OlafS25 Quote:

I think there are two different markets. 68k on amiga today is retro. It is unrealistic to modernize amiga in a way that it becomes usable in todays world because today you have very different requirements, f.e. full memory protection for security reasons. You cannot add that without breaking everything. If you do there is not much software left. The advantages amiga today has would be no longer exist. I do not see much sense in it.

The biggest limitation for a modern 68k Amiga is lack of value which is lack of performance and too high of cost. Sure the AmigaOS lacks modern desktop features on high performance large footprint computers but that technology only scales down so far. The 68k Amiga can scale down further into small footprint low cost computing while providing a surprisingly modern environment. It would have retro appeal but also modern usefulness if it was competitive enough in value.

OlafS25 Quote:

On the other hand there is certainly a market for a modern and usable OS in todays terms as alternative. I was not very convinced about the different projects and ideas up to now because they all end in a niche with limited resources and thus missing drivers and no modern software. When I read from Deadwood developing a platform to port amiga (aros) software to linux I thought nice but I am personally already using windows and for retro there is 68k already. But then he showed the aros shell running in linux and also wanderer and from discussions it became obvious that the goal will be that you can mix aros to linux compiled components with linux like you can do that on aros 68k with aros and amiga. That would be something cool and it will be possible to create a distribution hopefully that is based on linux but also offers many components and look & feel of amiga. Much more than it is possible right now with themed linux desktops. Deadwood f.e. created a version of mplayer that uses a zune gui but the base is from linux. The ideal case will be when there are no real differences so you can mix all together. I will do a distribution when this is possible because it is both interesting and logical. It solves all niche related problems, you get a modern base, drivers and modern software.

AxRuntime from Deadwood is clever. Running AmigaOS and Linux software simultaneously is appealing and would be more so if UAE was integrated to auto run Amiga software. Finally we see an AmigaOS solution that supports SMP. Being able to use Linux development tools on high performance hardware while testing AmigaOS software is valuable. AxRuntime is not a simple solution for AmigaOS compatibility and requires much more than a simple recompile. The ISA, ABI, API, endianess, pointer width and memory consistency models (for SMP) have all changed compared to the 68k Amiga which is daunting to deal with. The two AmigaOS like apps which have been converted to run in AxRuntime were likely already converted to run with AROS x86-64 which may have been no small task itself and they were likely easy candidates to convert. Finishing the AmigaOS like AxRuntime API will likely require considerable work. If the AxRuntime API ever has most AmigaOS functionality and enough open source software is modified to use AxRuntime, AROS x86-64 will be enhanced and should attract a few more users and developers. It should provide better integration than using hardware virtualization to run separate AROS and Linux environments. I don't see it as being a game changer for the Amiga though.

OlafS25 Quote:

So in short... for me two directions for future... 68k as a fun platform, that includes of course V4 and PiStorm, Aros 68k and AmigaOS and so on NG based on Linux as modern base that offers lots of look & feel of amiga but is of course a different base. For me dropping compatibility (and that is what modernizing means) only makes sense if you win something with it. If you win a huge software base like linux modernizing makes sense because people use a OS to run software and not to move windows. If you drop compatibility just for "being modern" winning nothing from a user point of view but loose lots of software it makes no sense to me. 68k should stay where it is, it is also a nice niche in itself with chances, to reach out to a bigger number of users outside the current community you need different concepts. In this sense Deadwood convinced me and I see a chance there like I did with Aros 68k years ago. I think there are 3 different groups... current community with mostly retro interest and some who also would use a system with amiga look & feel for everyday, people who used amiga many years ago and still have memories and people not knowing amiga at all. All have different interests and priorities. I am looking forward to do things (distributions) for that groups, priority at first of course current community.

Modern 68k Amiga hardware with good value and professional enhancements would allow the current 68k AmigaOS and Amiga software to run on it and allow for an incompatible but enhanced AmigaOS NG to be run on it. This is what x86 hardware provided where old DOS can be used or a more modern and enhanced OS with less compatibility. Amiga 68k hardware based on FPGA simulation and emulation doesn't offer enough performance for Amiga NG use though. Amiga hardware which is 1/10 the performance and has 10 times the cost of a Raspberry Pi has 1/100 the value (performance/cost) and is nowhere close to competitive.

@matthey

Quote:

matthey wrote: A native port likely would be 15+ fps on higher end 1990s era 68k hardware. Unfortunately, this hardware is rare, expensive and offers poor value so there isn't much motivation as a target.

All Amiga hardware is getting 'rare' and 'expensive' these days. But thousands of Vampire cards have been sold and they don't come up for sale second-hand much, which suggests their owners are still using them and some might be interested in a Tomb Raider port. Based on the popularity of games that need AGA or RTG and a fast 68060 I reckon there must be hundreds of suitably equipped Amigas in use today. These numbers are more likely to grow than diminish, despite the 'rarity' and 'expense' of Amiga stuff today (still way cheaper than what it was 25 years ago).

