Click Here
home features news forums classifieds faqs links search
5630 members 
Amiga Q&A /  Free for All /  Emulation /  Gaming / (Latest Posts)
Login

Nickname

Password

Lost Password?

Don't have an account yet?
Register now!

Support Amigaworld.net
Your support is needed and is appreciated as Amigaworld.net is primarily dependent upon the support of its users.
Donate

Menu
Main sections
» Home
» Features
» News
» Forums
» Classifieds
» Links
» Downloads
Extras
» OS4 Zone
» IRC Network
» AmigaWorld Radio
» Newsfeed
» Top Members
» Amiga Dealers
Information
» About Us
» FAQs
» Advertise
» Polls
» Terms of Service
» Search

IRC Channel

Who's Online
 78 guest(s) on-line.
 0 member(s) on-line.



You are an anonymous user.
Register Now!
 Argo:  7 mins ago
 zipper:  36 mins ago
 amigasociety:  52 mins ago
 ferrels:  1 hr 5 mins ago
 Rob:  1 hr 10 mins ago
 Fl@sh:  1 hr 11 mins ago
 smf:  1 hr 15 mins ago
 A1200:  1 hr 21 mins ago
 Trixie:  2 hrs 6 mins ago
 Tiw1:  2 hrs 21 mins ago

/  Forum Index
   /  Amiga General Chat
      /  Poll of CPU architecture interest for AmigaOS
Register To Post

Goto page ( Previous Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 Next Page )
Poll : Which CPU architecture are you most interested in for AmigaOS in the future?
68k
ARM
POWER
PowerPC
RISC-V
x86_64
other
 
PosterThread
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 7-Feb-2019 9:11:55
#141 ]
Cult Member
Joined: 14-Mar-2007
Posts: 654
From: Kansas

Quote:

Srtest wrote:
With respect mate, you're kinda doing the same thing some others here are doing and that (for the sake of the argument) combining tech with biz and markets and creating a mess.

Your figures about sales are impressive in the sense of sustainability (and for that their biz model should be commended). On that same note, you HAVE to know that ARM got its big brake because of developments which didn't stem from ARM. On one hand you can say they made themselves ready like an NBA team is clearing enough space to sign a player (only the opposite - they were available to get picked). You really can't compare ARM prior to smart phones and smart, small and functional devices (DuneHD for instance) to ARM at its current state. 68k and PPC were out there and they got tested as feasible mainstream processors. A whole infrastructure was built there around those cores so in the cultural sense they were the "soc" of an entire habitat. Nowadays that can happen if you get a multinational corp like M$ to decide it no longer wants to be attached to Intel (the sad irony).


I am aware of the big technology businesses and the power they wield from economies of scale (Apple, ARM, Intel, Google, Microsoft, Amazon, Samsung, Qualcomm, etc.). They matter not if a product can be created cheap enough and sold. Look at the Raspberry Pi for example. They didn't have to break any big players hold on a market, find a way into stores or even advertise much. They stealthily broke into the embedded market in a big way though. Did you see how much embedded hardware now uses the Raspberry Pi form factor? It didn't need to be ARM SoCs in the hardware. A product that offers similar value could do just as well. While it is difficult to match the subsidized mass produced price, it should be possible to come close and may be possible to have an advantage with a smaller footprint, more games and an FPGA for versatility. It would have more personality than the rather boring Raspberry Pi and most of the other SBCs on the market.

The 68k was tested and succeeded as a mainstream architecture for personal computers and the embedded market which is now split between x86_64 and ARM. While RISC hype probably would have successfully forced the 68k from the PC market, Motorola killed it for embedded too. Motorola new embedded designs switched to PPC cores and we know how successful that was for ARM. The failure of PPC to clock high didn't affect the failure in the embedded market. PPC being a resource hog and the unfriendly ISA had more to do with it. ARM has now gone big and fat with AArch64. While the ISA is friendlier than PPC, the footprint is closer to PPC than to the 68k. Perhaps ARMs energy efficient reputation will sustain it or perhaps they are abandoning part of the embedded market the same as Motorola did by switching from the 68k to PPC.

 Status: Offline
Profile     Report this post  
Srtest 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 7-Feb-2019 13:54:30
#142 ]
Regular Member
Joined: 15-Nov-2016
Posts: 228
From: Israel, Haderah

@matthey

You missed the point, mate. I didn't say that big biz are the facilitators, just that you need that kind of backing to start your own venture into something which challenges the entire way something is being done conventionally (like what is power and how the Amiga way can manifest into a powerful presence not needing 16gb of memory like Win10). If you needed smartphones and streamers and smart tvs to come forth then your developments or maybe the ones you are responsible for weren't the main facilitators.

Take Google for example. You can hardly pass over a page anywhere where they talk about tech and IT without mentioning them. Where would Google be in smartphones without Linux, Java and those processors you talk about? you see what I did there? I switched it because that's the truth, that Google with all their bells and whistles and press got their foothold in that market due to developments not coming from their own house. BTW, I just got a message that they are shutting down G+ because they thought that just because their name is Google they would be able to compete with FB. The only difference is their financial worth and backing which allows them to make those kind of mistakes. M$ built their entire operation based on that (and in-field tyrrany).

ARM has proven to be great for being in a certain casing adhering to needs. Before that combo of the right casing + needs became apparent they weren't in the mainstream consumerist sense. There are two completely different manifestations of what you call "embedded": there's embedded cause someone decided it will serve some purpose (or maybe a decision to change with the times a.k.a smart devices). Then there's embedded because products widely and universally accepted appreared for which they became known as embedded in the consumer market.

