Your support is needed and is appreciated as Amigaworld.net is primarily dependent upon the support of its users.
|
|
|
|
Poster | Thread | agami
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 21-Oct-2022 1:12:54
| | [ #401 ] |
| |
|
Super Member |
Joined: 30-Jun-2008 Posts: 1840
From: Melbourne, Australia | | |
|
| @DOOM
I thought we were discussing how good the AGA chipset was, an trying to be as objective as we can. Then what is the reason for constantly using Doom as the benchmark?
It has been well established that the lack of an official Doom port to Amiga is in large part due to the lack of critical mass in minimal spec. Sadly in 1993, the lowest and widest common denominator for the Amiga "platform" was still an OCS A500 with 1MB RAM, which has nothing to do specifically with AGA shortcomings as compared to VGA.
Yes the G in AGA is Graphics, but there is more to graphics than playing Doom. The Amiga, enabled by its chipset, goes beyond the mere tech spec of MIPS and colours on screen. Even in 1994, none of my friends with a 386 could create with their systems what I could with my expanded A1200HD at the same budget.
And I was not starved for gaming entertainment. I had plenty of fun in 93/94 playing Alien Breed II, Syndicate, Banshee, Pinball Fantasies, Superfrog, Lemmings 2, Skidmarks, Flashback, to name a few I recall fondly but there were others. They weren't all optimised for AGA, but I played them on the A1200 before Commodore's demise.
While I had been using 286, 386 and 486 PCs at work, it wasn't until late 1995 that I finally got a Win 95 Pentium 90 PC at home. I played the then popular Doom II twice. Once in single-player mode, and once because a friend wanted to try playing over a modem. It really wasn't until Quake in 1996 and GLQuake in 1997 that I got into FPS games.
Last edited by agami on 21-Oct-2022 at 01:21 AM. Last edited by agami on 21-Oct-2022 at 01:19 AM.
_________________ All the way, with 68k |
| Status: Offline |
| | cdimauro
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 21-Oct-2022 4:58:30
| | [ #402 ] |
| |
|
Elite Member |
Joined: 29-Oct-2012 Posts: 4111
From: Germany | | |
|
| @bhabbott
Quote:
bhabbott wrote: @cdimauro
Quote:
cdimauro wrote:
The CPU isn't a 68000 (AGA machines have at least a 68020), so it has always access to those slots (I mean: it can potentially access all of them). |
So these people are wrong?
A1200 CPU memory access speed (with DMA) Quote:
Toni Wilen WinUAE developer
Blitter timing has not changed. Chipset timing has not changed. Main difference is bus width, CPU has 32-bit access to chip ram... timing is same, chipset still needs 2 clocks to complete single CPU access just like OCS/ECS. |
Quote:
Kalms
Whenever the CPU - regardless of if it's a 68000/020/030/040/060 performs a write operation, it will spend 2 buscycles communicating with the custom chipset. This puts the peak throughput for the CPU at ~1.77M write operations in a frame. A 68020+ in an AGA system would perform 32-bit writes, thus, ~7.09MB/s write throughput.
The CPU will not be able to utilize more than every 2nd buscycle for the actual data transfer. That leaves the 1st buscycle in each buscycle pair free for other things - system DMA, bitplane DMA, blitter DMA. |
|
So, the they decided to keep the same allocation slots mechanism and castrated the CPU access to the RAM forcing it to wait two additional cycles before the required two for the concrete access.
Beautiful! Only the so smart Commodore engineers could have done it: making it worse that the prehistoric 16-bit ISA bus (since it runs at 8Mhz at least). Quote:
bhabbott wrote: @Hammer
Quote:
Hammer wrote:
For a new computer in Q3 1992, A3000's asking price is not competitive when compared to the 386DX-25 PC clone. |
Equating a crappy PC clone with slow 16 bit ISA bus to the A3000 with its 32 bit Zorro III bus and 32 bit Chip RAM? You can't be serious.
Quote:
386DX PC clone with an ISA slot can be upgraded with a fast SVGA card. |
Oh, you are serious! |
Indeed: he was serious. Absolutely
Whereas the brilliant Commodore engineers made the same prank as above also with the ECS chipset. So, even if you have a 32-bit access to the Chip mem and could do it in two cycles, the processor is forced to wait 2 additional cycles before the real access can happen. Quote:
I upgraded my A3000 with a Cyberstorm 060 at 50MHz and 32MB of local Fast RAM, and a Picasso-II RTG card with 2MB RAM (which is only Zorro II, but still pretty snappy due the Cirrus Logic GD5428's 32 bit blitter). |
Where you able to use its Blitter?
@agami
Quote:
agami wrote: @DOOM
I thought we were discussing how good the AGA chipset was, an trying to be as objective as we can. Then what is the reason for constantly using Doom as the benchmark? |
Because is the most used software as a benchmark, since it's also ported to the vast majority of hardware devices (even on a pregnancy test, recently). |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 21-Oct-2022 7:40:29
| | [ #403 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @Kronos
Quote:
Kronos wrote: @Hammer
Quote:
Hammer wrote:
Not with IBM VGA. It must be a fast 16-bit VGA clone.
|
And what does that mean to the price of the fish???
In 1993 16bit VGA with 0.5 or 1MB were dirt cheap so that was the competition.
The OCS had 2 major selling points:
- using RAM cycles that would otherwise be wasted with minimum impact on the CPU performance. --- those weren't a thing with an 020 and running the RAM at half the CPU's speed meant that the CPU performance was crippled even before AGA drew a single pixel
- it could do thing useful for 80s style 2D games with little to no CPU needed --- that type of gaming was going out of fashin fast by 1993
If it wasn't for backwards compatibility an A1200 with 1MB proper fast and fully memory mapped 1MB 16Bit VGA would have been much better at the early 3D games and productivity. Heck with the same level of HW coding it might have done the old-skool 2D just as fast as AGA (due to much better CPU performance).
|
ET4000AX comes in 16-bit fast VRAM or 32-bit DRAM.
From https://archive.org/details/bub_gb_hqQJaNzN9IcC/page/n603/mode/2up
PC Mag 1992-08, page 604 of 664, Diamond Speedstar 24 (ET4000AX ISA) has $169 USD.
https://www.youtube.com/watch?v=5o9yOBBWPgM PC 286-16 with a fast VGA clone and SB sound card as a retro gaming machine. This 286-16 based PC plays A500 class games with VGA's 256 colors.
PC components can be recycled for the 386DX-33 build while the entire Amiga 1500/2000/2500 with full 32-bit 68020/68030 accelerator is ditched for AGA SKUs.