Quote:

Multicore certainly isn't necessary. A 1GHz plus 68k Amiga ASIC SoC would allow Tomb Raider to have full playability even with emulation and without 3D hardware. That is only about a decade newer technology than the Vampire hardware and could have been produced for a few dollars per chip in quantity. 3D hardware allows the resolution and number of colors to increase of even old games which have been converted to 3D.

ASIC isn't necessary either. A fast 060 or Vampire V2 has the power to run Tomb Raider at a similar speed as the machine it was designed for (Pentium 60, 320x200 in 256 colors, no 3D hardware) and that is all I want. But it has to be native code, not emulated.

Quote:

It's amazing how much better Doom and Quake look in higher resolution especially with high resolution monitors today. GLQuake at 512x384x16 on my 68060@75MHz Amiga looked and played good enough that my nephew had no complaints compared to his older generation XBox. He was more absorbed by the game play...

Gameplay is more important than resolution, particularly with older games like Tomb Raider that were created for low resolution displays. These days I play user created Tomb Raider games on my PC in 1280x1024 24 bit, but I turn off smoothing because otherwise the textures look ugly.

I run my Amigas through an LCD TV. Tomb Raider on the Vampired A600 looks better in HDMI than it did on the PlayStation in composite, and I would be perfectly satisfied with it if the frame rate was reasonable. I have no need for higher resolution on the Amiga. When I do need it I will just use the PC (which itself is becoming 'vintage' not due to lack of processing power, but because XP is too old to run the latest browsers etc.).

Quote:

The grittyness probably has more to do with the dithering used and less to do with the pixelated jaggies. Higher resolutions practically anti-alias away pixelated jaggies.

But I want to see the pixels, not have them anti-aliased away into a blurry mess.


Quote:

There was talk of a "bounty" on the apollo-core forum for an Open Lara port. http://apollo-core.com/knowledge.php?b=3¬e=37536

Thank you. I read that thread and was somewhat disappointed (but not surprised) that the discussion went the way it always does on the Amiga. Nobody is ever satisfied with what can reasonably be done, they have to keep blowing up the specs until it's impossible. Only the Amiga suffers from this curse - nobody does it on other retro platforms...

Quote:

I'm not interested and my old 68k hardware is starting to fail. Without affordable higher performance 68k hardware, I'll likely lose interest and leave the Amiga.

That's fine, just don't throw your old 'failing' 68k Amiga stuff away like I did - give it to someone who will appreciate it.

Quote:

I believe there is enough interest in the Amiga to mass produce retro 68k Amiga hardware like a Raspberry Pi but the Amiga pirates would rather go down clenching their pieces of the treasure map rather than working together to relocate the lost treasure.

I believe there isn't enough interest to mass produce an 'Amiga' that is something like the Raspberry Pi. Why would the masses want it? The Amiga is now vintage hardware, like a classic car, an old sailing ship or a steam locomotive. People are restoring and reproducing those things too, but they aren't trying to mass market cheap modernized versions. Everybody enjoys the experience of riding in those old machines too, but few would buy one. And that's OK. It's time we stopped treating it as a popularity contest where your choices aren't validated unless millions of others are doing the same as you.

bhabbott Quote:

All Amiga hardware is getting 'rare' and 'expensive' these days. But thousands of Vampire cards have been sold and they don't come up for sale second-hand much, which suggests their owners are still using them and some might be interested in a Tomb Raider port. Based on the popularity of games that need AGA or RTG and a fast 68060 I reckon there must be hundreds of suitably equipped Amigas in use today. These numbers are more likely to grow than diminish, despite the 'rarity' and 'expense' of Amiga stuff today (still way cheaper than what it was 25 years ago).

I expect there are thousands of Amiga 68060 and Vampire hardware in use today. A 68060@50Mhz with slower memory in an older accelerator may not be enough to run some of the new ported games. There are at least 3 new 68060 accelerators which clock up to ~100MHz and support higher performance memory but there are not enough rev6 68060s. Vampire hardware has similar if not higher performance with similar if not lower core clocks but has higher performance memory than 68060 hardware. Amiga with 68060s often have bottlenecks, lack AGA, RTG or 3D and have issues with the old hardware. Vampire hardware removes most bottlenecks but has to deal with old hardware issues except the stand alone and lacks 3D. Amiga users often spend at least 500 dollars and sometimes thousand of dollars for real hardware with roughly double the performance of a 68060@50MHz. The value is not there and it is a pain to maintain and upgrade old equipment but hundreds of Amiga users per year are buying upgraded new hardware. Many times more people want the ultimate Amiga hardware but not at these prices and with maintenance issues of 30+ year old hardware. The best way to increase the chances of games being ported to the Amiga is to increase the user base by decreasing the cost of hardware and increase the performance of hardware so more games can be ported and less work is required to port and optimize them.

bhabbott Quote:

ASIC isn't necessary either. A fast 060 or Vampire V2 has the power to run Tomb Raider at a similar speed as the machine it was designed for (Pentium 60, 320x200 in 256 colors, no 3D hardware) and that is all I want. But it has to be native code, not emulated.

