@iggy

Quote:

iggy wrote: Everyone seems to be on the same page except for Hyperion. But as I don't see a future past the two 64 bit cores that Freescale offers, we will have to change ISAs eventually.

First read THIS.

If Hyperion AND the Linux guys are both working on converting their particular ASM code to C, then doesen't make sense to wait until a point is reached where as little as possible assembler would be required?

Jumping on Hyperion, or screaming about the death of PPC for 10 years will not hurry things along. Neither do lawsuits or angry words.

_________________
Tinkering with computers.

@Signal

I have some past experience with 68K assembly code.
On the 68K well written machine code definitely outperforms C.
But I wouldn't want the odious task of maintaining, upgrading, or moving code generated by an assembler to another ISA.

So I can understand your point, but as a member of the community I do reserve the same right as everyone else here to expect huge amounts time and effort to be expended in order to reach MY goals.

Beyond that perfectly silly suggestion, there is the practical point that the sooner they start planning for and implementing an ISA move, the sooner it will be ready.

@iggy

Quote:

iggy wrote: @Signal But I wouldn't want the odious task of maintaining, upgrading, or moving code generated by an assembler to another ISA. ........ there is the practical point that the sooner they start planning for and implementing an ISA move, the sooner it will be ready.

Which is why it makes sense to first move it to like endian CPU. Then, when as much as possible is C, make the move to a different ISA.

Can you imagine the number of 200 page threads that would clog this site if Hyperion even hinted at a change publicly?


Of course all changes would have to be approved by Mr. KnowItAll.

Last edited by Signal on 17-Dec-2015 at 10:12 PM.

_________________
Tinkering with computers.

@saimon69

Exactly. Back in the day ppc seemed promising, was up to date and probably not a bad choice (good (migration) support by Motorola, right endianess). Even in the Pegasos/A1 XE/SE days ppc was still not too stupid, abeit the drama of Articia showed that the market was virtually dead already. The gap between ppc and x86 was big already and steadily increasing. Then when Apple left the ppc stage ppc really had no promising future with desktop usage.

Somehow I like ppc, but thing is: Let it go when it's dead.
And ppc is virtually dead. _Could_ get resurrected (technicaly as it is not generally bad), but the Amiga market has _absolutely no_ impact on such resurrection. It's smaller than tiny. IMHO it's wiser not to hope for some help from outside (some 3rd player pushing ppc again), but walk the safe paths. I.e. x64. Cheap powerful and will stay on the market for the foreseeable future. ARM could be an alternative but includes too many uncertain factors. x64 is a constant.

Last edited by Zylesea on 17-Dec-2015 at 11:25 PM.

_________________
My programs: via.bckrs.de
MorphOS user since V0.4 (2001)

@Signal

Counter-question... do you think Trevor would invest in new PPC hardware like Tabor if there would be any discussion about a ISA change? I do not think that Hyperion would do such a move without approval by their business partners. Because of that I do not think that there will be any change in next years. From my view it would have made more sense than investing lots of money in new hardware but it is Trevors decisions and risk. Of course any change would have been risky but porting AmigaOS to either X64 or ARM would have offered at least chances. PPC now finally is dead end and a-eon and Acube are the last companies now offering PPC desktops.

Last edited by OlafS25 on 17-Dec-2015 at 11:18 PM.
Last edited by OlafS25 on 17-Dec-2015 at 11:11 PM.

@Zylesea

it is more wise to let all doors open and not try to play only one card. Nobody knows what will come. My two cents.

@OlafS25

Personally I don't think you can trust Ben as far as you could fling him, but that is just an opinion.
I do like Trevor and I have known Paul Gentle even longer so I am comfortable with Varisys.
AND, I like PPCs too (yes, I know some think that perverse).

But if we see new PPC cores after the e6500, I will be surprised.
I'd like a board based on that, although I may have to settle for the X5000's e5500 based cpu.

Unless...the community or a third party develops a new low cost alternative (which, come to think, I have been harping on you guys about - T10XX).

Last edited by iggy on 18-Dec-2015 at 01:50 AM.

