Click Here
home features news forums classifieds faqs links search
6071 members 
Amiga Q&A /  Free for All /  Emulation /  Gaming / (Latest Posts)
Login

Nickname

Password

Lost Password?

Don't have an account yet?
Register now!

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

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

IRC Channel
Server: irc.amigaworld.net
Ports: 1024,5555, 6665-6669
SSL port: 6697
Channel: #Amigaworld
Channel Policy and Guidelines

Who's Online
21 crawler(s) on-line.
 123 guest(s) on-line.
 1 member(s) on-line.


 amigakit

You are an anonymous user.
Register Now!
 amigakit:  2 mins ago
 OneTimer1:  8 mins ago
 NutsAboutAmiga:  11 mins ago
 kolla:  24 mins ago
 Gunnar:  27 mins ago
 Comi:  53 mins ago
 vox:  1 hr 39 mins ago
 zipper:  1 hr 42 mins ago
 BigD:  2 hrs 46 mins ago
 OlafS25:  2 hrs 48 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
Fl@sh 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 24-Feb-2019 9:52:49
#221 ]
Regular Member
Joined: 6-Oct-2004
Posts: 253
From: Napoli - Italy

@hardwaretech

Next year will be firsts macosx and windows notebooks on arm.
Maybe there will be also desktops to follow.
So Linus opinion about arm servers will change for sure.

Future is arm inside, it’s enough simply to see.

_________________
Pegasos II G4@1GHz 2GB Radeon 9250 256MB
AmigaOS4.1 fe - MorphOS - Debian 9 Jessie

 Status: Offline
Profile     Report this post  
Hypex 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 24-Feb-2019 15:19:06
#222 ]
Elite Member
Joined: 6-May-2007
Posts: 11180
From: Greensborough, Australia

@BigD

Quote:
It still leaves the door open for an Arm version if a port could be justified for use on the RPi.


It could be. But then Amiga buyers would expect better than a computer that might be as powerful as their phone but almost as small. Right now an AmigaOne Xxxxx would be considered less powerful than a budget phone* but also comes on a full sized motherboard.

* For comparison with an average entry level phone, of total effecitive computing power, a quad core ARM at 1Ghz.

 Status: Offline
Profile     Report this post  
Hypex 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 24-Feb-2019 16:18:16
#223 ]
Elite Member
Joined: 6-May-2007
Posts: 11180
From: Greensborough, Australia

@Fl@sh

It's rumoured yes. But I don't see it. It just looks too unrealistic to expect ARM to match x86/64.

Sure the A76 will get the gigahertz up. But what about cores? Address space? And RAM speed? Not to mention SIMD, do ARM vectors now match the huge size of x86 now?

It's said Apple want to converge on one CPU for the whole market. That's nice. They could do that on Intel right now. But if they chose ARM then there will still be universal binaries with Intel/ARM for years to come. But what does this matter? Android uses Java so why couldn't Apple develop a virtual CPU for a perfectly portable binary format? They've been in use for years. In any case, if all apps are to go through the app store, then a developer can simply compile a uni binary. The app store can "store" two versions of the app and deliver the one matched to the CPU on the device itself. Can't be too hard, websites can find out what CPU and OS a person is using.

Another thing is the need to emulate x86 and Windows, on ARM, apparenly. In this case, macOS is not doing it's job. Why should anyone these days still need Windows? In this case, developers need to throw off the Windows is all paradigm, since in the modern age this isn't the case any more. So perhaps there needs to be better integration for developers to easily develop cross platform applications.

 Status: Offline
Profile     Report this post  
Fl@sh 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 24-Feb-2019 16:38:55
#224 ]
Regular Member
Joined: 6-Oct-2004
Posts: 253
From: Napoli - Italy

@Hypex

Also windows will go to arm, at least for notebooks and tablets (or surface).
Next year autonomy will be delivered in days not more in hours.
Oh yes.. computers will be always on connected, 5g revolution is coming and all you have now is already obsolete.

All these features and others too will be possible due arm adoption.
So.. The future is arm!
if Amiga involved companies are enough smart they can agree each other and produce a new AmigaOS 5.0 with a low cost hardware based on arm and let the story continue..