That is likely true but a few thousand 1990s era 68k Amigas are off the radar for most of the world. If we had hundreds of thousands of 1-2GHz 68k Amigas we would likely have an Open Laura port available within weeks.

bhabbott Quote:

Gameplay is more important than resolution, particularly with older games like Tomb Raider that were created for low resolution displays. These days I play user created Tomb Raider games on my PC in 1280x1024 24 bit, but I turn off smoothing because otherwise the textures look ugly. I run my Amigas through an LCD TV. Tomb Raider on the Vampired A600 looks better in HDMI than it did on the PlayStation in composite, and I would be perfectly satisfied with it if the frame rate was reasonable. I have no need for higher resolution on the Amiga. When I do need it I will just use the PC (which itself is becoming 'vintage' not due to lack of processing power, but because XP is too old to run the latest browsers etc.).

I agree that gameplay is more important and what makes retro games special. Some of the pixel art may lose appeal at higher resolutions and sometimes a naive conversion to more colors can look worse. Most games with scalable graphics look so much better and more modern at higher resolutions though. If wanting to appeal to a larger retro audience, I encourage higher resolution enhancements although it should be possible to allow an original mode and there is the original game. We had a nice discussion about reimagined/enhanced games in a thread here awhile ago where I posted some comparisons of old vs new games.

https://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=44415&forum=9&start=20&viewmode=flat&order=0#847790

Similar gameplay and atmosphere can usually be maintained with more modern appeal. Sadly, the Amiga is not getting enough of these enhanced games, not even the Indie games created by ex-Amiga developers. The situation will not change with FPGA hardware, 68060 hardware or emulated hardware.

bhabbott Quote:

Thank you. I read that thread and was somewhat disappointed (but not surprised) that the discussion went the way it always does on the Amiga. Nobody is ever satisfied with what can reasonably be done, they have to keep blowing up the specs until it's impossible. Only the Amiga suffers from this curse - nobody does it on other retro platforms...

Some of the problem is the Vampire attitude. Supporters often want to make better quality games enhanced for Vampire hardware although this rarely happens. Most 68k Amiga game ports run on the 68060 and may have a few enhancements for the Vampire. This would be a good target for a Open Lara port too. Vampire games are mostly reliant on ports and it is better to port many games than enhance and optimize a few for the Apollo-core/Vampire. The Vampire user base is too small and FPGA
enhancement too minor to advance the top 68k spec so we are stuck with the least common denominator (practically 68060+RTG).

bhabbott Quote:

That's fine, just don't throw your old 'failing' 68k Amiga stuff away like I did - give it to someone who will appreciate it.

I threw away my CSMK3 and both my rev 6 68060s yesterday (joking). I could probably sell my Amiga stuff for $5000 and most of it is far from like new condition (1-4000T, 1-4000, 1-3000T, 2-3000s, 4+-2000s, 1-1000, 1-500 plus some desirable goodies). I'm not motivated to do anything with it currently.

bhabbott Quote:

I believe there isn't enough interest to mass produce an 'Amiga' that is something like the Raspberry Pi. Why would the masses want it? The Amiga is now vintage hardware, like a classic car, an old sailing ship or a steam locomotive. People are restoring and reproducing those things too, but they aren't trying to mass market cheap modernized versions. Everybody enjoys the experience of riding in those old machines too, but few would buy one. And that's OK. It's time we stopped treating it as a popularity contest where your choices aren't validated unless millions of others are doing the same as you.

Why did the Raspberry Pi sell millions of units? A cheaper and more powerful Amiga is more appealing (retro 68k games) and can have similar usefulness. The AmigaOS has fewer desktop features than Linux but it really doesn't matter when the Amiga can scale to a smaller footprint and has better responsiveness. People don't expect much at an impulse purchase price and the Amiga can deliver a surprising amount of semi-modern computing usefulness. Even the 68000 had a 32 bit ISA and 32 bit addressing compatibility which is better than ARM's 26 bit addressing limitation for RISC OS which is the much less popular Raspberry Pi retro heritage. The Natami "MX Bringup Thread" had 761,487 views at one point I recorded it and went higher after. I would be surprised if THEA500 Mini doesn't have at least 50k preorders and it is a cheap gimmick that lacks value compared to what could be done.

Car manufacturers have taken advantage of demand for vintage vehicles. Think Chevy Camaro (5.5 million sales), Dodge Challenger (673k sold), VW New Beetle (1.7 million sold), Mini Cooper (5,387,862 sold), Mazda Miata (modernized remake of lightweight British and Italian roadster sports cars selling over 1.1 million), etc. Some of the remakes have been more vintage than others but there is an appeal for modernized vintage vehicles. The success varies perhaps dependent on striking the right balance between vintage and modern. A spiritual successor must have enough of the look, feel, philosophy and soul of the original. There are still people who want the original vintage vehicles and enjoy tinkering with them but most don't want to adjust the lash on the lifters and steering rack, add octane booster to fuel, grease their wheel bearings, etc. Most want a more practical modern version of the vintage car. It's no different with computers except that it should be possible to make the Amiga cheap enough for more impulse buying and then it is just a matter of exceeding low expectations instead of disappointing customers.