Quote:

tlosm wrote: Exactly it was ... And Beos was claimed at that time the AmigaOS replacement when amiga goes to powerpc 1997-1998 ... but something was change ... Any way if PowerPc G5 was developed as 8 core at 3.5 ghz (like Xbox 360 ghz)...

G5 performance was lackluster for being a high end desktop CPU. For all the resources aimed at high performance, it produces as much heat as it does performance. The G3/G4 design (G4=G3+Altivec) is a better design for a lower end target which is a proven efficient design if not high performance (a shallow pipeline like the G3 is efficient but not high performance). The Tabor CPU is based on the G3 design with a few tweaks, 2 cores and more modern die size giving better integer performance but lacks the standard PPC FPU. This is a move in the wrong direction IMO.

G3 - integer+FPU+MMU
G4 - integer+FPU+MMU+Altivec
Tabor CPU - integer+MMU+SoC for embedded

IMO, the "elite" Amiga PPC market wants high end Amigas and are willing to pay the price premium. There hasn't been any new high end PPC designs (POWER processors are not really PPC but close) so where can A-EON/Hyperion go from here? They will continue to lose customers in this market as PPC technology falls farther and farther behind. Also, while PPC (and POWER) has relatively good performance for RISC processors, they need very good compiler support and this is starting to decline (it never met expectations even with direct help from the AIM partners many years ago). The Amiga elite are in danger of becoming more elite and then orphans.

The low end Amiga "masses" market wants affordable Amigas. FPGA boards like the FPGA Arcade and Mist are meeting the lowest end of this range. The 68k processor provides better Amiga compatibility and a smaller footprint here where initially processing power is not a priority (I believe better performance than the last 68k high end systems would greatly increase sales though). This market is currently blocked by the last generation Amiga technology hoarders but the damn is building.

Quote:

Signal wrote: Which is why it makes sense to first move it to like endian CPU. Then, when as much as possible is C, make the move to a different ISA.

The following is a list of CPU endian support.

68k - big endian
PPC - biendian (more efficient and supported as BE)
ARM Thumb 2 - bi-endian (more efficient and supported as LE)
ARM AARCH64 (ARMv8) - bi-endian (more efficient and supported as LE)
x86/x86_64 - little endian

I have not given the long list of RISC processors which are worse off than PPC. There is no other popular big endian (by default) architecture. The most suitable candidate of popular processors is ARM. ARMv8 is the most like PPC and the long list of past RISC processors that failed. The performance is less than x86 but at least the power efficiency is good. There are really only 3 good options for sustainability.

1) Switch to x86 (highest performance but least compatibility)
2) Switch to ARMv8 (moderate performance with fair compatibility)
3) Design own 68k SoC processors (unknown performance with good compatibility)

I like option #3 although #2 is a better option than PPC for the Amiga at this point. With option #1 you might as well select a different name than Amiga and break all compatibility.

Quote:

OlafS25 wrote: Counter-question... do you think Trevor would invest in new PPC hardware like Tabor if there would be any discussion about a ISA change? I do not think that Hyperion would do such a move without approval by their business partners. Because of that I do not think that there will be any change in next years. From my view it would have made more sense than investing lots of money in new hardware but it is Trevors decisions and risk. Of course any change would have been risky but porting AmigaOS to either X64 or ARM would have offered at least chances. PPC now finally is dead end and a-eon and Acube are the last companies now offering PPC desktops.

I expect Trevor has less than positive feedback about the Tabor board decision by now and is seeing how much resources and time is required for software to try to make up for poor hardware decisions. Would there be any ISA change announcements soon? I doubt there will be any announcements until after those Tabor boards are sold. I hope they have a better planning and technology committee before the next hardware is started.

Why are you picking on cdimauro? I think he has contributed the same as I did to 68k and Amiga development. Gunnar was stronger with a diverse group of help but he likes unquestioned subservience. Don't believe everything Gunnar tells you just because he understands the hardware in depth. He doesn't understand people nearly as well. Maybe cdimauro is too pessimistic and negative but the Amiga situation is dire and management seems clueless (the Amiga curse since C= days). The Amiga isn't everything so we just need to let it go if management can't get a clue. People, our friends are more important. I hope you haven't forgotten this.