Last edited by Hammer on 21-Oct-2022 at 07:47 AM. Last edited by Hammer on 21-Oct-2022 at 07:41 AM.
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 21-Oct-2022 9:20:34
| | [ #404 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @Hypex
Quote:
Hypex wrote: @cdimauro
Quote:
Indeed. But the main problem is that SuperHires using 8 bitplanes is utterly slow: it leaves only a few memory slots free for doing something other than displaying the screen. |
Given HAM was always a hardware mode I always wondered what incursion it had on the system. It uses max planes but aside from that does HAM8 mode take more out of the system? Though not as pretty, HAM6 would be better, if system load is considered.
C2P takes some time. Though not too much on an 060 I found out. But here it needs to render four times as much. So it does it's raytcasting in low res. But then it upscales it to super hi res! Ouch.
16 bpp, 24 bpp, either way, the framebuffer it renders is already in a bigger size against 8 bit. In the best quality case, it needs to render a long word per pixel, so 320 pixels across will be 1280 bytes to store per line. Then, to dump it in HAM, it needs to read in a long per pixel and output 4 bits in planar per plane. So, taking 8 pixels, scaling each to a nibble per plane, 8 long words of 24-bit RGB pixels becomes 8 long words in planar. Madness!
|
HAM acts like lossy color compression i.e. "HAM can be considered a lossy compression technique, similar in operation and efficiency to JPEG minus the DCT stage".
In modern PC GPUs, we have marketing spill on the latest video hardware decompression support and Delta Color Compression (DCC) everywhere.
Intel is making a deal out of AV1 hardware support on their ARC GPUs.
AMD's DCC marketing for RDNA
NVIDIA's DCC marketing for CUDA Pascal.
Delta Color Compression works on chunky pixel tiles and compression is applied. DCC adds logic complexity but it conserves bandwidth and cache.
Amiga's HAM mode was a good idea and should have been refined and evolved. Amiga's HAM color compression idea was ahead of its time.
There's significant latency between Agnus/Alice's Copper and Denise/Lisa's color registers since the chips are separated.
Brute force and efficiency win the graphics hardware game.
Commodore's management wouldn't have supported NVIDIA's relentless innovation business model. NVIDIA sells "toy" SoCs for Nintendo's Switch mobile games console i.e. a long time ago, Nintendo's NES has Ricoh 2A03 containing a second sourced MOS Technology/ CSG (Commodore Semiconductor Group) 6502 CPU core.
Switch's SoC ARM CPU family was Acorn's replacement for the uncompetitive CSG 65xx CPU R&D roadmap. ARM contributing its strength for Amiga's PiStorm software 68K accelerator.
Most participants in this topic wanted Commodore Semiconductor Group to be successful as modern-day NVIDIA or AMD or Intel that continued Amiga's technology evolution.
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 21-Oct-2022 10:13:59
| | [ #405 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @bhabbott
Quote:
bhabbott wrote: @Hammer
Quote:
Thanks for the link - another magazine I had forgotten about!
But...
Amiga 3000 Quote:
Release date June 1990 Discontinued 1992
Commodore replaced the A3000 six months behind schedule, in the fall of 1992, with the A4000 |
We knew something was coming even before the A4000 was announced, because Commodore slashed the price of the A3000 a few months beforehand. The price you see it selling for in July 1992 was not the launch price!
And yes, those of us who bought it at full price were a little miffed. But we understood why they had to dump it, and the promise of something new helped alleviate the pain of seeing the value of our 'investment'' drop dramatically. Such is life on the bleeding edge...
The magazine you should be looking at is ACAR Annual 1991, with a review of the A3000 quoting prices of:-
Amiga 3000-25-40 (25MHz with 40Mb HD) AU$6119 Amiga 3000-25-100 (25MHz with 100Mb HD) AU$7199
|
ECS wasn't a "major leap in technology."
Amiga 3000 was in the Australian Personal Computer magazine in June 1990
I was aware of the UK magazine's faster CPU-equipped Amigas with ECS rumors such as Amiga 1400(?). New A3000's 1991 asking price was off our radar.
From https://archive.org/details/amiga-world-1992-03/page/n65/mode/2up Commodore has adverts like this
In early 1992, my Dad was one of those unwise buyers who purchased an ex-corporate A3000 since my Dad wasn't aware of the A3000's discontinuation and AGA's existence. If my Dad has been aware of the A3000's EOL and AGA's road map, He wouldn't have purchased A3000. AGA's release in Q4 1992, the purchase of A3000 with a non-upgradeable ECS resulted in an unhappy customer, hence my Dad dumped A3000's complete ownership to me.
This situation is repeated for many full 32-bit CPU accelerated Amiga OCS/ECS owners.
Dave Haynie's A3000 AGA motherboard upgrade advocacy was important for existing A3000 owners i.e. we don't need to buy a new A3000 case, Amiga keyboard, mouse, PSU, Zorro III daughterboard, high-density FDD, and SCSI HDD.
Dave Haynie's A3000 AGA motherboard upgrade advocacy was rejected by Commodore's management.
If the Amiga platform was a clone market, somebody would have provided the upgrade solution e.g. AGA on ECS. https://github.com/nonarkitten/amiga_replacement_project or Commodore sells AGA chips to AIB partners like NVIDIA's or AMD's AIB partner business model.
Amiga's custom chip section wasn't designed to be AIB modular.
Before Amiga Hombre OpenGL PCI AIB (add-in-board) plans, Commodore doesn't like NVIDIA's or ATI's AIB business model e.g. driven off Checkmate A1500 case from the market with Commodore's A1500.
Since 1993, my Dad has banned any future IBM and Commodore products from our household. My Dad was fully focused on PC's modular design with defacto industry modular standards. Since I fully owned A3000, I sold it to a small TV studio that helps to fund Pentium 150-based PC in 1996. Since 1998, I took over PC purchasing upgrade cycles and my relatives receives my "ex-gaming PC hardware" hand-me-downs.
I only purchased "sold for parts" A1200 during the COVID-19 lockdown in 2020, TF1260 in 2021, and Indivision AGA MK3 in 2022. A1200 was been recapped in 2021.
I took the chance with "sold for parts" A1200 since it advertised with booting into prompt with clicking HDD (it's actually floppy drive's clicking check) since fully working A1200's prices are crazy.
Last edited by Hammer on 21-Oct-2022 at 10:42 AM. Last edited by Hammer on 21-Oct-2022 at 10:39 AM. Last edited by Hammer on 21-Oct-2022 at 10:34 AM. Last edited by Hammer on 21-Oct-2022 at 10:30 AM. Last edited by Hammer on 21-Oct-2022 at 10:27 AM. Last edited by Hammer on 21-Oct-2022 at 10:24 AM.