Top 10 narcissistic traits.

1). You truly believe your personal taste and opinion is every body elses reality and repeat it ad nauseam even though other people have point blank told you, time, time and time again that they don't share the same perspective.


Stay tuned for more of the top 10 in coming days.
There's guaranteed to be long winded examples just around the corner.

Last edited by fishy_fis on 15-Mar-2022 at 01:24 AM.

@matthey

Quote:

Why did the Raspberry Pi sell millions of units?

* Because a wider ecosystem was also built up
* Because Linux
* Because Open Source
* Because cheap

Quote:

A cheaper and more powerful Amiga is more appealing (retro 68k games) and can have similar usefulness.

* It won't be cheaper, no matter how much you try
* it won't be more powerful
* it won't be more appealing - retro 68k gaming isn't that a big thing
* only way it can have similar usefulness is if it provides similar interfaces, not a chance.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@OlafS25

Quote:

I think there are two different market. 68k on Amiga today is retro ... On the other hand there is certainly a market for a modern and usable OS in today's terms as alternative. I was not very convinced about the different projects and ideas up to now because they all end in a niche with limited resources and thus missing drivers and no modern software ... So in short ... for me two directions for future: - 68k as a fun platform, that includes of course V4 and PiStorm, AROS 68k and Amiga OS and so on - NG based on Linux as modern base that offers lots of look & feel of Amiga but is of course a different base ... In this sense Deadwood convinced me and I see a chance there like I did with AROS 68k years ago.

Yep.

Last edited by agami on 15-Mar-2022 at 03:20 AM.

_________________
All the way, with 68k

@bhabbott,

I disagree, I have a CSMK3 in a 4000T, and although it's very nice and fast by 68K standards it's just scratching the surface. The vampire shows it is solid enough to do much more. a proof of concept lets say.

Now imagine what kind of games are possible with a scaled up 1 - 2 G 68K would be capable of.

We know that clock for clock a 68K will outpace an x86. The game that would be possible I can't even Imagine, Lara would be easy. Think Bigger.

Chris

@QuikSanz

Quote:

QuikSanz wrote: @bhabbott, I disagree, I have a CSMK3 in a 4000T, and although it's very nice and fast by 68K standards it's just scratching the surface. The vampire shows it is solid enough to do much more. a proof of concept lets say. Now imagine what kind of games are possible with a scaled up 1 - 2 G 68K would be capable of. We know that clock for clock a 68K will outpace an x86. The game that would be possible I can't even Imagine, Lara would be easy. Think Bigger. Chris

I had a CSMK2 and RTG in my A3000, and it was disappointingly slow playing Quake. Unfortunately I sold that system (for a bargain price) so I don't know whether it would be better today with improved software. I do know that in general it's performance was similar to a Pentium 90, so it should have been able to do better.

Anyway my point is that a number of games and apps which have recently been produced for or ported to 68k Amiga 'require' an 060, so this has become the 'standard' high-end spec. The Vampire exceeds this performance by a small margin (on normal 68k code) so it makes sense to stick to this 'standard' for greatest inclusiveness. Anything above that risks alienating even more users - confirming the fears of regular Amiga users that the Vampire would divide the community.

I can imagine a 'scaled up 68k' system outperforming an x86, but it seems a bit pointless when x86 itself is not actually x86 anymore. Modern programs are written in high level languages so the CPU instruction set is irrelevant. I very much doubt that you could significantly improve on the performance of a modern CPU and compiler by using 68k code. And besides, isn't a modern PC fast enough?

I do enjoy using my Vampired A600 for web browsing and playing those games that need the faster CPU and RTG, but most of the time I prefer my A1200 with 50MHz 030. Why? I have been running this system for 25 years and feel more comfortable with it. It's fast enough for most of the Amiga stuff I do, and I actually enjoy the challenge of making my code more efficient. With a hyper-fast system I wouldn't bother, and so would be missing out on half the fun!

As for games - I like to try out new games that are released for the Amiga. But the more we divide the community the harder it gets to produce games with wide ranging appeal. I will not be 'upgrading' my systems just to play some game that needs a 1 - 2 G 68K (and no doubt some other stuff like fast 3D card etc.), and I bet a lot of other Amiga users feel the same. If I really wanted to play games that needed such power I would just get a PC or PlayStation.

@QuikSanz

Quote:

QuikSanz wrote: ... The vampire shows it is solid enough to do much more. a proof of concept lets say. Now imagine what kind of games are possible with a scaled up 1 - 2 G 68K would be capable of. We know that clock for clock a 68K will outpace an x86. The game that would be possible I can't even Imagine, Lara would be easy. Think Bigger.

The OpenLara port is an important first step. You might be pleased to know that Bigger Thinking is underway.


@bhabbott

Quote:

bhabbott wrote: ... The Vampire exceeds this performance by a small margin (on normal 68k code) so it makes sense to stick to this 'standard' for greatest inclusiveness. Anything above that risks alienating even more users - confirming the fears of regular Amiga users that the Vampire would divide the community. ... As for games - I like to try out new games that are released for the Amiga. But the more we divide the community the harder it gets to produce games with wide ranging appeal. I will not be 'upgrading' my systems just to play some game that needs a 1 - 2 G 68K (and no doubt some other stuff like fast 3D card etc.), and I bet a lot of other Amiga users feel the same. If I really wanted to play games that needed such power I would just get a PC or PlayStation.