_________________
Pegasos II G4@1GHz 2GB Radeon 9250 256MB
AmigaOS4.1 fe - MorphOS - Debian 9 Jessie

 Status: Offline
Profile     Report this post  
bison 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 24-Feb-2019 20:51:56
#225 ]
Elite Member
Joined: 18-Dec-2007
Posts: 2112
From: N-Space

@Fl@sh

Quote:
So.. The future is arm!

Well, maybe. Maybe not. It seems pretty safe to say that ARM will have a part in the future, though how dominant is hard to tell.

There's that famous quote, attributed to Yogi Berra: "It's tough to make predictions, especially about the future."

_________________
"Unix is supposed to fix that." -- Jay Miner

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 1:42:22
#226 ]
Super Member
Joined: 14-Mar-2007
Posts: 1968
From: Kansas

Quote:

Fl@sh wrote:
Also windows will go to arm, at least for notebooks and tablets (or surface).
Next year autonomy will be delivered in days not more in hours.
Oh yes.. computers will be always on connected, 5g revolution is coming and all you have now is already obsolete.


Let's take a closer look at some ARM literature directly from their website.

Accelerating mobile and laptop performance: Arm announces Client CPU roadmap
https://www.arm.com/company/news/2018/08/accelerating-mobile-and-laptop-performance

Quote:

News highlights

o Arm unveils its first-ever public CPU forward-looking roadmap and performance numbers


The "roadmap" is only next generation CPU code names and expected fab processes. The performance numbers are all projected.

Quote:

o Client CPU’s expected to deliver year-over-year performance improvements of >15% for compute through 2020


Projected improvements again.

Quote:

o Arm positioned for laptop share gain with roadmap designed for 5G always-on, always-connected devices


Propaganda and wishful thinking. Where is the news?

Quote:

Over the last five years, advances in Arm technology have brought desktop-class PC performance into our smartphones, fundamentally changing how we use technology in our daily lives. This is a direct result of Arm’s annual cadence of introducing new world-class CPU designs, which have delivered double-digit gains every year in instructions-per-clock (IPC) performance since 2013. Arm is now applying this same design leadership, along with optimizing for the latest advances in process technology from our foundry partners, to enable the PC industry to overcome their reliance on Moore’s law – which has definitely slowed – and deliver a high-performance, always-on, always-connected laptop experience. It’s an experience that will be a necessity as 5G enables an entirely new world of connected possibilities.


Commodity foundry die shrinks "enable the PC industry to overcome their reliance on Moore’s law"? Huh? What does this have to do with 5G technology? Let me translate. Die shrinks have made our performance acceptable for desktop and laptop use while Intel is having trouble with the 10nm fab process. Pay no attention to AMD behind the curtain.

Quote:

The latest example of this is our recently launched Cortex-A76 CPU. The Cortex-A76 is delivering an unprecedented 35% gain in performance compared to the previous generation, without compromising Arm’s efficiency leadership. This is an industry milestone for multiple reasons. Firstly, it’s the foundational CPU IP for the first 7nm SoCs expected to be in production later this year. More importantly, Cortex-A76 represents the continuation of the trajectory that will increase performance at a staggering pace, enabling consumers to do more with their smartphones and level the performance playing field against mass-market laptop CPUs from the competition.

Cortex-A76 laptop-class performance comparison


There is a chart showing the Cortex-A76 (3GHz, Projected) benchmarking at the same integer performance level as the Intel Core i5-7300U (3.5GHz, Turbo) and using less than 1/3 of the power. Pretty good until we read the fine print that the Cortex-A76 info is a projection at 7nm and looking up the Intel CPU we see it is at 14nm. Wow! It took 2 die shrinks to catch the performance? Embarrassing! ARM is insulting the intelligence of people who understand the technology here. Two die shrinks can give a core almost 50% performance boost and reduce power by 75% (15W * .25 = 3.75W). ARM isn't close!