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | bhabbott
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 1:16:32
| | [ #406 ] |
| |
|
Regular Member |
Joined: 6-Jun-2018 Posts: 453
From: Aotearoa | | |
|
| @cdimauro
Quote:
cdimauro wrote:
So, the they decided to keep the same allocation slots mechanism and castrated the CPU access to the RAM forcing it to wait two additional cycles before the required two for the concrete access.
|
I'm impressed by how smoothly you switched from "Which is plainly wrong... The CPU isn't a 68000 (AGA machines have at least a 68020), so it has always access to those slots" to "they decided to keep the same allocation slots mechanism and castrated the CPU access", without missing a beat! Well done sir. Have you thought of becoming a politician?
Quote:
Beautiful! Only the so smart Commodore engineers could have done it: making it worse that the prehistoric 16-bit ISA bus (since it runs at 8Mhz at least). |
According to the ISA bus 'standard' the maximum rated speed is 8MHz - anything over that is non-standard and at the user's risk. In 1992 the ISA bus wasn't 'prehistoric' - it was a standard feature of all PCs and the only bus in contemporary 386DX and 386SX machines.
But how fast was it really? In this video several 16 bit ISA VGA cards were tested in an 80MHz 486DX2 PC. The most interesting one is the TVGA9000B, which was released in 1992 and commonly fitted to many PC clones. At an ISA bus speed of 10MHz (25% overclocked) it achieved 2.9MB/s write speed and 2.44MB/s read speed.
In comparison a stock A1200 gets 5.33MB/s write speed and 3.0MB/s read speed. So a stock A1200 with 14MHz 020 can read and write Chip RAM faster than a PC with popular ISA VGA card of the day - even with a much faster 486 CPU.
An A1200 with 50MHz 030 (Blizzard 1230-IV) gets 7.07MB write speed and 4.84MB/s read speed, which matches or exceeds the performance of the 80MHz 486 with a GB5422 VL bus VGA card running in ISA bus mode. But does this affect real-world performance? My 50MHz 030 equipped A1200 gets a 25% higher frame rate in Doom than a 40MHz 386DX, despite the overhead of c2p. Part of the reason it can do this is that c2p'ing into ChipRAM is faster than directly rendering to a typical ISA bus VGA card. The 'oh so smart' engineers at Commodore tried to develop a much more advanced graphics chipset, and failed. The task then set for them was to make a less advanced chipset that was highly compatible with ECS, and they had to do it quickly. So they did the sensible thing - kept Agnus largely the same, with the same bus timing and blitter etc. IMO this is what they should have done from the start, easing into a more advanced architecture in achievable steps instead of trying to do it all at once.
Though the CPU in a stock A1200 doesn't achieve the maximum Chip RAM speed that would be possible if it had access to all slots, in practice this isn't a big deal. In a typical AGA application bitplane DMA will be using about half the slots anyway, so it isn't missing much. Slots it leaves free in lower resolutions will be used by the blitter, so they aren't necessarily wasted.
Even when 'handicapped' by not having Fast RAM, the 020 in the A1200 is twice as fast as the 68000 in an A500. Combined with AGA graphics this takes the A1200 to a higher level than the A500 without being divorced from it. For example Dread, the 'Doom clone' being developed for the A500, runs very well on a stock A1200 but has graphical glitches on my A1200 with 50MHz Blizzard 1230-IV, even when CPU caches are turned off. This applies to a large number of commercial games too. Being able to run existing Amiga titles was very important to the survival of the A1200, which would take time to build up a library of AGA titles.
Manufacturers of other platforms often found out the hard way that customers tend to value compatibility more than advanced features or performance. The Amiga 1000 was much more powerful than a typical PC-XT clone of the day, but completely useless to most potential buyers because it wasn't IBM compatible. Not only that, but even an 'almost 100%' compatible MSDOS machine was not acceptable when their favorite app glitched on it.
Similarly, many Amiga fans dinged the A500+ and A600 for not being 100% compatible with existing software. Some incompatibility was unavoidable on the A1200, but Commodore managed to achieve a reasonable balance at low cost. Having only ChipRAM avoided problems with data not being put in the correct memory type. Having a CPU with 24 bits address bus. no data cache and and not too fast made it more likely to run software designed to work on the A500. The AGA chipset was designed to be virtually the same as ECS except for the new AGA features. Having an incompatible chipset - even something 'harmless' like the CPU using more DMA slots - could have been very frustrating for users.
Last edited by bhabbott on 22-Oct-2022 at 03:45 AM.
|
| Status: Offline |
| | bhabbott
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 3:30:43
| | [ #407 ] |
| |
|
Regular Member |
Joined: 6-Jun-2018 Posts: 453
From: Aotearoa | | |
|
| @Hammer
Quote:
Hammer wrote:
ECS wasn't a "major leap in technology." |
Who should we give this week's prize for Gish Galloping to, you or cdimauro? I think you should get it for the size of the irrelevant images you posted (which BTW took ages to load in iBrowse on my 1200, due to the huge file size and enormous TLS decryption overhead. Why oh why do images have to be encrypted?).
Quote:
I was aware of the UK magazine's faster CPU-equipped Amigas with ECS rumors such as Amiga 1400(?). New A3000's 1991 asking price was off our radar. |
Well it certainly wasn't off my RADAR, because I pre-ordered it as soon as they were announced. At that time I had plenty of money, and a new job doing Amiga software development. A typical PC and software tools for cross-development targeting the Amiga would have cost more (and probably be a lot more frustrating too)..
Quote:
If my Dad has been aware of the A3000's EOL and AGA's road map, He wouldn't have purchased A3000. AGA's release in Q4 1992, the purchase of A3000 with a non-upgradeable ECS resulted in an unhappy customer, hence my Dad dumped A3000's complete ownership to me. |
An understandable reaction, but quite shortsighted. He got the machine cheap, and could have used some of the money saved to add an RTG card for better performance than AGA.
Quote:
This situation is repeated for many full 32-bit CPU accelerated Amiga OCS/ECS owners. |
Not really. The vast majority would be A2000 owners who had already chosen performance over compatibility with stock machines. Sure they would be missing out on whatever new AGA titles were produced, but with a graphics card and faster CPU etc. they could do much more. And many did.