Well it's a good thing PC and console makers didn't constantly fear "dividing" the community as they delivered advancements. When the 3D accelerators started appearing, it's a good thing PC game developers didn't just stick with the old chunky 256 colour VGA standard, for fear of dividing the PC gaming community.

What you call dividing, I call rallying behind a new direction.
The Vampire V4 HW can't do anything on its own to divide users. It will be up to market economics. If there is more software produced for the V4, then there will be more people buying V4. No one will stop anyone from making their software run on 060+RTG or even lower. It'll be up to the the developers to ascertain if it's worth their time to develop and provide ongoing support for a wider range of hardware spec targets.

As for some future 1GHz - 2GHz 68k silicon, with more improved 3D graphics capabilities, this will only reasonably come when the market is primed for it. Which means, when a decent amount of people are willing to upgrade to it because the price is favourable, and the software library is enticing. In all reality it'll take at least 2 years before such a thing could be released, if its going to have any chances at success.

And just like with any new tech, there'll be the Innovators, the Early Adopters, the Early Majority, the Late Majority, and eventually the Laggards.

You might miss a cycle. That's your prerogative. When the Early Adopters are upgrading from their 1st gen 1GHz board to a 2nd gen 2GHz board, you might think that the used price of couple hundred bucks for a 1st gen board is worth it, in order to play games that will not be found on the PC or Playstation.

Last edited by agami on 17-Mar-2022 at 09:18 AM.

_________________
All the way, with 68k

QuikSanz Quote:

I disagree, I have a CSMK3 in a 4000T, and although it's very nice and fast by 68K standards it's just scratching the surface. The vampire shows it is solid enough to do much more. a proof of concept lets say.

The CSMK3 is amazing. The memory and drive performance make it feel nearly modern compared to most other 68k accelerators which lack in this area. It's unfortunate that they didn't allow to use higher clocked 68060s like the rev 6 as there are 3 different oscillators on board. I suspect the memory controller won't tolerate clocking up and is known to burn out which is why I didn't try to push past a conservative clock of 68060@75MHz. The memory bus and caches are clocked up with this which partially uncorks the 68060 as CISC reg-mem designs are more likely to benefit from faster caches and memory being memory eating monsters compared to RISC.

QuikSanz Quote:

Now imagine what kind of games are possible with a scaled up 1 - 2 G 68K would be capable of. We know that clock for clock a 68K will outpace an x86. The game that would be possible I can't even Imagine, Lara would be easy. Think Bigger.

A CPU design has more to do with performance than the ISA but 68k designs were generally superior in PPA (Power, Performance, Area) than x86 designs most obvious with the semi-modern 68060 having a significant PPA advantage over the Pentium. Modern OoO x86-64 designs have ridiculous amounts of man hours of development with a cumulative cost of billions of dollars as each generation builds on the last. If a 68k ASIC was created, it would likely use a much simpler mostly in-order core design resembling a modernized 68060. While most Amiga users would likely prefer a full performance design, a mostly in-order core design that achieves better performance than RISC in-order designs while using less power than low power x86-64 cores would place the 68k in a nice middle ground for CPUs where there isn't as much competition. Mostly in-order core designs also use a fraction of the area of an OoO design allowing to add more cores or reduce cost. The 68060 core used a proven balanced in-order design with surprising performance and a more modern version should provide enough performance at 1-2GHz to revitalize the Amiga market while hopefully drawing interest for non-Amiga use.

bhabbott Quote:

I had a CSMK2 and RTG in my A3000, and it was disappointingly slow playing Quake. Unfortunately I sold that system (for a bargain price) so I don't know whether it would be better today with improved software. I do know that in general it's performance was similar to a Pentium 90, so it should have been able to do better.

The CSMK2 had slower memory than some other accelerators. I expect you would be surprised at how much better performance a clocked up 68060 with better memory performance has. The software is not that much better today but there is more especially in the free category. Performance can vary quite a bit depending on software setup too.

A 68060@50MHz being equivalent to a Pentium@90MHz is likely a stretch but may be true for some integer benchmarks. The ByteMark benchmark with a quick GCC compile showed a 40% integer performance advantage at the same clock speed so a 68060@50MHz has a similar integer performance as a Pentium@70MHz.

https://amigaworld.net/modules/newbb/viewtopic.php?topic_id=44391&forum=25#847418

