AmigaOne X1000/X5000 Remains Well Supported With PowerPC Linux 117
Mike Bouma writes: Despite being expensive and having been sold out for quite some time at the main Amiga Dealers, two days after Linus Torvalds' release of Linux 5.18, Christian "xeno74" Zigotzky made the latest PPC kernel available for the AmigaOne X1000/X5000. Here and here are some screenshots. Linux PPC performs well on AmigaOne computers. For example, here is a 5-year-old YouTube AmigaOne X5000 demonstration video.
AmigaOne Hardware News : Fifteen years of Sam440e (Score:2)
"Dear customers,
It was the year 2007 when we introduced our first “baby”, the Sam440ep. Fifteen years ago. We designed and produced the Sam440ep, a lovely mini-itx motherboard that many enjoyed running AmigaOS 4, AROS and Linux.
At that time we, a bunch of mad amigans, resolved to do our best to bring this hardware to life for all the Amiga Community because it was in dire need of a new computer platform to run AmigaOS.
Without the Sam440ep, the future of AmigaOS would have been very difficult.
Aft
M68K is still supported by the Linux kernel (Score:2)
Distros maybe not so much.
Re: (Score:3, Funny)
It still out-computes you.
Re: (Score:2)
Yeah, but he only has 640k, go easy on the poor fella ^_^
Re: (Score:2)
It's all FastRAM though...
Re:Just let it go (Score:5, Informative)
Re: (Score:2, Troll)
The original Amigas were great - technically well ahead of their time
It really depends on your perspective. In terms of user experience and what you could 'do' on a home computer - yes. In terms of design not really.
The whole package was a relatively slow compute unit with a bunch of custom peripheral chips strapped to it. The result was a machine that could do a lot of neat stuff but the programing model was complicated. That is why you had a demo scene, doing really cool stuff always meant voodoo. Look where the rest of the industry went in the decade that followed; far m
Re:Just let it go (Score:5, Informative)
What could you do with A4000 in 1992 that a Pentium PC from 1992 could not do as well or better by that point? - I can't think of anything
They used Amiga for the special effects on Babylon 5 in 1993 [youtube.com]. It wasn't until Windows 95 a few years later where PC really started to shine.
and I can't see C= ever catching up again.
No doubt. They've been out of business for 25 years or so.
Re:Just let it go (Score:4, Informative)
It wasn't until Windows 95 a few years later where PC really started to shine.
For more famous Amiga uses, look here:
https://www.amigareport.com/ar... [amigareport.com]
Re: (Score:2)
They used Amiga for the special effects on Babylon 5 in 1993.
For the generation of the 3d imagery, anyway... and only on the pilot [midwinter.com], where 16 A2000s with Fusion 40 accelerators did the rendering. After that they moved to PCs and Alphas for the 3d stuff, although ISTR that some people kept modeling on Amigas.
Re: (Score:2)
Re: (Score:3)
What could you do with A4000 in 1992 that a Pentium PC from 1992 could not do as well or better by that point?
Display more than 256 colors reliably. Play MIDI music that didn't sound like shit.
The parts of it that were "evolutionary dead-ends", as you call them, were not caught up with until PCs began using those same evolutionary dead-ends.
It was quite a bit later that you started getting CPUs powerful enough to execute software audio codecs that didn't suck, or rasterize graphics at anywhere near the performance levels of the Amiga.
You're confusing board peripherals as somehow being less custom than those th
Re:Just let it go (Score:4, Interesting)
It was quite a bit later that you started getting CPUs powerful enough to execute software audio codecs that didn't suck, or rasterize graphics at anywhere near the performance levels of the Amiga.
Of course it wasn't just that, it was also the graphics acceleration. When the Amiga came out, it was literally the only platform with anything like it. Everyone else either had a pure dumb frame buffer or only hardware that would do a couple of dumb tricks. Along came the Amiga with this fancy-ass architecture that sped up a bunch of bottleneck tasks, and it punched far above its class.
But sometime in the Windows 3.1 days of ISA-bus PCs the PC started to get accelerated graphics cards that would not only accelerate all the same stuff the Amiga would (for example, it was common for them to have hardware to accelerate rectangular copies) but also accelerate other common graphics tasks like drawing shapes and even font rendering. So even old ass PCs would offer excellent capabilities if you just had a decent graphics card... which probably cost more than the Amiga.
By the later days of the 486, though, even most budget PCs had all that stuff plus accelerated audio with sound fonts and such, so the Amiga's superiority is often grossly exaggerated.
The only evolutionary dead-end about the Amiga was the 68k- but at the time, it was a good choice. Motorola simply never stood a chance in the long run.
They should have found a partner to share the burden earlier. At one time they were the de facto standard for anyone who wanted to be taken seriously. Most of the Unix vendors produced 68k family systems at one point, they had a great name and reputation.
Re: (Score:3)
Of course it wasn't just that, it was also the graphics acceleration. When the Amiga came out, it was literally the only platform with anything like it. Everyone else either had a pure dumb frame buffer or only hardware that would do a couple of dumb tricks. Along came the Amiga with this fancy-ass architecture that sped up a bunch of bottleneck tasks, and it punched far above its class.
It's important to keep what that class is firmly in mind.
The Amiga came out in 86 (I got mine in 87).
It didn't just punch above its class, it re-defined the class. Contemporary PCs and Macs of the era were plotting along with low resolution 4-16 color graphics adapters.
But sometime in the Windows 3.1 days of ISA-bus PCs the PC started to get accelerated graphics cards that would not only accelerate all the same stuff the Amiga would (for example, it was common for them to have hardware to accelerate rectangular copies) but also accelerate other common graphics tasks like drawing shapes and even font rendering. So even old ass PCs would offer excellent capabilities if you just had a decent graphics card... which probably cost more than the Amiga.
That's mostly a matter of CPU. By the time of the 486, the PC was simply radically faster than the 68k. By multiples.