What are consoles? PS3 had the cell processor and was different than the competitor. The result being something which was different than the competitor. Fast forward to PS4 vs XBOX and they are practically the same and no one who wants one of these, cares. I, an Amigan, had to together with my dad convince my nephew and my sister in law to go for a pc instead of a console (I don't even remember which one it was). You know what he's doing today for fun? alternating between a tablet and a phone while using his pc for some online activities and some very specific tasks.

For all those who say Amiga is some kind of a relic of the past - they actually figured out how to tap into all of that in a single machine/device. I always get a laugh thinking about how Amiga was referred to as a console by "serious folk" who today look to make their pc more like that (and for all intents and purposes the pc market is a derivative of that and consoles are introducing pc-like features because of online activities).

Power didn't fail. It was never made to succeed in the mainstream like the 68k and eventually Apple ditched it (and on the way out sabotaged something like the PWRficient). I still remember though how those Transmeta processors couldn't find any takers cause it wasn't the right time (which means the big corps wouldn't allow it) even though they were x86. ARM not being x86 did manage to take in new markets. X86 nowadays is a market - nothing else and not even a main one.

Last edited by Srtest on 07-Feb-2019 at 02:02 PM.
Last edited by Srtest on 07-Feb-2019 at 01:56 PM.
Last edited by Srtest on 07-Feb-2019 at 01:56 PM.

 Status: Offline
Profile     Report this post  
Signal 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 7-Feb-2019 14:42:13
#143 ]
Cult Member
Joined: 1-Jun-2013
Posts: 657
From: USA

When I purchased my A500 I did not care what CPU it had. Did not matter.
The adverts on TV for a new crop of computers, pads, phones, etc, I do not see or hear anything about what CPU the devices have in them, except for the "Intel inside" as if it makes the device better somehow or able to leap tall buildings in a single bound.

When people 'must' show me their new device they do not start by bragging about which CPU it has in it. No. They demonstrate what it can do, or how nice the screen is, and probably 95% have no idea what is inside, and it does not matter as long as it does the job.

Now if you are going to make CPU selection on an existing OS and what restrictions it puts on design factors (BE, LE, bus width) then your options are limited but not down to a single choice.

The same approach can be applied in the automotive industry for mass produced vehicles. You don't start design/build by selecting the engine first then building the rest around it, that's for motorcycles in most cases. You start with function. High speed, low speed, mileage, comfort, safety, and then select a suitable power-plant and make as little change to the design as possible to incorporate the choice.

Just think. If the CPU was mounted on a card that went into a slot it would not matter what CPU was selected as long as it provided the necessary function. Lower cost motherboard, lower function CPU. More functional motherboard, more powerful CPU. The brand or design of the CPU does not matter, only its provided function.

What is really important to be able to call your computer an Amiga is a 3com Ethernet device. Can't live without one.

_________________
Tinkering with computers.

 Status: Offline
Profile     Report this post  
clusteruk 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 7-Feb-2019 14:58:41
#144 ]
Super Member
Joined: 20-Nov-2008
Posts: 1518
From: Marston Moretaine, England

@matthey

Obviously Amiga should be going ARM for cost and power considerations, I have thought this sine 2007. Right now Apple is planning the transition to ARM and the PR behind an Amiga comeback beating Apple would be hilarious.

X86 is old and slowly losing to ARM.

PowerPC ?

Also moving to ARM is really easy as it is pretty much running already and just needs fine tuning and supporting.

Steve

_________________
Amiga 1000, 3000D Toaster, Checkmate A1500 Plus
http://www.checkmate1500plus.com/

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 8-Feb-2019 0:06:54
#145 ]
Cult Member
Joined: 14-Mar-2007
Posts: 654
From: Kansas

Quote:

Srtest wrote:
You missed the point, mate. I didn't say that big biz are the facilitators, just that you need that kind of backing to start your own venture into something which challenges the entire way something is being done conventionally (like what is power and how the Amiga way can manifest into a powerful presence not needing 16gb of memory like Win10). If you needed smartphones and streamers and smart tvs to come forth then your developments or maybe the ones you are responsible for weren't the main facilitators.


Some investment is required in most businesses but I don't consider a low performance ASIC to be capital intensive.

low performance: 500-1500 MHz, millions of $ to develop
mid performance: 1500-3000 MHz, tens of millions of $ to develop
high performance: 3000+ MHz, hundreds of millions of $ to develop (This is capital intensive!)

Trevor Dickinson suggested $10 million would be enough to relaunch the Amiga. It should be possible to make a low performance 68k ASIC for a fraction of that even with bringing in some professional help. A veteran CPU designer should give the project the respect it needs to sway some potential embedded partners who look at projects like the Apollo Core as impressive but done by amateurs (true even though the developers could possibly do most of the work under the supervision of a professional). A high production embedded partner and underutilized fabs should make per unit production substantially cheaper than off the shelf commodity CPUs. The small footprint of the AmigaOS and 68k could then be used to push into low priced markets. There would be no revolutionary technology changes but rather starting simple, small and relatively cheap like ARM. The idea is to get Amiga back into technology which requires hardware. I prefer vertical integration again with Moore's Law ending as CPU designs can be used for longer. There should be enough capital left to invest in software including AmigaOS improvements.

Quote:

Take Google for example. You can hardly pass over a page anywhere where they talk about tech and IT without mentioning them. Where would Google be in smartphones without Linux, Java and those processors you talk about? you see what I did there? I switched it because that's the truth, that Google with all their bells and whistles and press got their foothold in that market due to developments not coming from their own house. BTW, I just got a message that they are shutting down G+ because they thought that just because their name is Google they would be able to compete with FB. The only difference is their financial worth and backing which allows them to make those kind of mistakes. M$ built their entire operation based on that (and in-field tyrrany).