Quote:

Some important details you need to know about our client CPU roadmap:

o The follow-up to Cortex-A76 will be codenamed ‘Deimos’ and delivered to our partners in 2018. Optimized for the latest 7nm nodes, ‘Deimos’ is based on Arm DynamIQ technology and is expected to deliver a 15+ percent increase in compute performance.

o In 2019, the CPU codenamed ‘Hercules’ will be available to Arm partners. ‘Hercules’, also based on DynamIQ technology, will be optimized for both the latest 5nm and 7nm nodes. ‘Hercules’ continues the trajectory of increased compute performance, while also improving power and area efficiency by 10 percent (in addition to the efficiency gains achievable from the 5nm process node).


Look what we can do with die shrinks. We don't have single core single thread performance so we will give you lots of our cores that use different amounts of power called DynamicIQ (used to be BIG.little). When your laptop battery is low, we will switch to a low power core which will drain your battery slower while doing even less work.

Quote:

2018 was an important first step in expanding the Arm PC ecosystem and showing the world that we’re no longer bound to the idea that process technology will only incrementally improve every two years and that a laptop will need charging every few hours. The pace of innovation that transformed smartphones into the compute platform of choice is now powering and transforming the larger screen devices. The question now for the broader PC industry is: Are you ready to liberate yourselves from the slowing pace of Moore’s law and deliver the mobile productivity experience consumers and businesses will need as we enter the 5G era?


Your laptop battery needs to last all day long because it takes that long for your programs to finish.

I'm sorry, most of this literature is propaganda and apples and oranges comparisons. It lacks news and anything of any real substance.

The only chance ARM has in the desktop and laptop markets is to get single core single thread performance up as users don't like to wait and it is good for energy efficiency. It is the Holy Grail of CPU performance and RISC doesn't have it. AArch64 needs 2 die shrinks to pull ahead of x86_64. PPC tried and failed against x86 (32 GP registers vs 8 GP registers) despite keeping pace with die shrinks for awhile. POWER9 is only about a die shrink behind.

 Status: Offline
Profile     Report this post  
Fl@sh 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 8:24:06
#227 ]
Regular Member
Joined: 6-Oct-2004
Posts: 253
From: Napoli - Italy

@matthey

In smartphone android market Intel x86 tried to compete with arm.
Intel cpu process was always ahead arm, but their performance per watt was always lower than arm.
The language used to compile programs was java, high general purpose language cpu independent where compilers tends to assemble better on x86 due to better know architecture.
As result we know Intel failed and exited from a very large smartphone market.

Another point about arm is their die size, it is about 1/4 of x86, this lead to a small price of cpu compared to Intel or amd.

We can follow up next 12 months and see if there will be a progress, or not, in laptop and maybe desktop market due arm incoming.

Last edited by Fl@sh on 25-Feb-2019 at 04:16 PM.
Last edited by Fl@sh on 25-Feb-2019 at 08:49 AM.

_________________
Pegasos II G4@1GHz 2GB Radeon 9250 256MB
AmigaOS4.1 fe - MorphOS - Debian 9 Jessie

 Status: Offline
Profile     Report this post  
megol 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 10:54:47
#228 ]
Regular Member
Joined: 17-Mar-2008
Posts: 355
From: Unknown

If we could find the die size of the Cortex A76 at 12nm we could have a good comparison with AMD Zen+ cores (which IIRC have the same size as Zen cores at 14nm*), 7mm^2.

Then we'd have to compensate for L2 cache size and performance optimizations, cores designed for higher clock rates tend to be larger.

(* not shrinking the cores themselves just taking advantage of the improved process)



 Status: Offline
Profile     Report this post  
Fl@sh 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 12:17:51
#229 ]
Regular Member
Joined: 6-Oct-2004
Posts: 253
From: Napoli - Italy

@megol

Yeah! You centered the target, very often we forgot to compare different cpu with same die size and same process fab.
Usually arm cpus are much smaller than x86, at same fab process, I remember they usually are about 1/4 die size.

Until now they were used for low energy devices and I agree their performances was always lower than x86, but cpu was much cheaper at same time.
In few words they targeted a different market.