Last edited by matthey on 18-Dec-2015 at 02:39 AM.

@OlafS25

Quote:

do you think Trevor would invest in new PPC hardware like Tabor if there would be any discussion about a ISA change?

AmigaOS port from 68k to PowerPC did take 3 years... with such speed, Trevor could release two more new boards.

@matthey

Quote:

The Tabor CPU is based on the G3 design

G3 and e500 have similar integer performance, but I don´t think there is close relation (could be wrong of course).

Quote:

The low end Amiga "masses" market wants affordable Amigas. FPGA boards like the FPGA Arcade and Mist are meeting the lowest end of this range.

We don´t know Tabor price yet. If more close to p-cubed pricing (200 USD), it would fully satisfy the low end Amiga "masses" market. Interesting to note, it wouldn´t be the first FPU lacking hardware OS4 was ported to (late eLAP).

@matthey

cdimauro is picking others and calling them liars... yes you are right it is a hobby, I am very comfortable with having a retro system and even retro can be fun. BTW retro could also have a living market, not mainstream of course. Certain people are forgetting that. I at least expect that people treat others politely like they should in real life. Just because it is "online" and you do not see address and telephone number not means you can do everything. When certain people almost call everyone a liar then the person should think if the person is right or not.

And as I wrote (and was attacked by a certain person for that) in my view all NG platforms will always lack the software even if the OS is modernized so 68k offers more chances expecially when new hardware is available (not to forget UAE). 68k (or classic how people here call it) is magnitudes bigger than NG, both in the software base and user base. So IF there is any chance to revive to a certain degree than it is 68k. But I also accept different opinions and do not call anyone liar or stupid when saying something different.

Regarding Gunnar, I know that you had some problems with him. He is a engineer and sometimes technicians are a little different. But at least he stayed on course and managed to complete something. In few weeks lots of people have the chance to use it and can judge it. Finally it was his project as Natami was that of Thomas Hirsch so they have to decide what to do.

@iggy

Quote:

I, because I am already using a PPC system, have no problem supporting and recommeding PPC systems because they work (for our current OS'). But as I don't see a future past the two 64 bit cores that Freescale offers, we will have to change ISAs eventually. While the X5000 is over priced, at $2100 it is a better buy than the X1000 was.

And how many will be jumping to buy future PPC boards when, because of big endian issues in the only modern browser that is left on the platform, they will not be able to surf the web as one would expect today, especially on a 2000+$ computer...
Since MOS is going x64, I doubt Fab will invest much time and resources on PPC Odyssey that is slowly becoming a dead end.

_________________

@OlafS25

Quote:

OlafS25 wrote: @Signal Counter-question... do you think Trevor would invest in new PPC hardware like Tabor if there would be any discussion about a ISA change?

Yes. Because AOS is not to a state yet where a ISA jump could be done quickly enough. If there are discussions or plans for a different CPU it would take years to finally make the jump.

I also believe they would have stayed with the PA6T had it not been for the unfortunate circumstances surrounding it. PPC still has some years to it and for them to say 'We are moving to SnapDragon in 4 years so just put yourself on hold until then' would be certain death.

_________________
Tinkering with computers.

Quote:

pavlor wrote: G3 and e500 have similar integer performance, but I don´t think there is close relation (could be wrong of course).

Compare the pipeline diagrams of the G3 and e500. Notice the shallow pipeline design with reservation stations on both. Look at the 4 cycle loop mode for multiply and divide of both (not good for integer multiply performance needed by multimedia code by the way). There are some differences of course with the dropping of the FPU and multiple core support but this is not a new design.

Quote:

We don´t know Tabor price yet. If more close to p-cubed pricing (200 USD), it would fully satisfy the low end Amiga "masses" market. Interesting to note, it wouldn´t be the first FPU lacking hardware OS4 was ported to (late eLAP).