You dad had a negative attitude. No matter what system you buy it will eventually become 'redundant' when newer models come out. No platform suffered more from this in the 90's than the PC. It was common for software to be produced assuming that the next generation of PCs would be fast enough for it, with the 'minimum' installation specs being a joke.
An editorial in the June 1994 issue of NZ PC World magazine (original copy - not sure if it's available online) argued that it was better to buy a low-end model every 2 years than buy a high-end model expecting it to last. The suggestions were:-
1990: 386SX-20, 1MB RAM, 20MB hard disk, monochrome display NZ$2495 1992: 386SX-20,2MB RAM, 105MB hard disk, color display NZ$2395 1994: 486SX-25,4MB RAM, 170MB hard disk, color display NZ$2399
As opposed to:- 1990 386DX-33, 2MB RAM, 120MB hard drive, color display NZ$7495
Three machines costing less than one, and ending up with a much higher spec model!
Many PC owners followed this advice involuntarily as they found their existing machines weren't good enough to run the latest software. retailers loved it. A few struggled to upgrade their existing machines instead. I did good business in this area, often replacing entire motherboards and other stuff, then using the old parts to upgrade other customers' older models!
But this was not the Amiga way. We didn't want a chameleon that kept changing its appearance to suit the environment, we wanted to keep using the machines we knew and loved. Now that the Amiga is retro this is even more true. We only have a few chipsets to support and they are pretty much the same, which is a good thing. On the PC you had a plethora graphics cards even before things started going crazy with 3D and different card slots etc. It's almost impossible to program a graphics card 'bare metal' these days, and if you do it probably won't work an any other card. Everything has to be done through libraries and 'blobs' provided by the manufacturers - no fun at all. Quote:
Dave Haynie's A3000 AGA motherboard upgrade advocacy was rejected by Commodore's management. |
And fair enough too. The A3000 was over-engineered, over-priced, and a money sink - and there weren't enough A3000 owners willing to upgrade to make it worthwhile. Better to produce a new model that doesn't have to conform to the limitations of the A3000. If you wanted AGA to play games then an A1200 or CD32 was a better choice. I you just wanted better graphics then an RTG card was more appropriate.
Perhaps if Commodore had lasted longer they might have looked at making upgrades for existing Amiga models, or more likely motherboards in PC form factor. Today some people are doing just that! I predict it won't be long before you can get a motherboard to put existing AGA chips into that is not just a clone of the A1200 or A4000 - and these will come in various form factors including PC cases as well as Amiga models such as the A1000/3000/500/600. Quote:
This is happening now only because the means to do is in the hands of amateurs (with FPGA etc.), and there is no parent company to placate. When Commodore was producing Amigas they couldn't afford unauthorized cloning as it would cut into their market, and the market wasn't big enough to make licensed cloning viable. Apple already found that out, and IBM lost all of their market due to cloning. 'Cloning' of the Amiga was only viable in high-end markets such as video production, and then only for a short time before PCs muscled into there too.
Quote:
Since 1993, my Dad has banned any future IBM and Commodore products from our household. My Dad was fully focused on PC's modular design with defacto industry modular standards. |
Your dad was a killjoy. Did he ban all video game consoles too? How about cell phones, TVs, microwave ovens, and all those other consumer products that aren't modular? Quote:
I only purchased "sold for parts" A1200 during the COVID-19 lockdown in 2020, TF1260 in 2021, and Indivision AGA MK3 in 2022. A1200 was been recapped in 2021.
I took the chance with "sold for parts" A1200 since it advertised with booting into prompt with clicking HDD (it's actually floppy drive's clicking check) since fully working A1200's prices are crazy. |
Yes, prices for sought-after Amigas are crazy these days. Still mostly cheaper than they were new though, and amazingly most of them are still working 25-30 years later! I would have more Amigas despite the prices, except that I wouldn't use them all so it would be a waste. Your A1200 setup is sweet!
Just because PCs left the Amiga in the duct years ago doesn't mean we can't still enjoy using and improving them. In fact it's much better because we don't have to deal with PC envy anymore. There is so much I never got to play with back in the day that is still interesting.
Today I downloaded a game called 'Paradroid 90' for a friend who has an A500 doing nothing that he would use again if he could get that game. I had never played it before. Then I downloaded some Amiga Format subscriber disks from The Zone that we didn't get with the magazines. One had an RC model jet simulator written in AMOS 3D by a New Zealander. I recognized the background scenery! It ran pretty good on the A1200 with 50MHz 030, and I flew that jet around for about 20 minutes. That's more simulated RC model flying than I have done in real life this year (our flying field got flooded in April and again this month - still hasn't recovered).
|
| Status: Offline |
| | cdimauro
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 5:04:04
| | [ #408 ] |
| |
|
Elite Member |
Joined: 29-Oct-2012 Posts: 4111
From: Germany | | |
|
| @Hammer
Quote:
Hammer wrote: If my Dad has been aware |
that his son wouldn't be able to use an image host service neither to reduce the size of the images that he posts, so continuously fu*king the forum layout, then maybe Hammer was not here...
@bhabbott
Quote:
bhabbott wrote:
Who should we give this week's prize for Gish Galloping to, you or cdimauro? |
Fly down, Bruce: it was the first time here for me, and for good reasons.
Whereas YOU have written piles of bullsh*t since you're here, which I've a finger pointed (and you stopped replied after a while).
So, and as we're used to say in Italy, don't exchange our clothes. |
| Status: Offline |
| | QuikSanz
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 6:18:03
| | [ #409 ] |
| |
|
Super Member |
Joined: 28-Mar-2003 Posts: 1236
From: Harbor Gateway, Gardena, Ca. | | |
|
| @bhabbott,
nonarkitten does not use any FPGA's, and yet still with more functionality. Well, maybe a small one.
Last edited by QuikSanz on 22-Oct-2022 at 06:22 AM.
|
| Status: Offline |
| | cdimauro
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 6:22:33
| | [ #410 ] |
| |
|
Elite Member |
Joined: 29-Oct-2012 Posts: 4111
From: Germany | | |
|
| @bhabbott
Quote:
bhabbott wrote: @cdimauro
Quote:
cdimauro wrote:
So, the they decided to keep the same allocation slots mechanism and castrated the CPU access to the RAM forcing it to wait two additional cycles before the required two for the concrete access.
|
I'm impressed by how smoothly you switched from "Which is plainly wrong... The CPU isn't a 68000 (AGA machines have at least a 68020), so it has always access to those slots" to "they decided to keep the same allocation slots mechanism and castrated the CPU access", without missing a beat! Well done sir. Have you thought of becoming a politician? |
No, I'm just a programmer which had no AGA hardware manual (since Commodore never published it) and that has only relying on Motorola's manuals for its processors.