But.. what happen if someone (i.e. Apple) will produce new arm cpu with same transistor count of x86 and same fab process and same thermal dissipation?
I guess we could have a much faster cpu, especially in multicore apps.
ARM ISA is very powerfull and it evolves in months, not in decades as x86 did until now.

_________________
Pegasos II G4@1GHz 2GB Radeon 9250 256MB
AmigaOS4.1 fe - MorphOS - Debian 9 Jessie

 Status: Offline
Profile     Report this post  
Hypex 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 15:50:19
#230 ]
Elite Member
Joined: 6-May-2007
Posts: 11180
From: Greensborough, Australia

@Fl@sh

Quote:
Also windows will go to arm, at least for notebooks and tablets (or surface).


They also ported to PowerPC once (Windows NT) but that disappeared. Perhaps they should have waited and released it for PPC Macs.

Quote:
Oh yes.. computers will be always on connected, 5g revolution is coming and all you have now is already obsolete.


That usually is always the case so no big deal there.

Quote:
if Amiga involved companies are enough smart they can agree each other and produce a new AmigaOS 5.0 with a low cost hardware based on arm and let the story continue..


But, Amiga Inc isn't around any more to promote the AmigaOS 5 idea. With those Amiga objects. So it would take a great deal of work to build a new AmigaOS from the ground up.

Then the Amiga people come out of the woodwork. But is it 68K compatible? Does it play all my A500 games?

Or. Following on. Does my OS4 software work on it?

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 21:10:40
#231 ]
Super Member
Joined: 14-Mar-2007
Posts: 1968
From: Kansas

Quote:

Fl@sh wrote:
In smartphone android market Intel x86 tried to compete with arm.
Intel cpu process was always ahead arm, but their performance per watt was always lower than arm.


Do you think ARM has better performance per watt than PowerPC?



Steve Jobs' WWDC 2005 Keynote presentation slide of performance per watt PowerPC vs Intel.
https://www.anandtech.com/show/1936/7

PPC fans like to say how Intel was first to a new fab size but POWER/PPC was more competitive in this regard than other RISC architectures.


https://www.realworldtech.com/escape-from-x86/3/

From various data, it looks to me like POWER is more competitive than ARM in performance per Watt although it has had a single thread performance deficit vs x86_64 for some time. The early Alpha was the only RISC CPU to outperform the x86 in single thread performance and that was with extreme tradeoffs of power and area for performance. After most x86 instructions became single cycle, the only major architecture I'm aware of that beat the x86 in Power, Performance and Area (PPA) was the 68k (the 68060 took the triple crown from the Pentium).

CPUs do have power thresholds. ARM CPUs are usually good at area (fewer transistors) and static (idle) power. The x86_64 architecture is fat and has major inefficiencies which use power. The single thread performance is so good that the improved energy efficiency (good performance per watt) more than offsets the power tax for high performance CPUs while at low power thresholds the tax is too much. ARM's AArch64 has also become fatter than Thumb2 reducing the area and static power advantages although likely improving the single thread performance (but it appears not enough to compete with x86_64). ARM advertises low dynamic power per MHz (Watts/MHz) but higher numbers are better within a tolerable threshold as work is done faster using less total energy. Intel Atom cores were doing work faster than ARM cores but had trouble meeting low power thresholds with the x86/x86_64 tax. Single thread performance is important for lower performance cores too as it improves energy efficiency with work being done faster before idling.

Quote:

The language used to compile programs was java, high general purpose language cpu independent where compilers tends to assemble better on x86 due to better know architecture.
As result we know Intel failed and exited from a very large smartphone market.


The Atom CPUs could probably not meet the low dynamic power requirements for smart phones. Also, smartphones are idle much of the time where ARM has been very good at reducing power.

Some ARM CPUs had the Jazelle DBX extension to execute byte code but it had limited success and is now deprecated.

https://en.wikipedia.org/wiki/Jazelle

Java byte code was very inefficient. Android used Dalvik (16 bit encoding) bytecode and then ART ahead of time compilation.

https://en.wikipedia.org/wiki/Android_Runtime

This is still inefficient but necessary due to ARMs lack of standardization. Android is far from efficient with its Linux monolithic kernel anyway. The 68k AmigaOS is way more efficient while using a much smaller footprint.