The big boys can afford to make big mistakes, for a while. I was not talking about going big but rather small. The big boys often aren't flexible or smart enough. They are often trying to push a square peg through a round hole. Sometimes they succeed but the result is often inefficient like Android, intermediate code, Java applets and byte code where a simpler OS and binaries for a standardized architecture could be faster, smaller and more energy efficient. Nobody is going to replace that anytime soon but selling enough cheap mass produced hardware might start to make a difference.

Quote:

ARM has proven to be great for being in a certain casing adhering to needs. Before that combo of the right casing + needs became apparent they weren't in the mainstream consumerist sense. There are two completely different manifestations of what you call "embedded": there's embedded cause someone decided it will serve some purpose (or maybe a decision to change with the times a.k.a smart devices). Then there's embedded because products widely and universally accepted appeared for which they became known as embedded in the consumer market.


ARM makes professional designs and they are good at customizing and marketing them. It is the easy choice. There are developers who are not happy with ARM products for various reasons or want something different. Look no further than RISC-V for example. The embedded market is large enough to support a much wider variety of architectures but they need to offer advantages.

Quote:

What are consoles? PS3 had the cell processor and was different than the competitor. The result being something which was different than the competitor. Fast forward to PS4 vs XBOX and they are practically the same and no one who wants one of these, cares. I, an Amigan, had to together with my dad convince my nephew and my sister in law to go for a pc instead of a console (I don't even remember which one it was). You know what he's doing today for fun? alternating between a tablet and a phone while using his pc for some online activities and some very specific tasks.

For all those who say Amiga is some kind of a relic of the past - they actually figured out how to tap into all of that in a single machine/device. I always get a laugh thinking about how Amiga was referred to as a console by "serious folk" who today look to make their pc more like that (and for all intents and purposes the pc market is a derivative of that and consoles are introducing pc-like features because of online activities).


The extra efficiency that is achieved out of the standardized console hardware is nice but it is disappointing that they could not turn into more of a PC with regular OS, keyboard and mouse (PS3 had it but Sony pulled the plug). The Amiga CD32 with expansions could be used as more of a general purpose multimedia PC than most consoles are capable of today. The Raspberry Pi has the standardization and flexibility but misses the game focus and is short on performance. I think there is an opportunity here for the right product at the right price.

Quote:

Power didn't fail. It was never made to succeed in the mainstream like the 68k and eventually Apple ditched it (and on the way out sabotaged something like the PWRficient). I still remember though how those Transmeta processors couldn't find any takers cause it wasn't the right time (which means the big corps wouldn't allow it) even though they were x86. ARM not being x86 did manage to take in new markets. X86 nowadays is a market - nothing else and not even a main one.


I think the PWRficient CPUs were some of the better PPC designs, especially from a new business, but the lack of performance disappointed some Amiga users as they were more energy efficient designs to try to compete with ARM. I have my doubts that they could have saved PPC but Apple taking them out certainly hastened the demise of PPC. I doubt that this was a deliberate attempt by Apple to end PPC though. They more than likely saw good engineering and took advantage of the cheap price due to lackluster PPC sales. They needed more talent to bring their chip designs in house as they became more vertically integrated. Apple did much more harm to PPC and likely deliberately with Exponential.

https://en.wikichip.org/wiki/exponential_technology/x704

Transmeta was a different story. Intel probably did use noncompetitive practices to discourage the adoption of Transmeta. Transmeta was over hyped and required a very complex software layer which was less than efficient though. They seemed to be competitive for a while but the kind of investment they received could make pigs fly. Intels attempt at VLIW for general purpose computing with Itanium also failed despite a huge budget.

Last edited by matthey on 08-Feb-2019 at 10:16 PM.
Last edited by matthey on 08-Feb-2019 at 10:15 PM.

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 8-Feb-2019 21:23:47
#146 ]
Cult Member
Joined: 14-Mar-2007
Posts: 654
From: Kansas

Quote:

clusteruk wrote:
Obviously Amiga should be going ARM for cost and power considerations, I have thought this sine 2007. Right now Apple is planning the transition to ARM and the PR behind an Amiga comeback beating Apple would be hilarious.

X86 is old and slowly losing to ARM.


What kind of "power considerations" do you have?

Energy efficiency is very important in servers yet ARM has failed numerous times to take any market share. We are talking about big players trying ARM including AMD, Broadcom, Calxeda, Applied Micro and now Qualcomm with their Centriq line. Let's look at where the new ARM based Qualcomm Centriq CPUs are energy efficient.

"Power – A 120W TDP is not impressive. Being able to drive down idle power consumption to 8W is. This is what Qualcomm has achieved with Centriq."

https://www.forbes.com/sites/moorinsights/2017/11/10/qualcomm-chasing-clouds-with-the-launch-of-centriq-2400/#1a0540f568eb

ARM cores have low power draw when they aren't doing anything! This is good for many embedded applications but not as important for a server or PC where code is waiting to execute. It is often more important to execute code quickly which needs single core performance.

"The reason why companies failed in the past came down to performance. Arm server chips just couldn’t match the performance of their x86 counterparts. Even in the low end of the market, Intel was able to stave off Arm with the position of Xeon D.

Things are different this time around. Qualcomm claimed pretty impressive performance numbers during its launch event. When running SPECint_2006, the Centriq 2460 saw about 7% better performance than the Intel Purley Platinum 8160. Granted, Qualcomm is comparing its 48 cores against 24 Skylake cores, but it is a competitive socket."