SVGA still used a singular framebuffer with no dedicated blitter hardware. You could DMA blocks of framebuffer around, but
Re: (Score:3)
The Amiga came out in 86 (I got mine in 87).
It didn't just punch above its class, it re-defined the class. Contemporary PCs and Macs of the era were plotting along with low resolution 4-16 color graphics adapters.
The Macintosh in particular is a tragedy of a story when it comes to graphics performance. Despite consistently being the most expensive hardware for a given level of performance they had no graphics acceleration whatsoever (doing absolutely everything in software) until the release of the $1340 8*24 GC NuBus card for the Macintosh II series, and didn't sell machines with any hardware graphics acceleration features onboard until the Quadra line.
sometime in the Windows 3.1 days of ISA-bus PCs the PC started to get accelerated graphics cards that would not only accelerate all the same stuff the Amiga would (for example, it was common for them to have hardware to accelerate rectangular copies) but also accelerate other common graphics tasks like drawing shapes and even font rendering.
That's mostly a matter of CPU
No, I mean the chip on the video card actually had hardware ac
Re: (Score:3)
No, I mean the chip on the video card actually had hardware acceleration which would outright perform many operations on the card that the Amiga had to do with the CPU, or with the CPU+Agnus, while the PC was thinking about something else. Even later ISA cards would do this, but by the time VLB rolled around (with bus bandwidth that absolutely smashed the Amiga, though the PC had always had a faster bus really) it was standard to have extensive graphics acceleration on the video card. A high-end 68k was just as fast as a 486, even faster at some operations. Even sticking with just Commodore stuff you could eventually get a 68040 in an Amiga, and it compared pretty well, and the cost remained reasonable. It wasn't until the Pentium class vs the '060 where the Amiga got whupped in processor power vs. price. An Am586 in particular would absolutely spank an '060 on that basis.
I was very active in video programming at the time.
I cannot recall any existence of any accelerator functionality prior to VBE. Do you have a citation?
SVGA only ever offered a flat or planar framebuffer to play with, and propriety methods to bootstrap it.
The 68040 was no comparison for the 486. It ran hotter and slower.
Re: (Score:2)
I cannot recall any existence of any accelerator functionality prior to VBE.
I remember several cards which advertised hardware acceleration features under Windows. They would improve stuff like window dragging. ISTR one of them even advertising a hardware bitblit. I am searching around for cites but I don't exactly have a stash of antiquated advertising. that's not my bag baby. But here is a brief stackexchange discussion about it [stackexchange.com]. One card I remember with windows acceleration was the one that came with the radius color pivot [sheley.org]. ATI Mach32 and Mach64 appeared in ISA variants, and off
Re: (Score:2)
Many cards were built with capabilities and drivers that Windows could use with its GDI.
From the protected-mode DOS perspective, these features simply did not exist.
There wasn't an API to use them. VESA was too limited.
Many operations completed more quickly on the '040
I can't think of one.
The 486 was 20% faster per-clock.
At the time, nobody I knew could figure out why. It was suspected that it was due to overhead from having the separate MMUs for instructions an
Re: (Score:2)
Alright, where you're coming from is fair. You're referring to from a Windows perspective.
Most importantly just a non-DOS perspective. If you ran Linux or SCO Unix or OS/2 or NT on machines with these cards then you could have graphics acceleration. If you ran DOS software, you could only rarely have support even for fancy color modes, and they had to be individually supported. But Windows gaming grew fairly quickly, and Microsoft developed the WinG API for doing fast graphics from Windows — which could importantly be mixed with the GDI interface. So you could have your fast bitmap graphic
Re: (Score:2)
Most importantly just a non-DOS perspective.
Also fair.
But Windows gaming grew fairly quickly, and Microsoft developed the WinG API for doing fast graphics from Windows
I don't think I played a Windows games that used any kind of graphics until Quake for Windows (WinQuake/QuakeWorld? I can't remember) in like 97 or 98. Prior to that, the most advanced Windows game I saw was Spaceward Ho!.
Re: (Score:2)
I don't think I played a Windows games that used any kind of graphics until Quake for Windows (WinQuake/QuakeWorld? I can't remember) in like 97 or 98. Prior to that, the most advanced Windows game I saw was Spaceward Ho!.
There is a partial list of games and apps using the WinG API [wikipedia.org] on WP, and here's another list [pcgamingwiki.com].
Re: (Score:2)
Re: (Score:2)
The 68040 could compete with the original 486 (ie the first DX/SX models), but whereas Intel kept developing (eg the later DX2, DX4 models) Motorola largely abandoned m68k and focused on PPC.
The 68040 topped out at 40mhz, and although there was a 68060 they didn't put much effort into its development, so it was late to the party and disappointing.
Sun, SGI and others abandoned m68k after the 68030, Apple & Next abandoned them after then 68040, aside from some embedded uses the Amiga was pretty much the o
Re: (Score:2)
The 486DX achieved +20% Drhystone MIPS per clock than the 68040.
+20% isn't generally regarded as competitive.
Re: (Score:2)
It depends on the benchmarks, the rest of the platform, compiler and the OS being used...
For instance the Dhrystone and Linpack scores quoted here show the 68040 outperforming the 80486 (both clocked at 25mhz) quite considerably:
https://www.nox-rhea.org/obsol... [nox-rhea.org]
An Amiga would also generally outperform a Mac with the same processor (you could even emulate macos on the amiga and achieve slightly better performance in cpu benchmarks than an equivalent mac using the same cpu).
Modern accelerator cards are still
Re: (Score:2)
The 486DX achieved +20% Drhystone MIPS per clock than the 68040.
+20% isn't generally regarded as competitive.
And MIPS aren't generally regarded as a worthwhile reflection of real-world performance, which is why the acronym is commonly expanded to "Meaningless Indicator of Processor Speed". One of the big problems with x86 is that it has no true general-purpose registers, and very few of them to boot, which means you waste a lot of time and cycles dicking around moving things from register to register until the PPro came along with register renaming, at which point register starvation is only a problem for the asse
Re: (Score:2)
This isn't to detract from the 68040, it's just that those benchmarks are poorly matched.