Quote:

Another point about arm is their die size, it is about 1/4 of x86, this lead to a small price of cpu compared to Intel or amd.


ARM cores are small because they are weak. Add aggressive OoO execution or fine grain multi-threading necessary to improve performance and their area advantage will be reduced. It was easier for them to change to a more powerful RISC ISA than to increase complexity in the cores even though this also increased the area of their cores. Advertising propaganda is also cheaper than making high performance cores. IBM actually made progress with fine grain multi-threading for POWER9 to improve efficiency even though the technology is specialized toward overall dynamic power reduction. ARM makes it easy for customers and gives lots of weak cores for the price. Maybe IBM could switch to ARM and make some progress as AArch64 looks to me like it has better performance traits than POWER with some increased complexity. It is unlikely any RISC CPU can compete in single core single thread performance with x86_64 but the overall efficiency could be enough for IBM ARM to take the server market.

Quote:

We can follow up next 12 months and see if there will be a progress, or not, in laptop and maybe desktop market due arm incoming.


Apple has been talking about moving to all ARM for almost a decade now. I think they would like to as profit margins would be higher but the performance per watt and especially single thread performance are not there. Maybe they could get away with it as Apple is not as big into games but then it would seem counter to the move away from PPC on the surface. Looking deeper, it would be more about profit margins and in house ARM customization. It would probably be safer to buy out AMD and use x86_64 if they could get the deal by the U.S. DoJ though.

Last edited by matthey on 25-Feb-2019 at 09:22 PM.

 Status: Offline
Profile     Report this post  
Rose 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 22:32:10
#232 ]
Cult Member
Joined: 5-Nov-2009
Posts: 982
From: Unknown

@matthey

Quote:
It would probably be safer to buy out AMD and use x86_64 if they could get the deal by the U.S. DoJ though.


There's a one big snag on that idea. It's been semi public information for years that AMD's x86 license isn't transferable in case of AMD getting aquired by 3rd party.

 Status: Offline
Profile     Report this post  
vision 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 23:10:41
#233 ]
Regular Member
Joined: 8-Jun-2005
Posts: 480
From: Unknown

@BigD

Quote:
It is very easy to demand these radical changes to an OS that has struggled to implement SMP and update classic apps to PPC despite that being a small challenge compared to porting the whole OS and apps and adding backwards compatibility to an entirely new CPU line I vote Hyperion and Trevor each give every last active Amiga owner $20 each and the AmigaOS 4.1 FE source code and a free X5000 and a pre-release Tabor board x86-64 didn't save BeOS and neither will a port of AmigaOS to x86-64 with no software.


I don't know if you believe those old lies saying those were "radical changes" impossible to do in the last 20 years, or just a cover up to those liers and responsibles of this ridiculous (even more than catastrophic or irreversible, being also both). While os4 fanatics keeps repeating that like mantra, I small group of aros developers already did that against your support. they already surpassed os4 in many areas, and keep improving and leaving os4 camp all the shame. And they were not even a 10% of amigaland devs. Imagine if 100% of devs were all supporting this same platform.

And that x86-64 won't save the day it is one of the most cynic statements I have seen around here. So why are THOUSANDS of WinUAE users out there? how does the situation improve by changing a x86 64 by some stupid/obscure/unavailable processor? Come on! speak the truth for once: You all don't wanna accept x86 64 just because it is wintel. Nothing else. And I have shouted intel outside! too, but in the 90s. We are headed for 2020 and XR. Grow up!!

Quote:

Just because it's cheap will not automatically bring new users. BootCamp is cheap but I only use it on my Mac for specific software. What specific software would tempt a new user to boot AmigaOS rather than Windows or Linux on their x-86-64 PC? If you want an Amiga(One) buy the whole package. The only machine that WOULD grow market share outside of PPC would have been the RPi but we would now be late to market and RiscOS has taken the glory there IMHO.


Cheap trap with cheap retoric: it is not about attracting new users in a day. It is FIXING the platform and retaining current and recently lost users. Only with a viable hardware and united platform can you attract developers (old or new) but with.. 68k? power9? the facepalms of any old amiga users or whatever serious developer reading those statements must be epic. Arm could be ok, but it is not yet a good performer.