I mentioned earlier how Intel had trouble as of late with their die shrinks. This allowed Qualcomm to make a comparison of the Centriq 2460 at a 10nm die size with Purley Platinum 8160 at 14nm. Adjust the performance and power for the die shrink and the Centriq 2460 doesn't look very good in either category. Maybe Qualcomm could have gained some ground but AMD has x86_64 EPYC at 10nm and is the biggest beneficiary of Intel's problems.

ARM saves energy when some of the many cores are idled but it uses more energy to complete a task in the cores it is using. This is because a higher single core performance allows doing more in less time and idling for for a longer time.

Quote:

Higher performance allows doing more in less time, and sleeping for a longer time
o Requires less energy to complete a given task

Higher performance allows lower clock rates
o Reduces clock tree and CPU power when active
o Enables use of HVT cells and allows smaller implementation, both decreasing power leakage


http://www.cast-inc.com/company/tech/CAST-ProcessorPower-Webinar-Dec2012.pdf

A CPU with better single core performance may draw more power but often uses less energy overall to complete a task. See slide number 17 from the above link. A CPU which has a low power/MHz is likely to use more energy and take longer to complete tasks as demonstrated by ARM cores in the slide. As a human at a computer, do you want your task to finish quickly and use less energy or finish slowly and use more energy?

Some people have predicted ARM to takeover the PC market too but it hasn't happened because single core performance is important, especially for games.

Quote:

Also moving to ARM is really easy as it is pretty much running already and just needs fine tuning and supporting.


AROS ARM may be easy but existing Amiga software would be doing good to execute at 1/10 the performance of a native CPU. It will be interesting to see how it turns out but I doubt it will be as compatible or fast as UAE (using a single core) anytime soon.

 Status: Offline
Profile     Report this post  
hth313 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 8-Feb-2019 23:50:26
#147 ]
Regular Member
Joined: 29-May-2018
Posts: 123
From: Delta, Canada

Quote:

matthey wrote:
AROS ARM may be easy but existing Amiga software would be doing good to execute at 1/10 the performance of a native CPU. It will be interesting to see how it turns out but I doubt it will be as compatible or fast as UAE (using a single core) anytime soon.


It is possible that another emulator is faster, but it is really not that important. I do not want to use an emulator, I want to run native on real hardware. A 68k emulation on ARM is a stop-gap for some software while it is either rebuilt to native ARM or in case source is not available, is replaced with something better and more modern.

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 9-Feb-2019 3:40:23
#148 ]
Cult Member
Joined: 14-Mar-2007
Posts: 654
From: Kansas

Quote:

hth313 wrote:
It is possible that another emulator is faster, but it is really not that important. I do not want to use an emulator, I want to run native on real hardware. A 68k emulation on ARM is a stop-gap for some software while it is either rebuilt to native ARM or in case source is not available, is replaced with something better and more modern.


Even the Raspberry Pi 3 has a low performance CPU designed for energy efficiency. Single core performance is poor. It is unlikely any emulator is going to have good performance unless it can use multiple cores in parallel but then there may be cache and memory consistency, coherency and contention issues. One core converting the 68k code to AArch64 code while another executes it may not be as efficient as hoped. It is possible to have a 68k emulator on each core but the maximum performance of a task/process does not improve.

I'm afraid much of the 68k software library does *not* have sources available. AmigaOS 4, MorphOS and AROS x86_64 assumed 68k emulation would be good enough but more compatible UAE, FPGA Amiga hardware and original Amiga hardware has more Amiga users. Starting on a new architecture with little or slow software is difficult enough without having a noncompetitive AmigaOS which is unlikely to attract new users and developers. The value of Amiga compatibility has often been underestimated.

 Status: Offline
Profile     Report this post  
gregthecanuck 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 9-Feb-2019 4:26:58
#149 ]
Cult Member
Joined: 30-Dec-2003
Posts: 840
From: Vancouver, Canada

@matthey

I think we need to be realistic going forward, at least for the near future (1-2 years).

Any "new" users are more than likely going to be "returning" users who had an Amiga but gave up on it due to many issues. What will they return to? An expensive PPC system? Or an entry-level 68K system to kick the tires first? Why are they returning? I think mostly retro/nostagia at this point.

A-EON has publicized the fact they have managed to get some brand-new Amigans onto PPC hardware. While this is very cool I don't think this will be a large factor in new users going forward. Maybe the Tabor board will help as it has lower costs. I also don't think the PPC hardware is flying off the shelves either, but to a large/smaller degree this is due to trouble getting OS4 updates out the door in a timely manner... (mostly out of A-EON's hands, unfortunately).

So that puts things back to 68K-land. What will new users come for? Reviving Amigas in the attic with all the issues and limitations? Some will. Taking an Amiga and accelerating it? Or jumping into one of the FPGA products? (Vampire, Mist, ...).

This is where I think the Vampire makes most sense... it is carrying the 68K and chipset forward and at a reasonable price. The demand continues to outstrip supply.

So I just don't see where any large discussion on CPU architecture make sense, at least for the 1-2 year time span. Maybe one day OS3.1 or OS4 is open-sourced and someone could work on making it portable to ARM in big-endian mode? Who knows.

But for the near future I think 68K is a great way to start rebuilding a (for now, retro/nostalgia) user base. With that growth will hopefully come some momentum that can be built upon.