They'd need to have put that 65040 25/50 up against a 486DX2-50
Re: (Score:2)
And MIPS aren't generally regarded as a worthwhile reflection of real-world performance, which is why the acronym is commonly expanded to "Meaningless Indicator of Processor Speed". One of the big problems with x86 is that it has no true general-purpose registers, and very few of them to boot, which means you waste a lot of time and cycles dicking around moving things from register to register until the PPro came along with register renaming, at which point register starvation is only a problem for the assembly programmer who still has to think about the lack of GPRs. So classic x86 tends to shit the bed when the chips are down, so to speak; it does well only on the simplest benchmarks, where its unfortunate architectural limitations do not come into play.
"MIPS" in general is a meaningless term.
However, Dhystone MIPS have a very real, useful meaning. As for GP registers, there's no evidence in existence that number of GP registers correlates with performance.
The '040 has significant advantages, not least more registers (which are also truly general purpose.) That's why the '040 beats the Dhrystone int and fp performance of the 486 clock for clock [preterhuman.net] even though it shows less Dhrystone MIPS, with almost double the integer performance, and three times the performance on the linpack benchmark. In typical use the difference is anywhere from 20-50%. The 68040 conclusively outperforms the 80486 clock for clock, pretty handily in int math but soundly when it comes to fp.
That is a comparison between a 50Mhz 68040 and a 25Mhz 486DX. ;)
Hardly fair
Re: (Score:2)
"MIPS" in general is a meaningless term.
However, Dhystone MIPS have a very real, useful meaning.
The existence of other Dhrystone benchmarks which present dramatically different results to Dhrystone MIPS suggests otherwise. At minimum it is not useful for comparisons of processor performance. DMIPS' correlation to actual performance varies from CPU to CPU, which makes it essentially worthless as a measurement of real performance. Though it might well be argued that none of the Dhrystone benchmarks are especially great today, which is why we have ever-updated versions of the SPEC benchmarks. Of course,
Re: (Score:2)
The existence of other Dhrystone benchmarks which present dramatically different results to Dhrystone MIPS suggests otherwise. At minimum it is not useful for comparisons of processor performance. DMIPS' correlation to actual performance varies from CPU to CPU, which makes it essentially worthless as a measurement of real performance. Though it might well be argued that none of the Dhrystone benchmarks are especially great today, which is why we have ever-updated versions of the SPEC benchmarks. Of course, those updates make comparisons to old tests invalid, but they still seem necessary.
No, they don't. A Dhrystone MIPS represents a specific type of integer workload. It doesn't say how fast a processor it, it says how fast it does that workload.
That's a perfectly meaningful measurement, since it's a specific unit of work.
Nonsense. The lack of GPRs inhibiting performance is a well-known [phatcode.net] phenomenon [wisc.edu] known as "register starvation". And if it wasn't a problem, register renaming wouldn't provide nearly as much of a performance benefit. This problem crops up even when doing mundane things like hashing [google.com].
How did you manage to provide so many links without reading any of them?
Register count does not correlate with performance except in identical architectures. The only thing it implies is more required instructions.
That may, or may not result in a performance decrease, o
Re: (Score:2)
That's a perfectly meaningful measurement, since it's a specific unit of work.
It isn't for reasons described in the link. It is affected by many factors.
Register count does not correlate with performance except in identical architectures
There is literally zero question that x86 suffered from register starvation-related performance inadequacy due to inadequate GPRs.
All 68040s run at a 2:1 pclk:bclk ratio.
Groovy. They weren't sold that way. They weren't described that way. They didn't compete that way. Now you're just picking nits.
Re: (Score:2)
It isn't for reasons described in the link. It is affected by many factors.
No, your link describes no such thing. It tells us something everyone already knows, and that I said many posts ago- DMIPS isn't a good indicator of "performance", but it does show how well a certain processor does a certain (useful to know) job.
There is literally zero question that x86 suffered from register starvation-related performance inadequacy due to inadequate GPRs.
This logic is fucking brainded. Any cpu with less than infinite registers suffers from register starvation-related performance inadequacy due to inadequate GPRs, because that bar can be pushed as far as you want. At the end of the day, it has zero relevance unless a
Re: (Score:2)
Any cpu with less than infinite registers suffers from register starvation-related performance inadequacy due to inadequate GPRs, because that bar can be pushed as far as you want. At the end of the day, it has zero relevance unless all things are equal.
At the end of the day, we literally know that register starvation is a problem during common tasks on x86 processors — especially prior to the PPro, but frankly even after that it's been known to affect performance. It's literally been studied extensively, and is well known, and is a big reason why AMD put [more] GPRs into amd64.
Why did Motorola market the BCLK, and Intel market the PCLK? No fucking idea. Maybe Motorola thought it would make their processors seem like they were doing more work with less cycles. Mission accomplished, eh? Maybe Intel was decided to go the route of "bigger numbers will sell better shit!" (as we know they did).
It was common to use the bus speed at the time, for whatever reason. It was probably because of what it implied for actually being able to feed data (meaningfully including in
Re: (Score:2)
At the end of the day, we literally know that register starvation is a problem during common tasks on x86 processors — especially prior to the PPro, but frankly even after that it's been known to affect performance. It's literally been studied extensively, and is well known, and is a big reason why AMD put [more] GPRs into amd64.
As I said, it's a problem on every processor. The bigger the register file, the better your micro-ops can be parallelized in the pipeline.
But it's meaningless between architectures. Completely meaningless.
IA-64 had 128 64-bit GPRs.
aarch32- 13!
aarch64- 32!
None of these devices outperformed contemporary x86 machines (well, until recently- Thanks, Apple, but that was a one trick pony. Alder lake has nearly closed the gap, and Zen4 is set to match or beat Alder Lake)
That's why I said it only matters, wit
Re: (Score:2)
You simply cannot say: "Arch A has more GPRs than Arch B, it's therefor faster, clock for clock."