Last edited by vision on 25-Feb-2019 at 11:16 PM.
Last edited by vision on 25-Feb-2019 at 11:12 PM.

 Status: Offline
Profile     Report this post  
BigD 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 25-Feb-2019 23:41:23
#234 ]
Elite Member
Joined: 11-Aug-2005
Posts: 7307
From: UK

@vision

Quote:
Cheap trap with cheap retoric: it is not about attracting new users in a day. It is FIXING the platform and retaining current and recently lost users. Only with a viable hardware and united platform can you attract developers (old or new) but with.. 68k? power9? the facepalms of any old amiga users or whatever serious developer reading those statements must be epic. Arm could be ok, but it is not yet a good performer.


It is pointless FIXING the platform if there's then no compatible software of interest (ala AROS) and people with a PC rig will 99.999999% of the time boot Windows or Linux which DO have applications!

I have NEVER tried MorphOS or WinUAE though I'd be interested in buying a Tabor or an A.L.I.C.E. laptop (if they can sort out a European warranty and distribution deal). I also have an iBook G4 and STILL MorphOS seems like too much trouble to ditch PPC Mac OS Tiger. That Mac OS has the Classic environment which is so useful compared to what MorphOS would offer me. I think I'd prefer to mess around with my old scanner and The Settlers 2 than play the Amiga version of WipeOut 2097 to be honest.

Also and most importantly....

....
1) WHO IS GOING TO PAY FOR THE X86-64 OS PORTING?

2) WHO WILL WRITE THE LAUNCH SOFTWARE / 68k / PPC ports or emulation layer (THEY'D HAVE TO DO THIS TO STAND ANY CHANCE OF RECOUPING THEIR COSTS)?

3) WHO IS GOING TO BUY THIS WHEN THE 10 PEOPLE INTERESTED HAVE ALREADY DOWNLOADED AROS FOR FREE!!!!

WHY? Why? Why?

Last edited by BigD on 25-Feb-2019 at 11:51 PM.
Last edited by BigD on 25-Feb-2019 at 11:44 PM.

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

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 26-Feb-2019 1:55:04
#235 ]
Super Member
Joined: 14-Mar-2007
Posts: 1968
From: Kansas

Quote:

Rose wrote:
There's a one big snag on that idea. It's been semi public information for years that AMD's x86 license isn't transferable in case of AMD getting aquired by 3rd party.


That is true but Intel would want to renegotiate the cross license worse than AMD under Apple ownership. Apple could say no more AMD64 for Intel and it would be the end of the x86_64 architecture. Apple could then have the AMD engineers go back to designing ARM CPUs (they have past experience) and low power GPUs. Gamers and the DoJ would not be happy with Apple though.

 Status: Offline
Profile     Report this post  
davidf215 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 26-Feb-2019 4:46:03
#236 ]
Member
Joined: 14-Feb-2010
Posts: 95
From: Texas

@matthey

Quote:
Let's say the AmigaOS developers pull off the miracle of adding SMP, adding 64 bit support, improving memory protection and adding security without breaking compatibility (AmigaOS and MorphOS developers have been unsuccessful after many years and AROS x86_64 breaks compatibility). The AmigaOS would *not* be attractive in the server and workstation markets due to lack of software. IBM has the same problem with Linux and BSD for POWER and that is why they are supporting ppc64le to ease endian issues with ports (most servers are x86_64 hardware and can use the same software as desktops). The AmigaOS would want big endian for compatibility and it is less POSIX compatible making it that much more difficult to port software. It would take an epic effort to bring the AmigaOS up to par and the best place to start would probably be breaking compatibility.

Is breaking backwards compatibility really a big issue considering 68k JIT and emulation are viable options for older software? Microsoft Windows and IBM OS/2 were able to add memory protection while allowing older software to run in isolated memory locations while helping to prevent crashing the OS. If Windows and OS/2 were able to do it, then AmigaOS should be able to do so as well. I would suggest to simply make the break, enhance JIT, and move on. Dave Pleasance has even said that, in time, the AmigaOS would have an end of life cycle and need to be replaced, so breaking backwards compatability with 68k via JIT is a good way to move forward.