Tabor should be around $200 U.S. but I doubt it will be unless it doesn't sell well. IMO, most people want affordable plus:

1) standard hardware with an FPU (preferably a standard SIMD unit too)
2) retro compatibility with 68k CPU and Amiga custom chips

If Tabor was successful then there would be multiple production runs. I don't see this happening but maybe I'm wrong. IMO, the game changer in hardware sales would come when a retro 68k Amiga is cheap enough to give away as Christmas presents with a bunch of games installed.

Quote:

OlafS25 wrote: cdimauro is picking others and calling them liars... yes you are right it is a hobby, I am very comfortable with having a retro system and even retro can be fun. BTW retro could also have a living market, not mainstream of course. Certain people are forgetting that. I at least expect that people treat others politely like they should in real life. Just because it is "online" and you do not see address and telephone number not means you can do everything. When certain people almost call everyone a liar then the person should think if the person is right or not.

Most of what cdimauro is saying is true. Some people can't handle the truth. There are a lot of people on here that don't understand the basics of processors or hardware. Yes, AIM cherry picked and gave deceptive benchmark numbers for the G5 (compilers doomed them instead of saving them despite the efforts of 3 large companies). Yes, Intel's x86/x86_64 has beat PPC badly since. Some people paint themselves in corners. I hope everyone decides to be polite and respectful instead of this thread deteriorating.

Quote:

And as I wrote (and was attacked by a certain person for that) in my view all NG platforms will always lack the software even if the OS is modernized so 68k offers more chances expecially when new hardware is available (not to forget UAE). 68k (or classic how people here call it) is magnitudes bigger than NG, both in the software base and user base. So IF there is any chance to revive to a certain degree than it is 68k. But I also accept different opinions and do not call anyone liar or stupid when saying something different.

I agree with most of what you said here. Software development and support comes with a large hardware base and "elite" PPC Amigas are not giving us this. There are many 68k users on diverse hardware (retro, UAE, FPGA) that are ignored despite probably significantly outnumbering PPC Amiga owners. Many would not buy PPC Amiga hardware no matter the price because of less retro compatibility.

Quote:

Regarding Gunnar, I know that you had some problems with him. He is a engineer and sometimes technicians are a little different. But at least he stayed on course and managed to complete something. In few weeks lots of people have the chance to use it and can judge it. Finally it was his project as Natami was that of Thomas Hirsch so they have to decide what to do.

I like Gunnar but he is his own worst enemy. I am pleased that he has continued to work on Apollo and even listened to many of our ideas finally. He is missing out on many good ideas as a committee of one chief and many underlings rather than a committee of peers. It is his toy project to sabotage though.

@Massi

Quote:

Massi wrote: @All I think PowerPC was a sort of natural successor to M68K, when all were going RISC.

Yes, after that Motorola abandoned its 68K family, it was one of the possible good choices.
Quote:

Was it a bad choice at its time? BeOS also used PPC (before the switch to Intel).

It was the 1998 when BeOS was ported on x86, after some years with the PowerPC (and other years before with another less known architecture).

It's another signal that the things with the PowerPCs weren't going so well anymore. In fact, and as I already reported several times, even Apple was about to abandon this processor family on 2000, just before MacOS 1.0 was released. And Apple was a co-founder of the PowerPC consortium.

So, with the new millennium, choosing a PowerPC wasn't a good a very good decision anymore.
Quote:

PPC is still well supported by various Linux.

I don't know the situation with Linux, but I doubt that the support is comparable with x86/x64 or ARM.

And you have no guarantee that this will continue, since this architecture is at a dead end, as reported by other people also.


@iggy

Quote:

iggy wrote: @cdimauro Everyone seems to be on the same page except for Hyperion. AROS is already using X86, and MorphOS' eventual shift to X64 has been announced.

Exactly.
Quote:

I, because I am already using a PPC system, have no problem supporting and recommeding PPC systems because they work (for our current OS'). But as I don't see a future past the two 64 bit cores that Freescale offers, we will have to change ISAs eventually. While the X5000 is over priced, at $2100 it is a better buy than the X1000 was. And, in light of Tabor's low specs, I may just have to buy an X5000 as my last PPC based system.

