PC world never gave AGA or Amiga RTG a 2nd glance even if they noticed it at all!


That's because the Amiga chipset was low end by the time AGA came out. The original Amiga chipset drew interest in 1985 but software was slow to come out for it and CBM had a poor reputation for higher end and professional use computers. The Amiga became popular as more of a toy computer after cost reductions and away from local markets where CBM's poor reputation was a handicap.

The Amiga chipset performance could have been upgraded to be better than most PC graphics options in the late '80s and '90s. The integrated graphics of the Amiga give a performance and cost advantage but CBM chose to lever this advantage by making it cheaper instead of higher performance. They wanted another C64 from the Amiga and cost reduction is all they had to do for the C64. Cost reductions worked great for Amiga sales at first but they needed to upgrade it faster to keep it from becoming obsolete.

BigD Quote:

The Amiga would have had to become an "Amiga on a card" to plug into PCs by 1995. Kinda like what Mick Tinker had planned!
Maybe that was the key to survival; the Amiga being a kind of VideoToaster Card for the PC!

DISCUSSION

Quote:
I exchanged some emails with Mick Tinker. Basically he gave up as the hardware was virtually done but there was no s/w from Paul Nolan (of Photogenics fame) to do the interface with PC hardware (the card uses the PC's VGA card as a display).

Personally I think it would have been a great piece of kit. But times move on and now with a fast PC and WinUAE you can get 040 speeds so whats the point (OK, 060 maybe).... Maybe if it had a Coldfire and patches like Oxypatcher etc it would be interesting....

Still, if someone made one I'd rather have real hardware rather than emulation inside my PC...


I believe there is a mix up of 2 Index Information products.

Inside Out (unreleased Amiga on a PCI card)
https://bigbookofamigahardware.com/bboah/product.aspx?id=41

BoXeR (unreleased Amiga compatible ATX motherboard)
https://bigbookofamigahardware.com/bboah/product.aspx?id=40

There was also the 3-pack which was a case that allowed 3 ATX motherboards the BoXeR may have been designed to fit in with Paul Nolan providing the Siamese software. That was the big marketing push but the BoXeR motherboard should be capable of standalone operations including providing Amiga chipset output compatible with AGA. The Amiga chipset was improved with some specs from CBM AA+ (and renamed AA+) including better chip memory performance, buffered I/O, etc. It also removed the chip memory limit and integrated the Amiga chipset into one chip already in the late '90s using FPGA technology (ASIC may have been planned for production or cost reduction). The 4 active PCI slots of the later board design would have eliminated the PCI to Zorro bottleneck of other solutions. It supported a 68040/68060 CPU socket and a CPU slot. It would have been a big upgrade for 68k Amigas which would likely be sought after still today. It's an amazing design and some of the FPGA VHDL code could probably reused today if it was available.

Boxer likely partially suffered from the same problem as the Natami. They were both professional level projects but there is no 68k CPU available which has a good enough performance/price to risk producing it. They both probably should have been produced anyway but a 68060@100MHz at the high end is going to limit sales. Even the 68060 is in short supply while there are 4 different 68060 accelerator cards available today which is crazy. FPGA technology can provide a chipset many times higher performance than AGA in a single FPGA but it can only help develop a 68k CPU that needs to be turned into an ASIC to be competitive.

 Status: Offline
Profile     Report this post  
Hammer 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 14-Oct-2022 3:57:24
#318 ]
Elite Member
Joined: 9-Mar-2003
Posts: 4656
From: Australia

@Karlos

Quote:

Karlos wrote:
@Hammer

Quote:

Hammer wrote:

From https://www.youtube.com/watch?v=yVAxSXM_5pU
Mateusz Staniszew has access to Vampire V1200 and Warp1260.

This is like throwing Pentium MMX and Pentium class CPUs at the problem.


I don't see where MMX comparison makes any sense here - no 68080 AMMX SIMD operations are used in the code.


I was referring to the general CPU performance.