The results are highly compiler dependent but likely give a rough idea as the overall score comes from several mini-benchmarks. The GCC results for floating point were significantly behind a Pentium at the same clock speed and Quake uses floating point. A vbcc compiled ByteMark benchmark showed a roughly equivalent floating point benchmark result for a similarly clocked 68060 and Pentium. This shows how poor the floating point compiler support and code generation in GCC and SAS/C was for the 68060 but, unfortunately, vbcc does not generate as good of integer code as some versions of GCC. We need a larger user base to improve compiler support. Even the 68060 generally did not get good floating point support or an instruction scheduler which is usually critical to performance of an in-order core but shows how good the 68060 is without recompiling code. The in-order integer performance, without using an instruction scheduler, of my (CSMK3) 68060@75MHz was actually better per clock than the OoO Pentium Pro and MMX. That's not supposed to happen considering how much lower power and area the 68060 is. The 68060 should have been clocked up to trounce the Pentium before the longer pipeline OoO Pentiums came out but the 68060 would have still been competitive afterward. Motorola already decided PPC was the high performance architecture despite difficulty clocking up the shallow pipeline designs.

bhabbott Quote:

Anyway my point is that a number of games and apps which have recently been produced for or ported to 68k Amiga 'require' an 060, so this has become the 'standard' high-end spec. The Vampire exceeds this performance by a small margin (on normal 68k code) so it makes sense to stick to this 'standard' for greatest inclusiveness. Anything above that risks alienating even more users - confirming the fears of regular Amiga users that the Vampire would divide the community.

The high end 68k Amiga spec has moved up from a high clocked 68040 or low clocked 68060 to an overclocked 68060, Apollo core or emulation, IMO. I agree that the 68k hardware has improved by a "small margin" but I see the bigger danger of not keeping up with technology advances as losing out to emulation and the hardware disappearing rather than leaving existing, dying and overpriced old 68k hardware behind.

bhabbott Quote:

I can imagine a 'scaled up 68k' system outperforming an x86, but it seems a bit pointless when x86 itself is not actually x86 anymore. Modern programs are written in high level languages so the CPU instruction set is irrelevant. I very much doubt that you could significantly improve on the performance of a modern CPU and compiler by using 68k code. And besides, isn't a modern PC fast enough? I do enjoy using my Vampired A600 for web browsing and playing those games that need the faster CPU and RTG, but most of the time I prefer my A1200 with 50MHz 030. Why? I have been running this system for 25 years and feel more comfortable with it. It's fast enough for most of the Amiga stuff I do, and I actually enjoy the challenge of making my code more efficient. With a hyper-fast system I wouldn't bother, and so would be missing out on half the fun!

There would still be a need for the Amiga to do more with less using a 1-2GHz 68k SoC. Hand assembler optimizations remained important for x86 games for quite a few CPU generations past the Pentium. In my opinion, it is a major contributing factor to x86 beating PPC where hand optimized code for games was not be done as often. One of the PPC objectives was to improve compilers so hand optimization and hardware optimization were unnecessary but that was a failure. Compilers are better today but will rarely beat a good programmer using hand optimized code especially with SIMD code. Auto-vectorization is very complex to implement in a compiler and still slows down code in some cases today. At the very least, it is good to use a profiler and debugger where looking at readable assembler code is very helpful.

bhabbott Quote:

As for games - I like to try out new games that are released for the Amiga. But the more we divide the community the harder it gets to produce games with wide ranging appeal. I will not be 'upgrading' my systems just to play some game that needs a 1 - 2 G 68K (and no doubt some other stuff like fast 3D card etc.), and I bet a lot of other Amiga users feel the same. If I really wanted to play games that needed such power I would just get a PC or PlayStation.

Cheap enough Amiga hardware which is powerful enough will add many times more new users than users lost to division. A 1-2GHz 68k+chipset SoC ASIC could be mass produced for less than $10 per chip. Think of the Flea FPGA Ohm which was $45 without mass production using a $10 1-2GHz SoC instead of a $10 FPGA.



Even with increasing the memory to say 512MiB, it should have been possible to mass produce this hardware for $40-$50 with a 68k SoC ASIC, at least before COVID. We may want more expansion but it should be possible to make very affordable Amiga hardware if it can be mass produced. In 1985, the Amiga 1000 had an introductory price of $1295 which was reasonable only costing roughly 25% of that of a car which was typically about $5000 then. Today, a new car costs roughly $40,000 on average and $40 mass produced hardware would be roughly 0.1% of the price of a new car or about the same as a nice meal out.

Last edited by matthey on 17-Mar-2022 at 07:16 PM.

@matthey

Quote:

Even with increasing the memory to say 512MiB, it should have been possible to mass produce this hardware for $40-$50 with a 68k SoC ASIC, at least before COVID. We may want more expansion but it should be possible to make very affordable Amiga hardware if it can be mass produced. In 1985, the Amiga 1000 had an introductory price of $1295 which was reasonable only costing roughly 25% of that of a car which was typically about $5000 then. Today, a new car costs roughly $40,000 on average and $40 mass produced hardware would be roughly 0.1% of the price of a new car or about the same as a nice meal out.

You either flunked economics or decided to conveniently leave out one important factor, or both. There has to be a large demand for said product to reach economies of scale for the price to be as low as you're quoting, and that demand simply doesn't exist, nor will it ever exist. The 68K architecture for home consumption died over 20 years ago and for good reasons and the quantity and lengths of your monologues will not change that.

ferrels Quote:

You either flunked economics or decided to conveniently leave out one important factor, or both. There has to be a large demand for said product to reach economies of scale for the price to be as low as you're quoting, and that demand simply doesn't exist, nor will it ever exist. The 68K architecture for home consumption died over 20 years ago and for good reasons and the quantity and lengths of your monologues will not change that.