So, I wasn't aware of those details and putting together the information that I had. But at least I wasn't the only one, looking at the thread that you give (which was opened just 3 years ago, BTW).
Anyway, wrong assumptions. Quote:
Quote:
Beautiful! Only the so smart Commodore engineers could have done it: making it worse that the prehistoric 16-bit ISA bus (since it runs at 8Mhz at least). |
According to the ISA bus 'standard' the maximum rated speed is 8MHz - anything over that is non-standard and at the user's risk. |
Which is wrong (and I can state it): there was no formal spec (and maybe that's why you've put the standard word in quotes) about the ISA bus telling that the maximum is 8Mhz.
In fact, it was a synchronous bus, so its clock was strictly bounded to the processor clock.
It was decoupled by it because of this caused issues with SOME cards.
But you're not forced to keep it as this maximum speed: you could go higher, as long as the cards on your PC can sustain the higher clock. Quote:
In 1992 the ISA bus wasn't 'prehistoric' - it was a standard feature of all PCs |
The 8-bit ISA was released on 1981 and the 16-bit version on 1984: that should be enough to define them prehistoric. Quote:
and the only bus in contemporary 386DX and 386SX machines. |
Only on the ones that you know, because other and more advanced buses were available: https://en.wikipedia.org/wiki/Micro_Channel_architecture https://en.wikipedia.org/wiki/Extended_Industry_Standard_Architecture https://en.wikipedia.org/wiki/VESA_Local_Bus Quote:
But how fast was it really? In this video several 16 bit ISA VGA cards were tested |
Several equals to just FOUR cards:
Trident TVGA8800CS Trident TVGA9000B Video Seven 456-0023 Cirrus Logic CL-GD5422
Which weren't the best ones. Some even with only 256kB, which usually were the cheapest ones (and low-performances). Quote:
From the video description:
I'm going to build a 386SX
which was also shown at the beginning of the video albeit for the tests he used a 486 DX2/66. Quote:
The most interesting one is the TVGA9000B, which was released in 1992 and commonly fitted to many PC clones. At an ISA bus speed of 10MHz (25% overclocked) it achieved 2.9MB/s write speed and 2.44MB/s read speed. |
Maybe because it's a crappy card? Man, you report only what's convenient for your propaganda!
The Cirrus Logic card got from 3.4 to 4.6MB/s for read speed and from 7.2 to 9.3MB/s for write speed.
So, the write speed was saturating the bus and that's the most important metric, since PC games were mostly writing the elaborated graphic to the video card. Quote:
In comparison a stock A1200 gets 5.33MB/s write speed and 3.0MB/s read speed. So a stock A1200 with 14MHz 020 can read and write Chip RAM faster than a PC with popular ISA VGA card of the day - even with a much faster 486 CPU.
An A1200 with 50MHz 030 (Blizzard 1230-IV) gets 7.07MB write speed and 4.84MB/s read speed, which matches or exceeds the performance of the 80MHz 486 with a GB5422 VL bus VGA card running in ISA bus mode. But does this affect real-world performance? | See above. And BTW even an 80386 could get the similar read/write speed on ISA cards, because it's the ISA bus the bottleneck here. Quote:
My 50MHz 030 equipped A1200 gets a 25% higher frame rate in Doom than a 40MHz 386DX, despite the overhead of c2p. Part of the reason it can do this is that c2p'ing into ChipRAM is faster than directly rendering to a typical ISA bus VGA card. | Again, see above: maybe because you used a crappy video card.
Plus, we don't know if the Amiga Doom port had other optimizations which the original Doom hadn't. Quote:
The 'oh so smart' engineers at Commodore tried to develop a much more advanced graphics chipset, and failed. | Do you understand that they failed TWICE? The first failure was with the ECS chipset, since they implemented the same crippled Chip-mem access.
But they weren't satisfied enough, so they decided to port the same crippled access to AGA... Quote:
The task then set for them was to make a less advanced chipset that was highly compatible with ECS, and they had to do it quickly. | Sure, I've no doubt about it: it's fully compatible with ECS. I see, I see. Quote:
So they did the sensible thing - kept Agnus largely the same, with the same bus timing and blitter etc. IMO this is what they should have done from the start, easing into a more advanced architecture in achievable steps instead of trying to do it all at once. | 7 (SEVEN) years recycling the same things. Thunderous applause, please! Quote:
Though the CPU in a stock A1200 doesn't achieve the maximum Chip RAM speed that would be possible if it had access to all slots, in practice this isn't a big deal. | Only if you had Fast mem, but this wasn't the case. Quote:
In a typical AGA application bitplane DMA will be using about half the slots anyway, so it isn't missing much. |
I've already gave you the math. When do you think that it's time to read AND understand it? Quote:
Slots it leaves free in lower resolutions will be used by the blitter, so they aren't necessarily wasted. |
They were wasted because the 68020 loses its capability to copy or fill Chip-Mem quicker than the Blitter.
Yes: some graphic operations could have been better managed by the CPU IFF the Chip-mem access wasn't crippled.
Considered that the stock Amiga 1200 is a very low-end machine, it could have helped a lot. Quote:
Even when 'handicapped' by not having Fast RAM, the 020 in the A1200 is twice as fast as the 68000 in an A500. |
Sure. A 32-bit 14Mhz CPU with much faster instructions execution and 256 byte instructions cache which was doubling a 16-bit 7Mhz CPU with no cache and were the minimum instruction was taking 4 clock cycles.
Only Commodore engineers made it possible! Quote:
Combined with AGA graphics this takes the A1200 to a higher level than the A500 without being divorced from it. |
This proves that sometimes a divorce is the best solution... Quote:
For example Dread, the 'Doom clone' being developed for the A500, runs very well on a stock A1200 |
It's not a Doom clone. Please don't sell absurdities.