The actual essence of what I said, if read and understood correctly, is that x86 would be faster if it had more GPRs — and what that implies is also whatever architectural changes are necessary to make that happen, but nothing else. And we know it to be true because of the time spent dicking around with the results of x86 not having any GPRs whatsoever, and precious few registers in general. x86 effectively improved the GPR ratios over time with newer instructions which were slightly less picky about
Re: (Score:2)
Otherwise we're back to, "all systems are held back by the fact that they use DRAM instead of flatly-addressed SRAM register files".
There are real-world considerations. More registers doesn't equal better in all cases.
If for example, you rely on tight fixed-word instruction encodings, you have to pick an amount of bits that gives you a good amount of registers, and a good amount of instructions. And there are costs to increasing both.
The formula for performanc
Re: (Score:2)
The Zorro-3 bus could theoretically perform slightly better than 32bit PCI, and the first implementations of Z3 were similar in performance to the first implementations of PCI (ie nowhere near the theoretical performance limits of the bus).
But Z3 stagnated (The A4000 had the same bus controller as the A3000 which was the first implementation of Z3), whereas PCI got better implementations.
Re: (Score:2)
Of course it wasn't just that, it was also the graphics acceleration. When the Amiga came out, it was literally the only platform with anything like it.
Only if you ignore the Acorn Archimedes.
The Archimedes kicked the Amiga's ass in every single department and it did it in software.
Re: (Score:2)
The Archimedes kicked the Amiga's ass in every single department and it did it in software.
I've never had one so I can't speak authoritatively, but you have a bunch of huge gaps in your Amiga knowledge which you've demonstrated in this discussion so you can't either.
However, the general consensus seems to be that the only thing in the Acorn unequivocally better than in the Amiga was the CPU, and the only thing the system did graphically superior to the Amiga was drawing filled polygons, because the Amiga didn't have any acceleration for that. The later Amigas had more powerful processors than the
Re: (Score:2)
The Archimedes kicked the Amiga's ass in every single department and it did it in software.
I've never had one so I can't speak authoritatively, but you have a bunch of huge gaps in your Amiga knowledge which you've demonstrated in this discussion so you can't either.
I've had both, so yes I can. The Archimedes totally kicked Amiga butt. It failed because of timing/marketing, but that's another story.
There may be gaps in my Amiga 2000 knowledge but that because I'm not an Amiga fanboy and never owned/used one. I did write commercial games on Amiga A500s back in the day because that's the Amiga that everybody had. The Amiga A500 hard disk situation was so bad that we use to edit/assemble the program on Atari STs and send the binaries over to the Amigas for running via the
I might add (Score:2)
Since why not, and I seem to have written something about this but edited it out... my 386 (due to late-life upgrades) had 24-bit color at VGA res, a RAMDAC that would do at least XGA (maybe more with fiddling), 1MB of zero-waitstate linearly addressed graphics RAM (and zero-waitstate bus communication) and really quite decent performance. It was my first Linux box. 8MB DIP DRAM and 120MB of ATA disk was enough to install about half the system (including the whole devel set) and run X and Netscape...
Re: (Score:3)
had 24-bit color at VGA res
Sure, that nearly nothing could use.
You could try to play with the after-market VESA drivers to get a standard SVGA frame buffer, but it rarely worked consistently.
End of the day, that fucker was using all its horsepower to blast Commander Keen onto the screen, while the Amiga was blowing literal fucking minds with Turrican.
1MB of zero-waitstate linearly addressed graphics RAM (and zero-waitstate bus communication) and really quite decent performance.
That was framebuffer RAM, and it was addressed via the ISA bus, so the waitstates were pretty irrelevant, since the slowest DRAM in the world was faster than the ISA bus.
It was my first Linux box.
That's good- t
Re: (Score:2)
1MB of zero-waitstate linearly addressed graphics RAM (and zero-waitstate bus communication) and really quite decent performance.
That was framebuffer RAM, and it was addressed via the ISA bus, so the waitstates were pretty irrelevant, since the slowest DRAM in the world was faster than the ISA bus.
It made a big performance difference over other ISA VGA cards in practice, the 8900D is widely regarded as one of the best-performing of these simpler cards and it can produce literally double the performance of many others. But the more expensive cards that did more were in practice obviously faster... for Windows, where you had a standardized driver etc.
That's good- then you were able to at least use the SVGA capabilities of that card. Games couldn't.
Not without programming to the "specific" card, no. But that was a common thing to do back when, and you'd run your little config program to select your v
Re: (Score:2)
I've owned A500, A1200, A2000, A2500 (separate machine) and A3000. Put fatter agnus and an '020 in my 500. I've been around the Amiga block. I've just been around most all of the other blocks as well :) (no RiscOS...)
As have I, and I still never saw a game that came close to comparing to my Amiga until probably Doom.
And even then, when you get right down to it, Doom was only impressive because of the CPU. The actual graphics requirement was nothing. Just rasterizing to a flat VGA framebuffer. But still- the Amiga simply didn't have the horsepower to do that.
Re: (Score:2)
There are and were Doom ports to Amiga [fandom.com]. ADoom [doomwiki.org] was the earliest credible port, and it will run on any "Amiga with AmigaOS 3.0+ along with a 020 processor and 6MB of RAM", although I understand framerates to be pretty terrible on those. The first machine I played much Doom on was a 486SX25 which managed it tolerably well... at ~320x240.
Re: (Score:2)
Yes, ID Software once claimed in an interview Doom couldn't be done on the Amiga. It got ported in days on Amigas older than PCs are able to run Doom after the source code got released. On most unexpanded PCs Doom ran also pretty terrible.