The non-mass produced Flea FPGA Ohm at $45 had 21%-22% of the cost go to payment and shipping. The actual hardware and assembly cost was closer to half of that.

https://www.indiegogo.com/projects/fleafpga-ohm-fpga-experimenter-board#/ Quote:

Our approximate campaign breakdown is as follows (Note - proportions subject to variation, exchange rates etc.): 50-55% allocated to component procurement and PCB assembly costs (estimated) 4% allocated to PCB production test costs (estimated) 5% allocated to indegogo fees 3% allocated to payment processing fees where applicable 5-6% allocated to an order fulfillment processor 8% allocated to shipping costs Balance to be allocated to me for project management, writing of related test and end-user documentation etc.

Even though the hardware above doesn't need mass production to be affordable, it is true that an ASIC SoC does need mass production but it can be used elsewhere. Something like this would be a low end product. It would be good to see how many ASIC SoC chips accelerator developers like iComp would want while giving an introductory price. An ASIC SoC should allow a 1-2GHz 68k for a fraction of the cost of a 68060 accelerator. It would be helpful to see how many THEA500s are preordered or have sold. I would be surprised if there wasn't already at least 50k units preordered and I expect total units sold will be in the hundreds of thousands. THEA500 doesn't even offer good value and lowering the cost has a profound affect on volumes according to Jeff Porter.

Jeff Porter Quote:

We got the cost of the A500 down to $200. So we could sell it for $400 to the retail channel and then mark it up to the retailer for $600 and everyone was really happy. In fact, someone asked me today, "Why do we call it the Amiga 500?" I said, "Because that is what it is supposed to sell for." It's supposed to sell for 500 bucks. If you can make a computer for 500 bucks vs a computer for 600 bucks, you'll sell twice as many at 500 as at 600. There is just some magic price points in the world. And every time you take another 100 bucks out you double the volume. It's pretty amazing there that if you can get the right price points you can move a whole bunch of these things. So we did that and in January of 1987 we had the first Amiga 500 prototypes in plastic cases. I think our tooling had just come in. And I still remember Winfried Hoffmann coming in from Germany and he looked at that and it was running Defender of the Crown, it was running Deluxe Paint, it was running all these famous Amiga titles and it looked like a Commodore 64 and his eyes just lit up. He said, "Jeff, I know exactly what to do with this and by the way I'm taking that one. It's going with me. I'm hand carrying back to Germany. There is no way this one is going back to Pennsylvania. Sorry." I said, "Please, and by the way I have another one here, do you want two?" "Yes Please, and I have a third one from the guy in the UK."

https://youtu.be/otuQq4iqFgg?t=1029

There is always risk. You can stick your money in a government guaranteed account earning a lower interest rate than inflation or you can take some risk to make money.

As far as your flunking economics accusation, maybe you should do a little research first. I was more than happy to share some investment tips back on Jan 9.

https://amigaworld.net/modules/newbb/viewtopic.php?topic_id=44253&forum=2&start=160&171#847508

Are you even capable of looking at what CVX and UAN have done since then?



This is almost as big of a shit storm as recent trolling and this was before the war started between two of the top fertilizer producers in the world and most fertilizer production needs natural gas. CVX has little exposure to Russia but has assets in the U.S. to ship LNG to Europe as well as being a major partner in the Energy Triangle in offshore Israel which is already providing electricity for Europe and a pipeline is possible. I have a guy in my 50 years and younger investment group talking about retiring. I quickly commented that this helps protect us from inflation not incompetent politicians. An investment report shows UAN is considered high risk and CVX medium risk while that government guaranteed interest account earning less interest than inflation is considered low risk. The annual dividend yield of UAN is about 12% and CVX around 4% if you bought back in January.

Let's not forget about you trolling SiFive as worthless.

ferrels Quote:

This CPU isn't going anywhere. Nobody, especially anyone who is "big", is going to throw money at this. This CPU is still-born and the reason that no benchmarks have been provided is because it doesn't even exist in real silicon. It only exists on paper. This is just another scam to rob investors of money. Until SciFive can provide some engineering samples of this CPU for independent verification, validation and testing, it's just someone's pipe dream.

Maybe you are sore from having your economic prowess exposed when I posted the following.

matthey Quote:

SiFive has had real silicon chips produced before. They have made deals with some pretty big companies. Intel may even like them enough to offer $2 billion for them if the rumor is true. If they aren't professional chip designers then they must be the best scammers I have ever heard of.

https://amigaworld.net/modules/newbb/viewtopic.php?post_id=845506&topic_id=44315&forum=17#845445

Instead of licking your wounds when you have been owned, you double down with the trolling. See you at my next post where I'll embarrass you some more.