It's a fine piece of software, no doubt about it according to what it's able to do even on a stock Amiga 500, but Doom is a completely different beast. Quote:
but has graphical glitches on my A1200 with 50MHz Blizzard 1230-IV, even when CPU caches are turned off. This applies to a large number of commercial games too. Being able to run existing Amiga titles was very important to the survival of the A1200, which would take time to build up a library of AGA titles. |
Don't tell me: I know it very well (I was putting "compatibility labels" on most of the games and the few of the demos that I had), and I think you already know my opinion about such stupid developers. Quote:
Manufacturers of other platforms often found out the hard way that customers tend to value compatibility more than advanced features or performance. The Amiga 1000 was much more powerful than a typical PC-XT clone of the day, but completely useless to most potential buyers because it wasn't IBM compatible. Not only that, but even an 'almost 100%' compatible MSDOS machine was not acceptable when their favorite app glitched on it. |
Well, PC games/demos weren't exempted: stupid developers were also there. Quote:
Similarly, many Amiga fans dinged the A500+ and A600 for not being 100% compatible with existing software. Some incompatibility was unavoidable on the A1200, |
I fully disagree: incompatibilities were avoidable. By following Commodore's guidelines. Quote:
but Commodore managed to achieve a reasonable balance at low cost. |
At least here Commodore wasn't the culprit: see above. Quote:
Having only ChipRAM avoided problems with data not being put in the correct memory type. Having a CPU with 24 bits address bus. no data cache and and not too fast made it more likely to run software designed to work on the A500. |
Should I cry? Please... Quote:
The AGA chipset was designed to be virtually the same as ECS except for the new AGA features. Having an incompatible chipset - |
The AGA chipset was NOT incompatible. Besides ECS's UltraHires feature (which nobody used). Quote:
even something 'harmless' like the CPU using more DMA slots - could have been very frustrating for users. |
I absolutely disagree: the guidelines were there to AVOID those issues and users frustration.
It was all about the idiots that developed these crappy games/demos that caused such frustrations. |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 8:50:18
| | [ #411 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @NutsAboutAmiga
https://www.youtube.com/watch?v=1LXfD3gQZHE&lc=UgyICwAtyTTrBkgjKoF4AaABAg.9hRr_cDhJ7P9hTuIRBWPJ3 Street Fighter 2 tech demo with parallax update for Commodore Amiga 500/600s.
TECH DEMO UPDATED 20/10/2022 - Screen mode changed to EHB allowing perfect sprite coloring - Background replaced with the SNES version (more accurate proportions) - Added parallax scrolling to floor (3 layers) and background roof - Added signs to the edge of the stage (not breakable atm) - Ken follows player backward - Splash screen removed from game loop - Optimized panels palettes - Optimized sounds
16-bit Fast RAM is recommended since 16-bit 2 MB Chip RAM with A500 Plus and A600 is not fast enough. With 16-bit Fast RAM, the Amiga OCS/ECS chipset and stock 68000 CPU can operate at their near full extent since there's significantly less shared CPU/IGP memory bus bottleneck.
Last edited by Hammer on 22-Oct-2022 at 08:51 AM.
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 8:54:40
| | [ #412 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @cdimauro
Quote:
hat his son wouldn't be able to use an image host service neither to reduce the size of the images that he posts, so continuously fu*king the forum layout, then maybe Hammer was not here...
|
What's the matter? You're not using a 4K monitor?
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 9:00:51
| | [ #413 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @bhabbott
Quote:
An understandable reaction, but quite shortsighted. He got the machine cheap, and could have used some of the money saved to add an RTG card for better performance than AGA.
|
From 1992 to 1995, RTG support is rubbish. My Dad is an "Xbox sports gamer" who played sports games on the Amiga from the late 1980s to the early 1990s and is an accountant (retired in 2016).
Reminder,
CyberGraphX was introduced in 1995.
Picasso96 was introduced in 1996.
386DX-33 with ET4000AX can play Amiga AGA game ports.
https://www.youtube.com/watch?v=KxMt-bIm5bk 386DX-25 PC (with FX-3000 motherboard, ET4000AX SVGA, Sound Blaster Pro 2) playing Wing Commander VGA.
For Wing Commander AGA, A1200 needs 68020 @ 25 Mhz with a Fast RAM accelerator for comparable results.
From https://www.vogons.org/viewtopic.php?t=37583 FX-3000 motherboard existed in 1992.
Faster VGA/SVGA clones have cloned IBM's VGA compatibility standard.
For games, your RTG remark is useless for the 1992 to 1994 time period.
Furthermore, 1996 era Phase 5 CyberGraphics 64/CyberStorm 060 upgrade for Amiga 3000 wasn't cost-effective when compared to a new PC clone with Pentium 150/S3 Trio 64/Sonata SR-S163 sound card (SoundBlaster Pro clone)/PC Chips M525 430VX motherboard when playing Quake. For the money, Pentium 150 crushed CyberStorm 060 upgrade when playing Quake.
Quote:
You dad had a negative attitude. No matter what system you buy it will eventually become 'redundant' when newer models come out. No platform suffered more from this in the 90's than the PC. It was common for software to be produced assuming that the next generation of PCs would be fast enough for it, with the 'minimum' installation specs being a joke.
An editorial in the June 1994 issue of NZ PC World magazine (original copy - not sure if it's available online) argued that it was better to buy a low-end model every 2 years than buy a high-end model expecting it to last. The suggestions were:-
1990: 386SX-20, 1MB RAM, 20MB hard disk, monochrome display NZ$2495 1992: 386SX-20,2MB RAM, 105MB hard disk, color display NZ$2395 1994: 486SX-25,4MB RAM, 170MB hard disk, color display NZ$2399
|
That's with a new PC purchase. For AGA games on Amiga from 1992-to-1993, there is a hard incompatibility difference between A3000's ECS (useless 32-bit Chip RAM) and A1200 AGA.
For games on a 386 PC with VGA, there is a performance difference between IBM VGA and faster ET4000AX, but they run common software with no hard incompatibility.
A slow IBM PS/2 Model 55SX (386-16, MCA 16-bit slots, IBM's onboard 16-bit VGA) with 5 MB of RAM can run the same software as a faster PC 386-DX clone with ET4000AX, but the main difference is performance. There is no hard incompatibility difference .
For 386-era software, there's a hard incompatibility difference between 286 and 386, but the PC clone allows component upgrades on either the CPU or graphics side.
When I upgrade to my PC later this year, I only need a new CPU, a new motherboard, and new DDR5-5600/6000 modules. Other components such as Sound Blaster Z PCIe, GeForce RTX 3080 Ti OC, two 1 TB Samsung NVMEs, two 10 TB WD Black HDDs, Phanteks P500A dRGB case, gaming keyboard, gaming mouse, Xbox One controller, Corsair i115 280 mm AIO) are recycled for the next hardware generation.
I can recycle my existing DDR4 modules with the Inte Rocket Lake DDR4 platform.
Intel Haswell Core i7-4790K and GTX 1660 Super will run the last games with slower performance.
A modern incompatibility example is the AVX requirement when Apple's Rosetta 2 doesn't emulate AVX.
---- Against your NZ PC prices.
Due to Australia's small and New Zealand's tiny market size, we don't dictate a computer platform's success or failure.