Doom running on an A1200 (AGA) / 68030 / 50 Mhz compared to an AMD 386 DX 40 MHZ with Tseng Labs ET4000 Isa video Card (very similar in performance, so ID Software was wrong):
https://www.youtube.com/watch?... [youtube.com]
Quake was even ported to 68k/AGA. With PowerPC and RTG upgrade ca
Re: (Score:2)
The difference is that modern computers have software abstraction layers to allow the same software to run on wildly different underlying hardware...
Systems before the Amiga generally didn't have any such abstraction, and required you to program the hardware directly.
The Amiga had both, but the abstraction layer (even as lightweight as AmigaOS was) had a significant impact on performance. With later systems the performance hit of the abstractions are largely ignored because very few people ever try to explo
Re: (Score:3)
I find that fans of the non-pc brands of this era always seem to forget what PCs were truly capable of. I don't know if that's just general ignorance of the platform and market or if its an assumption that PCs of the time worked like everything else and they conveniently forget its biggest selling point was its expandability.
Most definitely not.
The PC had arguably the best midi soundcard in the Roland Sound Canvas which launched in 1991. But it also had wavetable based synthesis available in the Gravis Ultrasound.
Ya- Roland was amazing. So was the Sound Blaster AWE32.
Bot both of those didn't come unto being until the Amiga was old kit.
As for 256 colors, meet the ATI Mach 32. It was a 2D accelerator capable of 24bit color in 1992.
Your Mach 32 could do 24-bit color, but no game could ever use it due to a lack of sufficient framebuffer, and you're comparing it to a machine that came out 6 years earlier.
Amiga's tight integration of first party ASICs to the platform is definitely an evolutionary dead end for the PC. The platform became generic and got its power from expandability. Who cares about some chip called linda or steve when I can buy this year's hot new sound card?
Speaking of ignorance.
Amigas did that too.
The evolutionary dead-end was the 68k, not the availability of expansion buses, which existed on all Amigas after the 1000.
Re: (Score:2)
We were talking about 1992. Don't move the goalposts.
What the Amiga could do n years before is irrelevant.
Amigas did that too.
With an expansion bus that relative to ISA, nobody supported? What kind of support did Zorro really have? ATI didn't support it. Neither did Soundblaster, or Adlib, or Roland. For every product on zorro there were five for PC.
Re: (Score:2)
We were talking about 1992. Don't move the goalposts.
I didn't, jackass.
They did when they used a term that only made sense if evaluated from a former era.
They said:
I find that fans of the non-pc brands of this era always seem to forget what PCs were truly capable of.
He clearly isn't talking about 1992, because in 1992, PCs were flatly superior to Amigas, and nobody alive thinks otherwise. Amiga nostalgia refers to the mid 80s, not the 90s.
With an expansion bus that relative to ISA, nobody supported? What kind of support did Zorro really have? ATI didn't support it. Neither did Soundblaster, or Adlib, or Roland. For every product on zorro there were five for PC.
Of course there were.
That's because for every Amiga there were 100 PCs, and that's because of IBM, no technical merit of the machine.
Computing magazines called the fight as soon as IBM started shipping x86 machines, and
Re: (Score:2)
for every Amiga there were 100 PCs, and that's because of IBM, no technical merit of the machine.
The IBM PC had technical merit where it counted, in the bus. Even in the PC and XT the use of an already-standard bus was a big benefit, but when the AT rolled around with its 16-bit ISA bus it was all over. They didn't have any autoconfiguration, and that was annoying, but it was actually of little import since most people weren't swapping expansion cards frequently. But interestingly, the 16 bit ISA bus is actually faster than the Zorro II bus. And while on paper it can theoretically keep up, Zorro III is
Re: (Score:2)
The IBM PC had technical merit where it counted, in the bus.
How do you figure?
Even in the PC and XT the use of an already-standard bus was a big benefit, but when the AT rolled around with its 16-bit ISA bus it was all over.
Again, I ask, how do you figure?
They didn't have any autoconfiguration, and that was annoying, but it was actually of little import since most people weren't swapping expansion cards frequently. But interestingly, the 16 bit ISA bus is actually faster than the Zorro II bus.
In what way?
The Zorro II bus had bus-mastering DMA, something the PC didn't have until PCI.
Then they had Zoro III, a full 32-bit bus-mastering DMA bus, 2 years before a PC shipped with PCI.
Zorro II did have a major pitfall in that it was locked to the CPU clock, so on some devices, you could get as low as 4MB/s, but on some devices you could get as high as 8MB/s. ISA was good for a fraction of that, but ran at a constant bus speed. And no bus-mastering.
Re: (Score:2)
the 16 bit ISA bus is actually faster than the Zorro II bus.
In what way?
In the really real world. Whether you're talking about synthetic benchmarks or just real world performance observations, the Zorro buses are slower than their PC competition.
Zorro II was superior to ISA in every way.
Except performance.
Even the original PS/2 didn't use ISA, it used the superior MCA.
So what? The MCA bus is an also-ran. Outside of business statistically nobody had one, because PS/2s were stupidly expensive. I've in fact only ever seen many of them in government and education, they were too expensive for everyone else.
Later on, with EISA, PC buses became competitive, performance wise, but not before that.
False [amiga.org]. The Zorro buses had really only two practical advantages; one was autocon
Re: (Score:2)
In the really real world. Whether you're talking about synthetic benchmarks or just real world performance observations, the Zorro buses are slower than their PC competition.
Not even remotely. [vcfed.org]
I seriously have no idea what you're going on about.
Pre-EISA ISA was the slowest bus of any computing platform in circulation.
False [amiga.org]. The Zorro buses had really only two practical advantages; one was autoconfiguration, and the other (which was relevant only on Amiga 1000 and 500) was that it lent itself to daisy chaining, so you could make a machine with zero slots which could have more than one expansion added. I suppose you could also call the cheaper connector an advantage (when compared to EISA, MCA, or VLB) but that's small comfort. The only thing we Amigans really got to brag about back in the day when it came to the bus was the autoconfig, until PCs with PCI came out with working interrupt sharing and autoconfiguration. Macs of course always had autoconfig on NuBus, but they also had an expensive and fragile connector, and NuBus performance is nothing to brag about either.