Last edited by matthey on 18-Mar-2022 at 11:01 AM.
Last edited by matthey on 17-Mar-2022 at 11:52 PM.
Last edited by matthey on 17-Mar-2022 at 11:51 PM.
Last edited by matthey on 17-Mar-2022 at 11:48 PM.
Last edited by matthey on 17-Mar-2022 at 11:24 PM.
Last edited by matthey on 17-Mar-2022 at 10:47 PM.

@agami

[quote]
agami wrote:
"The OpenLara port is an important first step. You might be pleased to know that Bigger Thinking is underway."

Elite Dangerous?

@matthey

Quote:

matthey wrote: The CSMK2 had slower memory than some other accelerators. I expect you would be surprised at how much better performance a clocked up 68060 with better memory performance has.

I won't be surprised, because I will never own one.

Quote:

A 68060@50MHz being equivalent to a Pentium@90MHz is likely a stretch but may be true for some integer benchmarks. The ByteMark benchmark with a quick GCC compile showed a 40% integer performance advantage at the same clock speed so a 68060@50MHz has a similar integer performance as a Pentium@70MHz.

70MHz, 90MHz - if 28% faster makes the difference then it's too slow anyway. I clocked my 060 up to 66MHz and Quake was still way too slow. Why? We may never know. I sold that system 20 years ago and won't be getting another one. But then Quake is a boring game anyway.

Quote:

I see the bigger danger of not keeping up with technology advances as losing out to emulation and the hardware disappearing rather than leaving existing, dying and overpriced old 68k hardware behind.

The hardware won't disappear. People are hoarding it now because its value can only go up. Emulation is to real iron like a print of the Mona Lisa is to the real thing. Those of us who have the real thing want to get more use out of it. We are not interested in mass-marketed 'Amigas', only in stuff that works with what we have. The further we get from that, the more pointless it becomes.

Why would I be interested in a GHz speed 'Amiga' designed to run software that has nothing to do with the Amiga? If I want to create and run games or apps that need that power, I will just do it on a PC etc. Porting stuff from other platforms is all about showing that can be done with what we have, not 'cheating' by creating vastly more powerful hardware to run it on - which is pointless because we already have vastly more powerful hardware we can run it on.

Quote:

Compilers are better today but will rarely beat a good programmer using hand optimized code especially with SIMD code. Auto-vectorization is very complex to implement in a compiler and still slows down code in some cases today. At the very least, it is good to use a profiler and debugger where looking at readable assembler code is very helpful.

I'm sure you are right, but with processors becoming ever more powerful the hand-optimization you do today becomes unnecessary tomorrow. Now programmers are switching to Java or Python (the 2 most popular languages at present) because they can't handle 'low level' languages. The performance hit is enormous, but worth it for the convenience - and they get away with it because the systems are so powerful.

Quote:

Cheap enough Amiga hardware which is powerful enough will add many times more new users than users lost to division. A 1-2GHz 68k+chipset SoC ASIC could be mass produced for less than $10 per chip. Think of the Flea FPGA Ohm which was $45 without mass production using a $10 1-2GHz SoC instead of a $10 FPGA.

Well good luck with that project, but if you do manage to get it off the ground your users won't be Amiga users so it's uninteresting to me.

Quote:

Even with increasing the memory to say 512MiB, it should have been possible to mass produce this hardware for $40-$50 with a 68k SoC ASIC, at least before COVID. We may want more expansion but it should be possible to make very affordable Amiga hardware if it can be mass produced.

But we aren't before COVID, we're in it. Manufacturers are having enough trouble making existing stuff. Yet another reason to get into retro computing. I am buying up old PC and Mac junk, and harvesting chips from it so I can make more truly retro hardware!

Quote:

In 1985, the Amiga 1000 had an introductory price of $1295 which was reasonable only costing roughly 25% of that of a car which was typically about $5000 then. Today, a new car costs roughly $40,000 on average...

Computers were more expensive back then for sure, but US$1,295 in 1985 only equates to $3,415 today - not much more than a high end gaming PC. Still too much for most retro enthusiasts, but you can buy an old A500 for a few hundred dollars and have a huge amount of fun with it.

Quote:

...and $40 mass produced hardware would be roughly 0.1% of the price of a new car or about the same as a nice meal out.

Only one sticking point, you might need a few million to get it started - better hope the demand is there!

Last edited by bhabbott on 29-Mar-2022 at 12:58 PM.

@QuikSanz

Quote:

Elite Dangerous?

What? No.
The idea is to move away from ports and produce new original content.

_________________
All the way, with 68k

Just going back to the original wish list of this thread, I'm going to call out the "32 channel 16-bit Paula".

Absolutely not! There are soundcards for this and they are extremely good clinically clean audio reproduction. And utterly soulless as a result.

I would definitely like more channels of Paula's uniquely coloured, nonlinear 8-bit audio (and all the crazy varying period and modulation controls they have). I would also like for those channels to have independent left/right volume control (for free panning) and if I can carry on wishing, an audio filter per channel. For maximum pleasure, a filter with continuously controllable cutoff and resonance.

I'm just going to put this cushion on my lap because I'm cold, umkay? No other reason.

_________________
Doing stupid things for fun...