For Quake demo3 benchmark example with PiStorm/RPI 3a https://www.youtube.com/watch?v=gJwdColUxUs
PiStorm/RPI 3a delivered performance comparable to Pentium II 266/300 results.

68060 rev 6 @ 100 Mhz is about classic Pentium 83 Mhz Overdrive.

AC68080 falls between PiStorm/RPI 3a (~Pentium II 266/300) and 68060 rev 6 @ 100 Mhz (~Pentium 83 Mhz Overdrive).

Minus the MMU and FP80 issues, AC68080's microarchitecture is superior when compared to 68060 rev 6.


Last edited by Hammer on 14-Oct-2022 at 03:58 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  
Hypex 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 14-Oct-2022 4:55:11
#319 ]
Elite Member
Joined: 6-May-2007
Posts: 10970
From: Greensborough, Australia

@Gunnar

Quote:
A4000 was supposed to support faster IDE modes
A4000 Register $DD1020 is for enable it.
The Vampire has this register working and does over 20MB/sec


Must be a custom register in A4000 as transfer speed was set via commands.

And at 20MB/s the Vampire is working at greater than PIO4 speeds. Is it using a DMA mode?

 Status: Offline
Profile     Report this post  
bhabbott 
Re: How good or bad was the AGA chipset in 1992/1993.
Posted on 14-Oct-2022 5:44:53
#320 ]
Regular Member
Joined: 6-Jun-2018
Posts: 251
From: Aotearoa

@matthey

Quote:

matthey wrote:

Quote:
The PC world never gave AGA or Amiga RTG a 2nd glance even if they noticed it at all!


That's because the Amiga chipset was low end by the time AGA came out. The original Amiga chipset drew interest in 1985 but

The Amiga wasn't IBM compatible, and so the PC world didn't give it a second thought.

Those of us who were into various home computers during the 80's may not appreciate how insular the PC world was. Where did PC users get information about other platforms? PC World, PC Magazine, Computer Shopper? Unlikely. Certainly not Amiga World or Amazing Computing, and probably not Compute!. Byte was probably the most inclusive computer magazine of the time, but after the PC was released in 1981 it rapidly became more PC-centric. The August 1985 issue's cover article featured the A1000. After that, nothing - like most of the other 'exotic' architectures they covered. I bought every issue of Byte from the late 70's to the mid 90's when it became just another PC magazine. The transition was disappointing, but inevitable. Before 1981 its advertising pages were filled with vendors trying to attract business users to the various systems they produced, afterwards it was the same - except every system was a PC.

Quote:
The Amiga chipset performance could have been upgraded to be better than most PC graphics options in the late '80s and '90s. The integrated graphics of the Amiga give a performance and cost advantage but CBM chose to lever this advantage by making it cheaper instead of higher performance. They wanted another C64 from the Amiga and cost reduction is all they had to do for the C64. Cost reductions worked great for Amiga sales at first but they needed to upgrade it faster to keep it from becoming obsolete.

This is nonsense. The Amiga was about making it cheaper even before Commodore acquired it. The integrated chipset was key to this goal (and not conducive to providing the highest possible future performance). But Commodore didn't just concentrate on the low end. The A2000 provided for similar open-ended upgrades as the PC. And as with the PC, 3rd party manufacturers were soon producing cards with performance rivaling or exceeding typical PC specs. Then they produced their first major improvement over the A1000, the A3000. The onboard chipset was much the same for compatibility, but it was intended to take high-end Zorro III graphics cards.

The A3000 was expensive, and high performance cards for it and other 'big box' Amigas were also expensive. This should counter the complaint of Commodore 'choosing to make it cheaper instead', but of course it doesn't because Amiga fans wanted it both ways - they expected the Amiga to be both cheaper and better than a typical PC.

I love how Amiga fans are still arguing about Commodore's supposed misdeeds in not producing the machine of their dreams. But they did manage to produce some wonderful machines than we still enjoy using today. Are they perfect? Of course not. But they could have been so much worse. How much worse? Just look at all the other home computers designed in the 80's.

 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