 Status: Offline
Profile     Report this post  
Fl@sh 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 9-Feb-2019 12:05:55
#150 ]
Regular Member
Joined: 6-Oct-2004
Posts: 104
From: Napoli - Italy

@matthey

Quote:
I think hardware has to be part of the plan. Proposal Create a 4 to 8 core 68k SoC ASIC 500MHz-1500MHz. Partner with a high production embedded partner to reduce per unit cost to a fraction of that of an FPGA used in the Vampire. Produce with underutilized fabs using older die sizes to reduce cost per unit. Older die size and extra cores in case of defects increases yields. 68k SoC ASIC 500MHz-1500MHz (CPU, Amiga custom chips, RTG, I/O, 3D?) 1 core used for OSD and management 1 or more cores used for AmigaOS (more than one would require a new AmigaOS with SMP) 1 or more cores used for embedded or retro use with a small FPGA Small FPGA can be reset independently of the CPU cores and new logic loaded. The logic could be a custom chip set for retro use (Atari ST, Sega Genesis, NeoGeo, x68000, etc.), custom FPGA logic for embedded use with IO pins connected externally or a custom codec for an OS. Memory: 512MiB-1GiB of memory I/O: HDMI/DVI, USB, ethernet, GPIO, memory card drives, maybe SATA Markets: retro, toy, hobbyist, embedded, education Cost: $50-$100 U.S. The hardware would be somewhat like a Raspberry Pi but would have the huge library of 68k software and especially games from the start.


Usually I agree with you in many many things and also this time you could be right, but.. a good plan is always a simple plan and your proposal is a bit difficult to realize. I think your best result could lead to a renewed retrocomputing interest, just like vampire project, nothing more.
A home made 68k asic will never be faster than any arm core present in today low price consumer devices.

So I say: why reinvent the wheel?

We already have a good start: the RPi3 developer machine (and next comings), is module expandilbe, is opensource and is linux compatible; if the plan will fail you'll have a linux developer machine to fun with or you'll can simply resell it in few days...

We already have right hardware.. the only thing we need is an AmigaOS4 port on it.
If you really need 68k compatibility you'll have Petunia or UAE, really you'll don't need more.

A lot of recent projects can be easily recompiled to gain full speed and a lot of new modern software can be developed using an arm cpu and a lot of ram and expansion ports.
Yes, you can get all these for few money!

This plan will lead to a fast grow in userbase, injecting renewed interest in amiga platform, also with some free marketing help, like using widespead socials, thus recalling to home many many old amiga users and, at same time I'm sure, a lot of new guys.

With a price under 100$ you can sell many thousand of these new Amiga in few months and make money on it.

You can also sell new project to all industries with time critical processes and/or all developers involved in robotics and automation processes, due some peculiatities of RPi/AmigaOS like small size (mirokernel), modularity, simplicity, next to metal programming with C/ASM code, good documentation, etc..

So I really can't explain why we have to complicate things.
Everything can be really simple, we already have an hw solution, we need only a simple porting of AmigaOS on it, and use customizations permitted by hw modularity to repropose old amiga fashion/style use.

Last edited by Fl@sh on 10-Feb-2019 at 09:00 AM.
Last edited by Fl@sh on 10-Feb-2019 at 08:57 AM.
Last edited by Fl@sh on 10-Feb-2019 at 08:50 AM.

 Status: Offline
Profile     Report this post  
hth313 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 9-Feb-2019 18:46:20
#151 ]
Regular Member
Joined: 29-May-2018
Posts: 123
From: Delta, Canada

@matthey

The Amiga situation is more of a software problem than a hardware problem.

While a 68k ASIC would be nice, I think any such approach need to be done with a focus on the embedded field. This is where it may be needed. There have certainly been a huge range of architectures used in this field over the years, so one more is not anything strange. This market is very different from the desktop market in this regard, as it is very diverse.

Any potential Amiga coming out from it would be a lucky spin-off.

Going to the software side, we have RPi/ARM and it is here now. You can order one today for little money and have it delivered home in a week or so. It may not be perfect in every (performance) aspect, and it will not fit everyones desire, but it certainly is common, it can do big endian and there are lots of appealing things with it. Future improvements will be done to it without us having to get involved in it.

I could go on and on why the CPU (and especially variant used) in the device is highly irrelevant to virtually everyone. I do not have to go any further than myself for that. But meh, no one cares.

 Status: Offline
Profile     Report this post  
ne_one 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 9-Feb-2019 20:16:04
#152 ]
Cult Member
Joined: 13-Jun-2005
Posts: 854
From: Unknown

@hth313

Quote:
I could go on and on why the CPU (and especially variant used) in the device is highly irrelevant to virtually everyone. I do not have to go any further than myself for that. But meh, no one cares.


I wouldn't be quite so dismissive. There is a large contingent that shares the same sentiment.

Preserving the Amiga is no longer a concern but to move forward, the OS needs to be reinvented.

I can already hear gasps and screams about the lack of resources and money.

Pfft! If anyone still thinks that going back to 1985 is the way to go, it's time to teleport into the present.

It's eminently doable.

 Status: Offline
Profile     Report this post  
davidf215 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 10-Feb-2019 9:04:05
#153 ]
Member
Joined: 14-Feb-2010
Posts: 76
From: Texas

@matthey

Sorry for responding to a post 3+ pages back, but I've been focused working on merging my code with ReAction code generated with Emperor (on OS 4.1).

Quote:
matthey said:
The largest CPU market in the world is embedded where the 32 bit CPU is the most popular by far. Not that polls are the most accurate, but the AspenCore gives the following poll data.

In what way is the AmigaOS part of the Embedded market? The biggest part of the embedded market (according to the chart you referenced) is the Industrial industry. I don't know of any Amiga solution that is available for that market.

Also why would you see the Amiga as part of the embedded market when the Amiga has been a desktop solution? It could be an embedded solution, but that is difficult as there aren't many (any) embedded solutions for it that I've seen. Maybe this is an area of opportunity.