That link doesn't show what you think it shows.
You're not wrong that Zorro 2 was slow. Where you're wrong is this bizarre idea that ISA was fast(er).
It didn't support bus-mastering, meaning all transfers had to be coordinated with the CPU, so you either used PIO or CPU-programmed DMA, and to even get above 2MB/s required highly aligned access and large enough
Re: (Score:2)
You're not wrong that Zorro 2 was slow. Where you're wrong is this bizarre idea that ISA was fast(er).
Except it was.
It didn't support bus-mastering,
hahaha no, the 16 bit ISA bus supports bus mastering [wikipedia.org].
Re: (Score:2)
Except it was.
By the compiled list of best real-world achieved speeds in links that we both provided, Zorro II was faster. I away an example from you of something saying otherwise.
hahaha no, the 16 bit ISA bus supports bus mastering [wikipedia.org].
Not really. A device on the ISA bus could force a bus master, but nothing ever did it because it was very dangerous. There were no safety mechanisms on the bus to avoid contention, and as such, if the OS get every other card on the bus to agree not to talk while it was happening, you could end up with any other card on the system having its wri
Re: (Score:2)
Some very high end SCSI cards did this.
Some very mid range SCSI cards did this too, I had an Adaptec 1540CP which did bus mastering.
The fact that you're not familiar with this tells me you don't actually know what bus mastering is.
It doesn't take a rocket surgeon to understand that bus mastering is DMA initiated by an add-in card instead of by whatever controls the bus, dude. This isn't secrets of the luftwaffe or whatever. The fact you think that features that were used weren't used tells me you know a lot less than you think you do. This conversation has borne that out repeatedly.
Re: (Score:2)
Some very mid range SCSI cards did this too, I had an Adaptec 1540CP which did bus mastering.
I'm quite sure that's literally one of the most expensive ISA SCSI cards you could possibly own at the time.
It doesn't take a rocket surgeon to understand that bus mastering is DMA initiated by an add-in card instead of by whatever controls the bus, dude. This isn't secrets of the luftwaffe or whatever. The fact you think that features that were used weren't used tells me you know a lot less than you think you do. This conversation has borne that out repeatedly.
You're right, it doesn't take a rocket surgeon.
However, someone with a little more knowledge might roll their eyes at the characterization of ISA's "bus mastering", with the bus mastering available on Zorro, VLB, and PCI buses, since it requires cascaded 8237s to block the bus- meaning it's quirky at best, and won't even work on all motherboards, or may crash other cards on the bus. As I said, it's
Re: (Score:2)
I'm quite sure that's literally one of the most expensive ISA SCSI cards you could possibly own at the time.
If it were, I wouldn't have owned it, because I have never bought the most expensive kit, literally. But I probably got it well after its introduction, because as someone who was quite poor back in those days I often used elderly hardware. Until I was in my 20s my A500 was the only new computer I ever owned.
As I said, it's not a real feature of the bus.
It was designed in on purpose so that's questionable, and it doesn't work reasonably with multiple masters either so it's not that wonderful, but it's still possible to get over 5MB/sec on the bus.
Re: (Score:2)
It was designed in on purpose so that's questionable, and it doesn't work reasonably with multiple masters either so it's not that wonderful, but it's still possible to get over 5MB/sec on the bus.
DMA is a feature of the bus (via the 8237). You can misuse 2 8237s to emulate something approaching bus mastering, but you will not find it in the ISA specification [bitsavers.org], because the bus wasn't intended to be used that way.
Used normally, DMA is fixed to 4.77Mhz, period, no exceptions, and it's slower than doing PIO (which is why the XT uses PIO for its BIOS disk access instead of DMA, since the CPU in PIO can actually do ~6Mhz of bus transfers)
Re: (Score:2)
Used normally, DMA is fixed to 4.77Mhz, period, no exceptions
Yes, and this trait followed all the way to the MCA and EISA buses, although they can do 32 bit transfers. But that's another reason why those buses were not as good as they want to be on paper. Expensive and slow at DMA is just not a good epitaph. The ISA bus was cheap and surprisingly good at the time. It doesn't matter if it took weird hacks to make it work if you're comparing to the Amiga, which is basically a collection of hacks, and which required extensive hackery to get the maximum performance as we
Re: (Score:3)
For a lot of software, the custom chips on the Amiga meant that it was actually more effective CPU wise. Point in case, emulating Apple systems were faster than actual Apple systems with the same specs, especially for graphics editing, music etc. I/O was also a lot faster.
Speaking about buses, the Zorro buses heavily influenced PCI. As for Pentium, the comparison would be the 68060, which featured on some accelerator boards. And an A4k with a 68060 board and a Picasso card still cost a bit less than a Penti
Re: (Score:2)
The result has surprised me - a lot. The 286 never got a lot of love at the time but in this configuration, it's not bad at all. Games are snappy and the sound is pretty good. I've compared Ultima IV between the t
Re: (Score:2)
What could you do with A4000 in 1992 that a Pentium PC from 1992 could not do as well or better by that point?
Well you'd probably be $2000 better off. Tricking out a Pentium PC with Soundblaster Pro / 16, decent SVGA card, & memory expansion necessary to make a comparable product would not have been cheap.
But in reality Commodore was dying by that point. I actually rang a dealer back then to buy an Amiga A4000 and discovered Commodore had jacked up the price by £100 and so I bought a PC instead. I guess I should thank Commodore since they went bust not long after. The PC I bought for the same money was
Re: (Score:2)
My thoughts exactly, wanting to play with newer versions of AmigaOS purely for the novelty value is the only reason you'd buy one of these machines.
If you want a Linux box, there are many faster and cheaper options out there.
If you want PPC Linux specifically, a used PowerMac is dirt cheap.
Or if you want to play with AmigaOS, you could buy something like a Vampire which implements a fast m68k and amiga style chipset in an FPGA, giving you compatibility with the original software.