Quote:
Steve Jobs' WWDC 2005 Keynote presentation slide of performance per watt PowerPC vs Intel.
https://www.anandtech.com/show/1936/7

Apple eventually moved to x86 in spite, though.

Quote:
After most x86 instructions became single cycle, the only major architecture I'm aware of that beat the x86 in Power, Performance and Area (PPA) was the 68k (the 68060 took the triple crown from the Pentium).

If this is true, then it's interesting as to why Motorola hasn't released a multicore 68k chip.

Quote:
It would probably be safer to buy out AMD and use x86_64 if they could get the deal by the U.S. DoJ though.

That would be very interesting. And, as you said, gamers would not be happy about it.

@BigD

Quote:
I vote Hyperion and Trevor each give every last active Amiga owner $20 each and the AmigaOS 4.1 FE source code and a free X5000 and a pre-release Tabor board

Paying at cost for a Tabor as Trevor has mentioned would be nice although I'll buy one regardless.

Quote:
x86-64 didn't save BeOS and neither will a port of AmigaOS to x86-64 with no software. Just because it's cheap will not automatically bring new users.
What specific software would tempt a new user to boot AmigaOS rather than Windows or Linux on their x-86-64 PC? If you want an Amiga(One) buy the whole package.

Yes, available software and a "whole package" are very important.

Quote:
The only machine that WOULD grow market share outside of PPC would have been the RPi but we would now be late to market and RiscOS has taken the glory there IMHO.

ARM, I think, would still be a good pathway moving forward.

@bennymee 

Quote:
Eehh, the price of Xeon Gold is 1000USD+, and that is only the cpu.

Yeah, I saw it's expensive. But it's at least an option for those who want that kind of speed.

@Hypex 

Quote:
So, by the looks of things, talk of anything x86 woud remain equally as a hypothesis if Acer put it to an instant halt.

This would definitely be a show stopper for AmigaOS on x86.

Quote:
Android uses Java so why couldn't Apple develop a virtual CPU for a perfectly portable binary format?

Wasn't the virtual CPU similar to Bill McEwen's idea in his work with the Tao Group? Software that could run on any CPU? I still think this is a good idea. It would render the current poll as unnecessary.

@Fl@ash

Quote:
In smartphone android market Intel x86 tried to compete with arm.

This is a good point

@vision

Quote:
So why are THOUSANDS of WinUAE users out there?

I've often wondered about this, too. But then I wonder that since Aros is available on x86, then why don't those who run WinUAE jump over to Aros for a more native experience on the x86? If the reason that prevents the switch is an issue of compatibility with legacy software, then, IMHO, enhancing JIT and 68k emulation should be a high priority for Aros dev. Perhaps better emulation on Aros would encourage WinUAE users to switch to a native x86 Amiga experience.

 Status: Offline
Profile     Report this post  
Rose 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 26-Feb-2019 14:36:05
#237 ]
Cult Member
Joined: 5-Nov-2009
Posts: 982
From: Unknown

@matthey

Quote:

matthey wrote:
Quote:

Rose wrote:
There's a one big snag on that idea. It's been semi public information for years that AMD's x86 license isn't transferable in case of AMD getting aquired by 3rd party.


That is true but Intel would want to renegotiate the cross license worse than AMD under Apple ownership. Apple could say no more AMD64 for Intel and it would be the end of the x86_64 architecture. Apple could then have the AMD engineers go back to designing ARM CPUs (they have past experience) and low power GPUs. Gamers and the DoJ would not be happy with Apple though.


Unlike Amiga companies, Intel and AMD have competent laywers. It's pretty standard in these kind of contracts to have a safeguard that in this case would protect Intel's right to continue using AMD64.

 Status: Offline
Profile     Report this post  
Hypex 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 26-Feb-2019 16:06:50
#238 ]
Elite Member
Joined: 6-May-2007
Posts: 11180
From: Greensborough, Australia

@davidf215

Quote:
Wasn't the virtual CPU similar to Bill McEwen's idea in his work with the Tao Group? Software that could run on any CPU? I still think this is a good idea. It would render the current poll as unnecessary.