Quote:
matthey said:
32 bit CPUs in embedded are healthy and dominating 64 bit CPUs by nearly 5 to 1 as 64 bit CPUs are often overkill and wasteful. The 64 bit embedded market nearly doubled from 2012 to 2017 though. The 8 bit CPU is holding strong because sometimes it is adequate and often most energy efficient and cheapest. The Adafruits Trinket board with Microchip ATtiny85 (8 bit AVR) MCU is only 1.2" x 0.6" x 0.2" (31mm x 15.5mm x 5mm). The following is a pic.

I guess until Amiga software really pushes the envelope of the available CPU power, then 32 bit would probably be okay. I'd go along with a multicore 68k chip if one was available, but if a multicore 68k chip isn't available, skip to the next 32bit multicore option (preferably one to which AmigaOS could be quickly ported).

That's a neat 8bit chip. It's cool. Not sure how a next gen Amiga could make use of it.

Quote:
matthey said:
How many 64 bit CPUs have that small of foot print?

I agree. But the footprint will shrink in time, and, admittedly, probably never be as small as a 32. I think many people don't really care about power consumption, though. The PowerMac G5 was a good example of that.

Quote:
matthey said:
We can already see the sales potential of the x5000 at its current price and it is unimpressive. It's nice enough hardware but too expensive. Tabor will likely sell a little better but it is cheapened hardware that is not cheap enough.

Unimpressive, sure. But I don't think that's entirely caused by price issues. I suspect; however, that it's (specifically with the X5000) price and software. For example, more would buy if an office suite, like LibreOffice, was available. This way the computer has more value as it could be used for retro gaming and some "real-life" productivity work.

"Too expensive" is relative. I bought my A1200 for about the same price that the Tabor will be. And "not cheap enough" is also relative. Not everyone wants to experience gaming (or whatever) on slow systems, and a Pi Amiga would be slower than Tabor. I do think a standalone Amiga device around the $100 price (or that of a RPi) would be a good addition to the Amiga market.

Quote:
matthey said:
It is possible for 68k hardware to be in the same ballpark for price and performance as Raspberry Pi 3 hardware. Raspberry Pi owners feel restricted by 1GiB of memory but that is a huge amount of memory for a 68k Amiga.

Yes, but I wouldn't consider any 68k Amiga a NG Amiga unless it was 4 or 8 core in the GHz range.

Quote:
matthey said:
It is more important to have the source available for embedded as the results of polling from the link above show.

An NDA for AmigaOS 4 source code between two companies could work for this. Otherwise, use Aros. Aros source is already available.

Quote:
matthey wrote:
Also note that the Raspberry Pi form factor is 23% of the embedded market in the 2017 poll. See why I switch between embedded and entry level hardware?

Sure, port AmigaOS 4 to ARM. But the ARM processor in the Pi should only be a stepping stone to Amiga systems with more powerful processing for more advanced (faster) solutions. The A1222 and X5000 could sell more as the low, entry level AmigaOS users want a more powerful desktop solution rather than a low entry level system. Once the A1222 and X5000 PPC product line is completed/sold out, newer models could use Arm instead (like the one that @tlosm mentioned).

Quote:
matthey wrote:
Users bought Raspberry Pi 1 for $20-$25 U.S. with 256-512MiB of memory and often complained about the lack of memory. The Raspberry Pi price was what attracted many users yet it was upgraded to 1MiB of memory and the price increased to $35 which discouraged some price conscious buyers. Many users don't think there was any choice but 68k Pi hardware with AmigaOS and 512MiB of memory may have allowed the cheaper price without the memory shortage problems.

Yes, but where can you buy a newly manufactured Amiga running on 68k for $25 or $35? I'd probably (90% likely) buy one it it was available and if it did what I wanted it for.

Quote:
matthey wrote:
Proposal

Create a 4 to 8 core 68k SoC ASIC 500MHz-1500MHz. Partner with a high production embedded partner to reduce per unit cost to a fraction of that of an FPGA used in the Vampire. Produce with underutilized fabs using older die sizes to reduce cost per unit. Older die size and extra cores in case of defects increases yields.
(system specs removed)

Now I understand the goal of what you're trying to create. If this was standalone and came with AmigaOS, then I'd add it to my budget to buy one for $50-$100. Running preferrably OS 4.1, of course.

Add a software bundle with some games and an artpack like @Signal mentioned and it may help sell more units. An AmigaOS SDK with programming documentation/tutorials also should be included (like is done on the RPi).

@tlosm

Quote:
tlosm wrote:
just for evalutation ... check this arm machine specs

https://amperecomputing.com/wp-content/uploads/2019/01/eMAG8180_PB_v0.5_20180914.pdf

Now that's a NG Amiga. It would make a nice NG Video Toaster system, too.

@Signal

Quote:
Signal wrote:
Get back to music, art, video, presentation, games, and the many other things that made people want to be at home with their Amiga, especially if they are using a mainline computer/OS at their job.

This is my thinking also. I would like to see more updated software to help with artistic creation. These art forms of music, art, presentations, games are what really helped to drive the Amiga platform. The multi-tasking and performance of the machine helped provide a good solution for such activities. I'm glad that at Amiwest 2018 someone asked Trevor about re-releasing some of the older creative-type apps (some of which A-Eon owns) to be available for purchase again. It's a good idea, I think.

 Status: Offline
Profile     Report this post  
OlafS25 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 10-Feb-2019 9:57:41
#154 ]
Elite Member
Joined: 12-May-2010
Posts: 5451
From: Unknown

@ne_one

it is the wrong question... first you must answer what customer needs you want to satisfy, basically "why" someone should buy your product and where it has a advantage to competition. Even if the beloved AmigaOS would suddenly have all modern features (SMP, 64bit, MP) it would still have no advantages to other platforms but still not much software. So first decide where to go and then what hardware is needed and what OS features and software has to be implemented. For me we are all in a retro market, no chance to become mainstream again. Even worse implementing f.e. SMP will break software (on Aros that already was tested). So I prefer what we have now and ARM with RPi because there is a huge community open for alternatives. Will it make Aros mainstream? Of course not. Will it potentially bring more users. Yes propably, expecially with the cheap hardware. Vampire + standalone are a big chance in retro market too. Regarding AmigaOS, I never heared anyone offering a serious plan where to go.