Re:Just let it go (Score:5, Informative)
Where it really failed was in support for hard disks, etc. An Atari ST worked by just connecting a hard disk to the SCSI connector on the back and it booted instantly. Amiga hard disks required you to boot from floppy and jump through all sorts of hoops to try to get it to boot quickly.
This is mostly false. The Amiga 1000 needed a kickstart floppy, and it's true that some disk controllers wouldn't boot the Amiga, so you had to have a boot floppy. But by far, most Amiga hard disk controllers would boot. You could put a rom on the device, put a driver in the rom, the Amiga was microkernel based so it loads the driver and boom it knows how to boot from your HDD regardless of whether it's SCSI or what. I had a Commodore SCSI/MFM board in an A2000 that would boot from SCSI but not from MFM, so before I got a SCSI device I had to boot from a floppy, but I got a 44MB syquest eventually and I was able to boot from that. Before that I was using a 21MB ST-225 half-height MFM disk...
Even having to boot from floppy, literally only the boot sector was loaded, then the disk was read, the startup-sequence was loaded. Then you'd load the device driver, make a couple of assigns, and execute the startup-sequence on the HDD. The process was easy to implement, and practically nothing was read from the floppy except for your device driver.
Re: (Score:2)
All that as opposed to... connecting a cable to the back of the machine? Simple!
(Oh, yeah, the Amiga expansion connector where the hard disks went. Don't get me started on that)
Re:Just let it go (Score:5, Informative)
Oh, yeah, the Amiga expansion connector where the hard disks went. Don't get me started on that
Don't get you started on the Amiga having a processor speed full width expansion bus that could be used for hdds, memory expansions, sidecars, etc?
You do know that there were other Amiga models, some with HDD controllers on the motherboard, right? And others with whole rows of expansion slots?
Re:Just let it go (Score:5, Insightful)
You really can't compare an Amiga 1000 to an Atari ST. First, the ST is 8 years newer, and laughably inferior in multimedia capabilities.
Re: (Score:2)
The first ST (the 520ST) beat the first Amiga (the A1000) to market by a month, but it was a much simpler design with significantly worse performance in many areas.
It did not have SCSI, it had a proprietary "ACSI" port for connecting an Atari specific drive (or other peripherals) to.
Re: (Score:2)
Re: (Score:2)
Yeah, I meant ACSI.
It was a SCSI-like thing with full DMA and it was on the back of every single Atari ST.
Re: (Score:2)
The ST was released in 1985, the same year as the Amiga.
Yes, it was. I should have clarified. He mentioned direct SCSI port, which didn't exist on the ST line until the Falcon, so I assumed that's what he was talking about.
Reading other comments, I see he was confused, and meant the ACSI port.
So you just lied your mother fucking ass off, claiming the ST came out... in 1993?
A misunderstanding.
What kind of habitual fucking liar thought that would sell? The DamnOregonian (963763) thats who.
OK, you're a fucking moron.
and the Apple IIgs had the best audio, and of all the tech gaps of the era, this was the biggest... the primary things amiga boy bragged about, actually fucking sucked with its only 4 channels and shitty sample rate, compared to the IIgs's 16 channel Ensoniq DOC2 chipset.
The fuck?
How stupid are you?
You're conflating the PCM capabilities with the synth capabilities. Fucking adorable.
The Amiga had no wavetable synthesis capabilities (at least initially). What it has 2 hardware DMA PCM streams,
Re: (Score:2)
Yes, it was. I should have clarified. He mentioned direct SCSI port, which didn't exist on the ST line until the Falcon
Wrong: https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
Re: (Score:3)
The Amiga had no wavetable synthesis capabilities (at least initially). What it has 2 hardware DMA PCM streams, so that it could play sampled music.
Actually 4, but who's counting?
(two on the left and two on the right)
Re: (Score:2)
It wasn't really 4 entirely separate channels, though you could certainly synthesize the data as such.
It was 2 hardware DMA units that read 2-channel interleaved stereo.
Re: (Score:2)
It was 2 hardware DMA units that read 2-channel interleaved stereo.
Go check again: It has 4 channels, no interleaving.
It does read bytes in pairs on each channel so your waveforms had to be even numbers of samples in length.
If you go look for videos of "Micky Mouse on the Amiga" you'll hear the music is a bit clicky/noisy. That's because it was my first time programming the Amiga and I didn't know about the "all samples must be even length" rule. The assembler we were using at the put a garbage byte at the end of all the the odd-length sound samples, hence the clicks (I fi
Re: (Score:2)
4 buffer pointers, 4 lengths.
I had assumed since the DMA channels were all hard-panned (meaning it was obviously intended to be 2x stereo), and that it had enough horsepower to play LPCM interleaved, that it was pulling them apart at the hardware level.
Re: (Score:2)
Upon researching why that is, I haven't found any good answers.
There were bizarre limitations due to the way the chip was interfaced (byte-at-a-time sample upload through the GLU), and sample size limitations (32K). A general consensus that it was a pain in the ass to program. Sounds like it was great for sampled MIDI synthesis, though
Re: (Score:2)
So you just lied your mother fucking ass off, claiming the ST came out... in 1993?
...and then he went on to compare it to a super-expensive Amiga that didn't come out until a few years later and that hardly anybody had.
Re: (Score:2)
Even my Apple IIgs would boot from a hard drive. Drop in the SCSI card, connect the drive (with OS on it and ready to go) and boot...no issues.