Just one question: do you really need a so much expensive system? Because continuing with supporting them means that there's no interest on pushing for a direction change, and you have to continue pay a lot for an underpowered machine.

Consumers have the key to drive the market with their choices. If you think that it's time to change ISA, and you want to give a strong signal to the interested entities, the only and best thing that you can do is to keep your wallet closed.


@tlosm

Quote:

tlosm wrote: @Massi Exactly it was ... And Beos was claimed at that time the AmigaOS replacment when amiga goes to powerpc 1997-1998 ... but something was change ...

The change was the introduction of an x86 version on 1998, which enlarged the BeOS market and allowed many more people to take a look and buy this wonderful o.s..

However it wasn't enough to let it survive, and unfortunately it's acquisition by Palm was the worst thing that happened to it.
Quote:

Any way if PowerPc G5 was developed as 8 core at 3.5 ghz (like Xbox 360 ghz)...

The story cannot be made with "what if". G5 was a very hot chip, and that's the reason why Apple decided to discard it just after 2 years of having presented its G5 machines, embracing Intel.

3.5Ghz and 8 cores can be reachable today, with the newest productive processes, but it'll not change the situation: the processor will remain uncompetitive with the x86/x64 offer.


@Signal

Quote:

Signal wrote: @iggy Which is why it makes sense to first move it to like endian CPU. Then, when as much as possible is C, make the move to a different ISA.

Which isn't that much work, since most of the code was already rewritten in C.

Of course, we cannot expect that Petunia is ported as well, since it doesn't work on a little-endian architecture, as well as a 64-bit one (even PowerPC).


@iggy

Quote:

iggy wrote: But if we see new PPC cores after the e6500, I will be surprised.

AFAIK there was an e6500 successor in Freescale's roadmap, which I saw some years ago. However it was just a minor revision of the same chip, introducing a new feature (for virtualization? I don't remember now).


@matthey

Quote:

matthey wrote: The following is a list of CPU endian support. 68k - big endian PPC - biendian (more efficient and supported as BE) ARM Thumb 2 - bi-endian (more efficient and supported as LE) ARM AARCH64 (ARMv8) - bi-endian (more efficient and supported as LE) x86/x86_64 - little endian I have not given the long list of RISC processors which are worse off than PPC. There is no other popular big endian (by default) architecture. The most suitable candidate of popular processors is ARM. ARMv8 is the most like PPC and the long list of past RISC processors that failed. The performance is less than x86 but at least the power efficiency is good. There are really only 3 good options for sustainability. 1) Switch to x86 (highest performance but least compatibility) 2) Switch to ARMv8 (moderate performance with fair compatibility) 3) Design own 68k SoC processors (unknown performance with good compatibility) I like option #3 although #2 is a better option than PPC for the Amiga at this point. With option #1 you might as well select a different name than Amiga and break all compatibility.

IMO it has to be clarified what to do with the new ISA.

If the idea is to switch to a new one, recycling everything as is (same 2GB/31-bit address space limit, no memory protection, no SMP, no resource tracking, etc.) only for a question of costs, then it doesn't make sense, and it's better to stay there and let the platform slowingly dying (because this is its doom in this conditions).

However if the people wants to break the 2GB barrier and finally have a 64-bit o.s. with plenty of GBs directly and linearly addressable, then point 1 and 2 are equivalent, because for running the old software you need to put it in a sandbox, like currently AROS does with it's AmiBridge/Janus.

That's because you cannot have an hybrid system, where 64-bit and 32-bit applications can be "first class citizens", running indifferently and mutually cooperating like it happens today. The reasons are more well and detailed expressed in the article that Zylesea has already reported.
Quote:

Maybe cdimauro is too pessimistic and negative but the Amiga situation is dire and management seems clueless (the Amiga curse since C= days)

Let's say this, in short: I stopped evaluating things from an emotional point-of-view, and I prefer to follow a more rational approach, coldly evaluating pros and cons of the things.