 Status: Offline
Profile     Report this post  
OlafS25 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 10-Feb-2019 10:05:02
#155 ]
Elite Member
Joined: 12-May-2010
Posts: 5451
From: Unknown

@davidf215

to get updated software on AmigaOS running natively you would need someone risking really big money. We saw that with Trevor acquiring several software projects. If the software would have stayed 68k and bugfixes and new features would be added there would have been some potential but to get it running natively you must invest a lot of work by skilled developers (who are rare and of course want to get money) and then you have no new features and just running it natively. To get something that can compete with modern software you would then need to invest lots of time in it with high risk to to get the money back.

 Status: Offline
Profile     Report this post  
Hypex 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 10-Feb-2019 15:23:16
#156 ]
Elite Member
Joined: 6-May-2007
Posts: 9484
From: Greensborough, Australia

@OlafS25



 Status: Offline
Profile     Report this post  
ne_one 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 10-Feb-2019 19:50:26
#157 ]
Cult Member
Joined: 13-Jun-2005
Posts: 854
From: Unknown

@OlafS25

Quote:
it is the wrong question... first you must answer what customer needs you want to satisfy, basically "why" someone should buy your product and where it has a advantage to competition


Twenty five years is enough time to conclude whether or not the Amiga market is viable. The sellability of the brand has never been the issue, it's the succession of incompetents and fraudsters that have controlled it.

It's also not an either/or proposition.

It simply doesn't make sense to invest any further in a 35-year-old OS, whether it's being emulated or running natively. If you want to preserve the past, there are already lots of options that are being actively supported to live forever in the Commodore or Hyperion eras of the platform.

But... if you want to attract revenue beyond the retro/hobbyist community, you need general purpose appeal that offers more than the ability to run old games. That doesn't mean being competitive, it means being useful and opening up the door for more than a one-time novelty purchase.

If anyone truly believes that moving from the bottom up makes more sense I'm anxiously awaiting suggestions.

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 10-Feb-2019 22:47:19
#158 ]
Cult Member
Joined: 14-Mar-2007
Posts: 654
From: Kansas

Quote:

Fl@sh wrote:
Usually I agree with you in many many things and also this time you could be right, but.. a good plan is always a simple plan and your proposal is a bit difficult to realize. I think your best result could lead to a renewed retrocomputing interest, just like vampire project, nothing more.
A home made 68k asic will never be faster than any arm core present in today low price consumer devices.


The primary objectives of a 68k ASIC would be to reduce the cost of the CPU below that of off the shelf ARM CPUs, reduce the cost of a board with a SoC and add custom hardware support to retain AmigaOS compatibility while adding modern features like SMP and security. Using an older die size practically guarantees that the ASIC would not be competitive in performance or energy efficiency to hardware at a smaller die size. Performance and energy efficiency could be compared to older hardware at the same die size to see how competitive it is as the technology could be brought forward. The performance and energy efficiency of 68k CPUs has historically been impressive. I mentioned hiring an experienced and professional CPU designer so this would be a "professional" design. Fabless semiconductor businesses have become popular with some pretty good "home made" designs (Qualcomm, Broadcom, Nvidia, MediaTek, Apple, AMD, HiSilicon, Xilinx, Marvell, UniGroup).

I consider it a necessity to target markets outside of the "retro computing" Amiga. The Amiga SoC ASIC likely could be used to produce a simple Raspberry Pi like device in the $25-$50 U.S. range (FleaFPGA was $45 without an ASIC or mass production). While this may be worth producing after an ASIC is created, its main appeal would be to Amiga users and I worry that it alone would not pay for development costs to produce an ASIC. The price needs to be low enough to trigger impulse buying but not just from Amiga users. I expect it is "simpler" to trigger impulse buying from a wider range of "retro computing" users than to upgrade AmigaOS features and try to convert non-Amiga users. It is probably easier to sneak a functional AmigaOS into a cheap (embedded) product to get users hooked than to *pay* users to switch to AmigaOS on commodity hardware.

Quote:

So I say: why reinvent the wheel?

We already have a good start: the RPi3 developer machine (and next comings), is module expandilbe, is opensource and is linux compatible; if the plan will fail you'll have a linux developer machine to fun with or you'll can simply resell it in few days...


I agree that there is not much user risk with the Pi due to the price. I worry that the compatibility and retro appeal will not be good enough for 68k Amiga users to upgrade and that performance will not be good enough for Amiga PPC users to upgrade. The Pi is cheap but better compatibility and performance are likely in UAE on x86_64 hardware which most users already own. Some Amiga users will convert though dividing the Amiga user base more. Converting existing Pi users without Amiga roots will be difficult. There is a lot to like about the Pi but it is far from a perfect solution for the Amiga.

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 11-Feb-2019 0:20:04
#159 ]
Cult Member
Joined: 14-Mar-2007
Posts: 654
From: Kansas

Quote:

hth313 wrote:
The Amiga situation is more of a software problem than a hardware problem.

While a 68k ASIC would be nice, I think any such approach need to be done with a focus on the embedded field. This is where it may be needed. There have certainly been a huge range of architectures used in this field over the years, so one more is not anything strange. This market is very different from the desktop market in this regard, as it is very diverse.