Re:Just let it go (Score:5, Informative)
Just as there were multiple Apple computers, there were multiple Amigas. A couple came both as desktop or tower, with different motherboards with different numbers of expansion slots. Some of had an ATA bus onboard. One desktop model (The 2000) was also sold as the 2500, which meant it came with an accelerator. It was also branded as the 2000HD (or 2500HD) if it came with a hard disk, early in its lifespan, when that was an unusual feature. The machine also had both a native ("Zorro") bus and an ISA bus backplane; By default the ISA bus wasn't used for anything but you could get a "bridgecard" for it that made it useful. A variety of different HDD controllers were used with this system, and a couple of them including one I had (A2091) had interfaces that would not boot, specifically the ST-506/MFM interface. I mentioned the PC bridgecard and ISA bus above because that's ostensibly what that functionality was for; most PCs of the day had MFM disks. It was not expected that you would boot from a MFM disk. Instead, you would boot your DOS installation in the bridgecard, which was itself essentially a complete PC with eventually up to a 386sx [resource.cx]. But the A2091 also had a WD SCSI interface and a config rom and it would boot from SCSI just fine.
The Amiga 500 and 1000 don't have SCSI because they're trying to be cheap, the 2000 was the first no-compromises design and the 3000 was the last, sadly. Arguably the 3000 is the peak of Amiga. It's got your onboard SCSI. Even the little bitty desktop model has a bridgecard slot and a couple of ISA slots, and there was a bridgecard with a 486. It's got a built in scan doubler (flicker fixer) with a VGA output, so you don't need a weird monitor, though it still pairs best with an NEC Multisync or similar. It's the only Amiga model I would dick with today. Unfortunately I long ago traded mine for a BeBox, then later traded that for an Indigo, which I also don't have any more.
Re: (Score:3)
Concerning your point about using a hard disk. Perhaps that was an issue ea
Re:Just let it go (Score:4, Informative)
Where it really failed was in support for hard disks, etc. An Atari ST worked by just connecting a hard disk to the SCSI connector on the back and it booted instantly. Amiga hard disks required you to boot from floppy and jump through all sorts of hoops to try to get it to boot quickly.
Untrue except for a few exceptions.
Most Amiga drive controllers (both IDE and SCSI) could autoboot, and later models had them built in (A3000 had SCSI, A600/A1200/A4000 had IDE). Some of these models also shipped with hard drives preinstalled at the factory.
It would have been a good fit for business users from a technology standpoint with a multitasking gui, but was never really promoted to that segment.
Re: (Score:2)
I posted above about my A2000 rig that I've been playing with and I have to disagree that the Amiga would have been a good fit for business.
My modern experience has shown that it would have been too unreliable for business use. So many issues: guru meditations that take the whole system down, no protected memory, and no process management. Sure, you can do multitasking but what good is it if by running more processes you're just increasing the risk that something corrupts the whole system? I'd argue that ve
Re: (Score:2)
The competing operating systems of the day (Dos, MacOS, Win3.x etc) also had no protected memory or process management, errant programs commonly did take the entire system down. Microsoft are still trying to shake off the reputation for instability they earned in those days. If you wanted a stable system, you had to go highend - a unix workstation from the likes of sun or sgi etc and only a small percentage of specialised businesses tended to buy these as they were a lot more expensive.
The default resolutio
Re: (Score:2)
Also if you are using an A2000 now it was likely built in the late 80s, the hardware could have deteriorated with age - making it more unstable than it would have been when new. The capacitors in particular tend to fail, which will manifest itself with instability at first and eventually failure to boot at all.
Try running the same software in UAE and see if you encounter less crashes.
Re: (Score:2)
It was that great. It got done for a pittance what the other guys couldn't accomplish at all for hundreds to thousands more dollars. But now it's a historical curiosity best experienced through emulation, if at all. Any rinky dink $100 tablet is going to deliver orders of magnitude more of everything except hackability, which most people will never utilize anyway.
It was an amazingly performant system with an extremely well-architected operating system and an astounding level of aftermarket support, and anyo
Re: (Score:2)
No kidding.
But this is great news though. Some Amigas can be set up as email and ftp servers.
Re: (Score:2)
You don't need Linux PowerPC to setup a FTP server.
You can do this as well with AmigaOS4 and ZitaFTP Server (also available for Windows):
https://www.youtube.com/watch?... [youtube.com]
https://keasigmadelta.com/stor... [keasigmadelta.com]
Re: Just let it go (Score:2)
Woosh!!!
Re: Just let it go (Score:2)
Get over it. The Amiga wasn't THAT great.
Even if it was, that was so long ago that it isn't relevant in the slightest.
However, I don't think anyone needs to worry about people still using them as a hobby. They surely have another modern computer of some sort as well.
Re: Just let it go (Score:2)
It must've been good if it has such a continuing userbase and NEW accessories still being made and sold for it.
Can't say the same about an old Windows 95 machine even with the current retrogaming going on for that platform.
Re: (Score:3)
Most AmigaOS4 / AmigaOne owners dual boot into Linux PPC as well. Most seem to be programmers and technically educated people. Some even triple boot into MorphOS:
Public Release of MorphOS 3.17
"X5000 hardware support was improved to handle two Radeon cards and the HDD activity LED. Several text layout and rendering issues have been fixed in MUI."
https://morph.zone/modules/new... [morph.zone]
For mainstream usage, here's an opinion from a hardcore AmigaOS4 user:
https://www.youtube.com/watch?... [youtube.com]
Most are using AmigaOS4 as a
Re: (Score:2)
Re: (Score:2)
Not fiction, but here's an interesting read:
Commodore The Inside Story: The Untold Tale of a Computer Giant:
https://www.amazon.com/Commodo... [amazon.com]
Re: (Score:2)
I think the Linux skiddies will vapor-lock while trying to digest the phrase "...freedom fighters the Windows people." After all, their world view depends on "M$ bad" as an immutable constant.
As a former Amiga dev (I got to use one of the early A1000 prototypes with the keyboard that was housed in a black-painted balsawood case) I fell in love with it right away as it was miles from MacOS and X Windows and a huge step up from PC world of Windows on DOS, OS/2, GEM etc. I don't think the 68k evolutionary dead
Re: Alternate Universe (Score:2)
Personally, I've been waiting for the fanfic about Craftsman tools being the only tool brand people use.
Re: Alternate Universe (Score:2)
Hahahabhahaah