Yes it is. Called AmigaDE. IIRC it only ran on a PC. Well that AmigaDE we were given. That was never ported to OS4.

Java is the same idea. Which is why I wondered back then why they were reinventing the wheel. But given how clumsy it is to run a Java program not in a web browser I could see how the user interface would be improved.

Right now, JavaScript JIT has taken over. Which might deliver a superior experience. But it still looks inefficient compling from source at run time compared to loading virtual machine code directly. Cutting out a middleman or frontman.

 Status: Offline
Profile     Report this post  
megol 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 26-Feb-2019 17:05:40
#239 ]
Regular Member
Joined: 17-Mar-2008
Posts: 355
From: Unknown

@Hypex
https://webassembly.org/

 Status: Offline
Profile     Report this post  
matthey 
Re: Poll of CPU architecture interest for AmigaOS
Posted on 26-Feb-2019 21:12:59
#240 ]
Super Member
Joined: 14-Mar-2007
Posts: 1968
From: Kansas

Quote:

vision wrote:
I don't know if you believe those old lies saying those were "radical changes" impossible to do in the last 20 years, or just a cover up to those liers and responsibles of this ridiculous (even more than catastrophic or irreversible, being also both). While os4 fanatics keeps repeating that like mantra, I small group of aros developers already did that against your support. they already surpassed os4 in many areas, and keep improving and leaving os4 camp all the shame. And they were not even a 10% of amigaland devs. Imagine if 100% of devs were all supporting this same platform.


AROS x86_64 broke compatibility which made it easier to implement 64 bit support and SMP (the SMP support still has issues and there is no full memory protection). Even with the best AmigaOS like technical features and excellent performance/price hardware, AROS x86_64 has not been a huge success. Could it be due to the compatibility they broke?

Quote:

And that x86-64 won't save the day it is one of the most cynic statements I have seen around here. So why are THOUSANDS of WinUAE users out there? how does the situation improve by changing a x86 64 by some stupid/obscure/unavailable processor? Come on! speak the truth for once: You all don't wanna accept x86 64 just because it is wintel. Nothing else. And I have shouted intel outside! too, but in the 90s. We are headed for 2020 and XR. Grow up!!


WinUAE is popular because it offers the best performance/price Amiga hardware which is compatible. It has nothing to do with users liking the x86_64 architecture underneath. In fact, several UAE users I know love the 68k and program it in assembler. Peter Keunecke, the author of the replacement icon.library written in 100% 68k assembler, has no Amiga hardware anymore and only uses UAE. Frank Wille, the author of vasm (based on his 68k assembler phx), vlink, responsible for the vbcc compiler free for all Amiga use, Solid Gold (Amiga original game in 100% 68k assembler), Sqrxz ports, Giana Sisters remake, etc. uses UAE. He thought he found bugs in my vbcc FPU support but they turned out to be lack of precision in the UAE FPU emulation which he realized when he finally tested on his Amiga 3000. UAE feels more like a fast 68k Amiga under the hood as the x86_64 architecture is hidden. If they liked x86_64 then they would use AROS x86_64 and try to make it better but they don't.

Quote:

Cheap trap with cheap retoric: it is not about attracting new users in a day. It is FIXING the platform and retaining current and recently lost users. Only with a viable hardware and united platform can you attract developers (old or new) but with.. 68k? power9? the facepalms of any old amiga users or whatever serious developer reading those statements must be epic. Arm could be ok, but it is not yet a good performer.


You aren't going to unify Amiga users on an x86_64 AmigaOS. It didn't even work to unify Amiga users on PPC where better compatibility is possible. The largest group of Amiga users are still using the 68k and want compatibility even if an x86_64 is hidden somewhere underneath. Look no further than the success of the Vampire and attraction of the Natami (Natami "MX Bringup Thread" had 761487 views) which don't even have a good performance/price. The Amiga users who stuck with the Amiga primarily liked it because of the 68k and/or AmigaOS. Separating these two divides the community.

 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