The US state of Texas's population is about 28 million with just over $2 trillion in GDP while the entire Australian population is only 26 million with $1.5 trillion in GDP.
Australia-New Zealand's common market population size is about 30 million in the year 2021.
Last edited by Hammer on 22-Oct-2022 at 11:05 AM. Last edited by Hammer on 22-Oct-2022 at 10:37 AM. Last edited by Hammer on 22-Oct-2022 at 10:10 AM. Last edited by Hammer on 22-Oct-2022 at 10:07 AM. Last edited by Hammer on 22-Oct-2022 at 09:48 AM.
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | cdimauro
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 10:17:45
| | [ #414 ] |
| |
|
Elite Member |
Joined: 29-Oct-2012 Posts: 4111
From: Germany | | |
|
| @Hammer
Quote:
Not really.
And the EHB wasn't used even for the two characters' shadows. Quote:
- Background replaced with the SNES version (more accurate proportions) |
The resolution looks also the same. Quote:
- Added parallax scrolling to floor (3 layers) and background roof |
Only the floor one is barely noticeable.
They need to do a lot of work. Quote:
16-bit Fast RAM is recommended since 16-bit 2 MB Chip RAM with A500 Plus and A600 is not fast enough. |
Only for that?!? Quote:
[b]With 16-bit Fast RAM, the Amiga OCS/ECS chipset and stock 68000 CPU can operate at their near full extent since there's significantly less shared CPU/IGP memory bus bottleneck.[/b |
I wonder how they are using the Fast RAM, since this game almost entirely dominated by Chip-mem access & Blitter usage.
Quote:
Hammer wrote: @cdimauro
Quote:
hat his son wouldn't be able to use an image host service neither to reduce the size of the images that he posts, so continuously fu*king the forum layout, then maybe Hammer was not here...
|
What's the matter? You're not using a 4K monitor? |
Classy (cit.)
No: I've 3 x FullHD and 2 x HDReady monitors.
I don't think that I'm the only one in this situation. Plus,. some amigans might be using even lower resolutions.
If you want to add 4K pictures it's absolutely fine, as long as you use a (good) image hosting service: then you have a reasonably sized image in the post, and people can see the full image by clicking on it. |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 10:56:19
| | [ #415 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @cdimauro
Quote:
And the EHB wasn't used even for the two characters' shadows. |
Not my claim. EHB colors are not limited to shadows.
Quote:
Only the floor one is barely noticeable. |
Work in progress, the parallax background movement concept is working.
Quote:
wonder how they are using the Fast RAM, since this game almost entirely dominated by Chip-mem access & Blitter usage.
|
Apparently, the game's frame rate is smoother with a faster CPU and Fast RAM or on the stock A1200.
I configured WinUAE with Amiga 500 Plus with 2 MB 16-bit Chip RAM configuration and the SF2 tech demo runs slow. Fast RAM is needed.
Last edited by Hammer on 22-Oct-2022 at 11:06 AM.
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | Kronos
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 11:06:55
| | [ #416 ] |
| |
|
Elite Member |
Joined: 8-Mar-2003 Posts: 2676
From: Unknown | | |
|
| @bhabbott
Quote:
bhabbott wrote: @ In comparison a stock A1200 gets 5.33MB/s write speed and 3.0MB/s read speed. So a stock A1200 with 14MHz 020 can read and write Chip RAM faster than a PC with popular ISA VGA card of the day - even with a much faster 486 CPU.
|
So the 020 could access it's local RAM faster them a 386/486 going through ISA bus.
Shock horror!!!
Now tell us how fast could these x86 CPUs access the RAM on the motherboard?
-> AGA crippled CPU performance -> AGA was a bad idea (but maybe still the best C= could have had at that time)._________________ - We don't need good ideas, we haven't run out on bad ones yet - blame Canada |
| Status: Offline |
| | Hammer
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 11:40:00
| | [ #417 ] |
| |
|
Elite Member |
Joined: 9-Mar-2003 Posts: 5972
From: Australia | | |
|
| @bhabbott
Quote:
ot really. The vast majority would be A2000 owners who had already chosen performance over compatibility with stock machines. Sure they would be missing out on whatever new AGA titles were produced, but with a graphics card and faster CPU etc. they could do much more. And many did.
You dad had a negative attitude. No matter what system you buy it will eventually become 'redundant' when newer models come out. No platform suffered more from this in the 90's than the PC. It was common for software to be produced assuming that the next generation of PCs would be fast enough for it, with the 'minimum' installation specs being a joke.
An editorial in the June 1994 issue of NZ PC World magazine (original copy - not sure if it's available online) argued that it was better to buy a low-end model every 2 years than buy a high-end model expecting it to last. The suggestions were:-
1990: 386SX-20, 1MB RAM, 20MB hard disk, monochrome display NZ$2495 1992: 386SX-20,2MB RAM, 105MB hard disk, color display NZ$2395 1994: 486SX-25,4MB RAM, 170MB hard disk, color display NZ$2399
As opposed to:- 1990 386DX-33, 2MB RAM, 120MB hard drive, color display NZ$7495
Three machines costing less than one, and ending up with a much higher spec model!
Many PC owners followed this advice involuntarily as they found their existing machines weren't good enough to run the latest software. retailers loved it. A few struggled to upgrade their existing machines instead. I did good business in this area, often replacing entire motherboards and other stuff, then using the old parts to upgrade other customers' older models!
But this was not the Amiga way. We didn't want a chameleon that kept changing its appearance to suit the environment, we wanted to keep using the machines we knew and loved. Now that the Amiga is retro this is even more true. We only have a few chipsets to support and they are pretty much the same, which is a good thing. On the PC you had a plethora graphics cards even before things started going crazy with 3D and different card slots etc. It's almost impossible to program a graphics card 'bare metal' these days, and if you do it probably won't work an any other card. Everything has to be done through libraries and 'blobs' provided by the manufacturers - no fun at all.
|
For this topic's 1992 to 1993 year date period,
For Sept 1993 in AUD, Brisbane, state of Queensland and Australia.
486SX-33-based PC priced at $1,545 AUD. Three months away from Doom's December 1993 release.
Find a better deal with the Amiga AGA!
https://i.ibb.co/44zfhyK/Australia-QLD-A1200-price-with-trade-in-June-1993.png In June 1993, near Brisbane, Australia. Toowoomba is south of Brisbane.