Any potential Amiga coming out from it would be a lucky spin-off.


The embedded market is very diverse too. Specific (custom) processing may be required or more general purpose processing closer to desktop use. For specific embedded processing, the FPGA has become more popular. Even some servers have moved to using FPGAs for processing which is not efficiently handled by a CPU (this is an area Gunnar von Boehn worked in at IBM in Germany). The one place FPGAs have not been used much for custom processing is the desktop yet the addition of an FPGA could add value for both desktop and embedded use.

Quote:

Going to the software side, we have RPi/ARM and it is here now. You can order one today for little money and have it delivered home in a week or so. It may not be perfect in every (performance) aspect, and it will not fit everyones desire, but it certainly is common, it can do big endian and there are lots of appealing things with it. Future improvements will be done to it without us having to get involved in it.

I could go on and on why the CPU (and especially variant used) in the device is highly irrelevant to virtually everyone. I do not have to go any further than myself for that. But meh, no one cares.


I have no argument about the price and availability of the Pi. Big endian data accesses may help the emulation performance but compatibility will be through emulation at a fraction of the performance of an energy efficient ARM CPU. I expect a major "software problem" until most of the software is converted to BE ARM which is an unusual compiler target (code remains in LE which is weird). I am not convinced that BE (really mixed BE/LE) ARM is a better choice than LE ARM.

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 11-Feb-2019 2:05:58
#160 ]
Cult Member
Joined: 14-Mar-2007
Posts: 654
From: Kansas

Quote:

davidf215 wrote:
In what way is the AmigaOS part of the Embedded market? The biggest part of the embedded market (according to the chart you referenced) is the Industrial industry. I don't know of any Amiga solution that is available for that market.

Also why would you see the Amiga as part of the embedded market when the Amiga has been a desktop solution? It could be an embedded solution, but that is difficult as there aren't many (any) embedded solutions for it that I've seen. Maybe this is an area of opportunity.


I believe the 68k AmigaOS has more competitive features for the embedded market than the PPC AmigaOS does for the desktop market. It is possible to be successful in the embedded market by being very good in some areas even if weak in others. The desktop market doesn't allow trade-offs like this. Lack of SMP support, full memory protection and security are devastating to the desktop market but there are embedded applications where they are not needed. The 68k AmigaOS has not been used in recent embedded products because of lack of affordable hardware and lack of support.

The 68k has been recently used for industrial embedded markets with the Fido.

http://www.innovasic.com/products/fido1100-communication-controller

Quote:

I guess until Amiga software really pushes the envelope of the available CPU power, then 32 bit would probably be okay. I'd go along with a multicore 68k chip if one was available, but if a multicore 68k chip isn't available, skip to the next 32bit multicore option (preferably one to which AmigaOS could be quickly ported).


I believe it is important to have an upgrade path to 64 bit even for embedded use. A 32 bit CPU can often be better performance, smaller footprint, more energy efficient and cheaper though. A 32 bit CPU with a small footprint can go further without needing a 64 bit CPU. The Raspberry Pi certainly didn't need a 64 bit CPU as it lacks the memory and expansion options to take advantage of it.

Quote:

That's a neat 8bit chip. It's cool. Not sure how a next gen Amiga could make use of it.


I don't want to go back to 8 bit CPUs as they are a pain to program and use an OS but the embedded market makes some harsh trade-offs for features that are important to that application. There is a large enough market with 32 and 64 bit CPUs as ARM has shown.

Quote:

I agree. But the footprint will shrink in time, and, admittedly, probably never be as small as a 32. I think many people don't really care about power consumption, though. The PowerMac G5 was a good example of that.


A smaller footprint usually gives performance (more on low end hardware) as well as energy efficiency. The high "power consumption" of the G5 made the whole system more expensive and louder which some people do care about.

Quote:

Unimpressive, sure. But I don't think that's entirely caused by price issues. I suspect; however, that it's (specifically with the X5000) price and software. For example, more would buy if an office suite, like LibreOffice, was available. This way the computer has more value as it could be used for retro gaming and some "real-life" productivity work.


AmigaOS PPC hardware would need SMP support as well as more software to realize the full value of the hardware but then it would still be overpriced.

Quote:

"Too expensive" is relative. I bought my A1200 for about the same price that the Tabor will be. And "not cheap enough" is also relative. Not everyone wants to experience gaming (or whatever) on slow systems, and a Pi Amiga would be slower than Tabor. I do think a standalone Amiga device around the $100 price (or that of a RPi) would be a good addition to the Amiga market.


Classic 68k Amiga hardware does not offer good value either. The situation is not conducive to expanding the Amiga user base.

Quote:

Yes, but I wouldn't consider any 68k Amiga a NG Amiga unless it was 4 or 8 core in the GHz range.


I expect a 68k ASIC would have better performance than some of the older "NG" Amiga PPC CPUs. Hardware customizations to enable SMP in an Amiga compatible way could make a 68k ASIC seem higher performance. A standard SIMD unit could help too.

Quote:

An NDA for AmigaOS 4 source code between two companies could work for this. Otherwise, use Aros. Aros source is already available.


I expect the AmigaOS will leave the clenched fist of Hyperion at some point considering their business ineptness. I hope the sources from AmigaOS 4 are not lost. Maybe the next owner will realize hiding away the sources does not help proliferation.

 Status: Offline
Profile     Report this post  
Goto page ( Previous Page 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 Next Page )

[ home ][ about us ][ privacy ] [ forums ][ classifieds ] [ links ][ news archive ] [ link to us ][ user account ]
Copyright (C) 2000 - 2019 Amigaworld.net.
Amigaworld.net was originally founded by David Doyle