Amiga 1200 with trade-in's asking price is $1,199 AUD.
https://archive.org/details/Australian_Commodore_and_Amiga_Review_The_Volume_10_Issue_6_1993-06_Saturday_Magazine_AU/page/n41/mode/2up For June 1993, Page 42 of 84
A1200 bare bones from $1,195. AUD Derringer 030 @ 50 Mhz with 2MB Fast RAM is $1,395 AUD
----------------- June 1992 from Your Computer (Australia)
Barebone 386DX-33 with 1MB RAM, 1.44 FDD, 512K SVGA, SVGA monitor, keyboard, Baby AT case = $1,335 AUD. It's bare-bones as the later A1200.
Last edited by Hammer on 22-Oct-2022 at 12:32 PM. Last edited by Hammer on 22-Oct-2022 at 12:13 PM. Last edited by Hammer on 22-Oct-2022 at 12:12 PM. Last edited by Hammer on 22-Oct-2022 at 12:11 PM. Last edited by Hammer on 22-Oct-2022 at 12:10 PM. Last edited by Hammer on 22-Oct-2022 at 12:05 PM. Last edited by Hammer on 22-Oct-2022 at 11:47 AM. Last edited by Hammer on 22-Oct-2022 at 11:44 AM.
_________________ Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68) Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68) Ryzen 9 7900X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB |
| Status: Offline |
| | cdimauro
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 19:02:15
| | [ #418 ] |
| |
|
Elite Member |
Joined: 29-Oct-2012 Posts: 4111
From: Germany | | |
|
| @Hammer
Quote:
Hammer wrote: @cdimauro
Quote:
And the EHB wasn't used even for the two characters' shadows. |
Not my claim. EHB colors are not limited to shadows. |
Let's see when they will be used. Actually the palette isn't that colorful, despite the claim:
Screen mode changed to EHB allowing perfect sprite coloring
It's certainly not perfect, looking at the SNES version: https://www.youtube.com/watch?v=GPP_RcMdqW4 Quote:
Quote:
Only the floor one is barely noticeable. |
Work in progress, the parallax background movement concept is working. |
OK, then it's better that the description is changed, because reading it looks like that all claims are valid. Quote:
Quote:
wonder how they are using the Fast RAM, since this game almost entirely dominated by Chip-mem access & Blitter usage.
|
Apparently, the game's frame rate is smoother with a faster CPU and Fast RAM or on the stock A1200. |
Have you checked on WinUAE? Because emulation isn't cycle-accurate on everything which isn't a 68000 with OCS/ECS. Quote:
I configured WinUAE with Amiga 500 Plus with 2 MB 16-bit Chip RAM configuration and the SF2 tech demo runs slow. Fast RAM is needed |
Could be accurately checked even with Fast Mem (should be correctly emulated).
P.S. The last image is still big. Even on a smartphone (which usually aren't 4k). |
| Status: Offline |
| | bhabbott
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 22:47:58
| | [ #419 ] |
| |
|
Regular Member |
Joined: 6-Jun-2018 Posts: 453
From: Aotearoa | | |
|
| @Kronos
Quote:
Kronos wrote:
So the 020 could access it's local RAM faster them a 386/486 going through ISA bus.
Shock horror!!! |
Amiga fans. When the Amiga can do something better than a PC, they always minimize it.
Quote:
Now tell us how fast could these x86 CPUs access the RAM on the motherboard? |
Why should I? Do your own research.
Quote:
-> AGA crippled CPU performance -> AGA was a bad idea (but maybe still the best C= could have had at that time). |
Perhaps Commodore should have just stuck a 7MHz 68000 in there, then we wouldn't have Amiga fans whining about 'crippled' CPU performance.
Nobody complained about the 386SX being crippled by having a 16 bit bus and no cache RAM. Instead they praised it for bringing 386 enhanced mode to low-end PCs. Similarly, Commodore brought true 32 bit computing to low-end Amigas. But that wasn't good enough for Amiga fans, who are never satisfied and always want more.
To realize the 020's full potential in the A1200 you simply have to add a RAM board in the trapdoor slot, just like A500 users did. For more advanced computing you will be needing that extra RAM anyway, so the fact that you need it for maximum CPU speed isn't a big deal. By not putting that RAM on the motherboard, the price was reduced and compatibility increased, and the options for expansion dramatically improved. That meant people could buy an AGA Amiga at the lowest possible price, but with unlimited expansion capability if they wanted it.
Many users were quite satisfied with the stock machine, which was at least twice as capable as an A500+ in CPU performance, memory, and graphics features. Install a 4MB RAM board and it was at least 4 times better. Put an accelerator card in it and the possibilities are limitless.
But of course none of that was good enough for Amiga fans. It had to have the processing and graphics power of a fast 486DX, along with 16 bit audio (with 32 voice synth), an HD floppy drive, DMA hard drive, and all the same games that were coming out on the PC - at a price well below the cheapest 386SX. And even then they would find something to complain about (...it won't run my entire collection of pirated A500 games!). The A1200 wasn't the most perfect home computer possible, but what was? None of them. Could you have done better? I doubt it. We are lucky that we got what we did. Just look at other 'famous' home computers of the day to see how bad it could have been.
"Oh but look at the PC!", you say enviously. Commodore made those too. In my shop you could have had a 386SX-16 with 2MB RAM, 80MB hard drive and VGA monitor for the same price as an A1200 with 2MB RAM, 40MB hard drive and 1084 monitor. No sound card or speakers in the PC, but hey - they're a cheap upgrade, right? And you can upgrade the RAM on the motherboard to 16MB too with the right SIMMs - but not the CPU, which is forever stuck in 16 bits. But you don't care about that - it can run Microsoft Windows 3.1, and Doom! (in a postage stamp sized window).
Last edited by bhabbott on 22-Oct-2022 at 10:51 PM.
|
| Status: Offline |
| | Karlos
| |
Re: How good or bad was the AGA chipset in 1992/1993. Posted on 22-Oct-2022 22:56:44
| | [ #420 ] |
| |
|
Elite Member |
Joined: 24-Aug-2003 Posts: 4620
From: As-sassin-aaate! As-sassin-aaate! Ooh! We forgot the ammunition! | | |
|
| @bhabbott
Quote:
Install a 4MB RAM board and it was at least 4 times better. Put an accelerator card in it and the possibilities |
Obviously I've said it before but if the machine had shipped with 1MB of fast memory on board (in conjunction with 2MB of chip), this would have been a little beast of a machine in 1992._________________ Doing stupid things for fun... |
| Status: Offline |
| |
|
|
|
[ home ][ about us ][ privacy ]
[ forums ][ classifieds ]
[ links ][ news archive ]
[ link to us ][ user account ]
|