@pavlor

Quote:

pavlor wrote: @OlafS25 Quote: do you think Trevor would invest in new PPC hardware like Tabor if there would be any discussion about a ISA change? AmigaOS port from 68k to PowerPC did take 3 years... with such speed, Trevor could release two more new boards.

AROS proven differently, with a very small team, which had to rewrite everything from scratch.


@OlafS25

Quote:

OlafS25 wrote: @matthey cdimauro is picking others

Don't confuse criticizing with picking. If you say that the 68Ks will go mainstream again, and I reply to use, that's a critic, not because I'm picking you.

Picking is the continuous, free attacks that I receive only because I'm writing something which people doesn't digest.
Quote:

and calling them liars...

Take a look at page 2: someone told me that I was a liar only because I made a statement (which was true, BTW), and without I had replied to guy previously.

When I did it, is because someone has put on my mouth words that I never stated. And that's pure lie, right? And if someone tell a lie is... a liar, by simply logic.

Especially if such guy continues to report lies, even if I asked him to quote me and prove his statements against me.
Quote:

Just because it is "online" and you do not see address and telephone number not means you can do everything. When certain people almost call everyone a liar then the person should think if the person is right or not.

If such person stops to report lies in the face of that people, I can assure you that there'll be no chance for the person to be accused of being lair.
Quote:

And as I wrote (and was attacked by a certain person for that) in my view all NG platforms will always lack the software even if the OS is modernized so 68k offers more chances expecially when new hardware is available (not to forget UAE). 68k (or classic how people here call it) is magnitudes bigger than NG, both in the software base and user base. So IF there is any chance to revive to a certain degree than it is 68k.

Nobody stated the contrary. The only thing which was reported, is the 68K software is "crystallized": it's as is, mostly binary, and with little chance of evolution.

But, well, it's a retro-platform, at the end.
Quote:

But I also accept different opinions and do not call anyone liar or stupid when saying something different.

Continuing to spread the same lies without reporting quotes to prove them will not change the situation.

"If you repeat a lie often enough, people will believe it, and you will even come to believe it yourself." - Adolf Hitler (Mein Kampf)

Now either prove your claims against me, or stop doing the victim. Take the responsibility for what YOU, not me (unless you can prove it), have written.

And it'll be a much better continuing with the technical discussion.
Quote:

Regarding Gunnar, I know that you had some problems with him. He is a engineer and sometimes technicians are a little different.

He's an hardware engineer. Other peoples here are engineers too.

I doesn't mean which one or the others regarding some technical questions. As usual, what should matters are facts and arguments.

Just for example, it was very known that he was against 64-bits (even here he has written it). After years, he presented a 68K softcore which supports them...

Being an hardware engineer doesn't mean that someone has a good vision on how to do things.

P.S. Sorry, no time to re-read and check errors.

@matthey

Quote:

matthey wrote: The following is a list of CPU endian support. 68k - big endian PPC - biendian (more efficient and supported as BE) ARM Thumb 2 - bi-endian (more efficient and supported as LE) ARM AARCH64 (ARMv8) - bi-endian (more efficient and supported as LE) x86/x86_64 - little endian I have not given the long list of RISC processors which are worse off than PPC. There is no other popular big endian (by default) architecture. The most suitable candidate of popular processors is ARM. ARMv8 is the most like PPC and the long list of past RISC processors that failed. The performance is less than x86 but at least the power efficiency is good. There are really only 3 good options for sustainability. 1) Switch to x86 (highest performance but least compatibility) 2) Switch to ARMv8 (moderate performance with fair compatibility) 3) Design own 68k SoC processors (unknown performance with good compatibility) I like option #3 although #2 is a better option than PPC for the Amiga at this point. With option #1 you might as well select a different name than Amiga and break all compatibility.

While x86 still is little endian and that is unlikely to change keeping compatibility is still possible, I've mentioned one way some years ago (address space inversion) and all modern AMD64 variants have support for the MOVBE instruction which does a big endian memory access. This means emulated 68k and "native" big endian AMD32 (64 bit mode using 32 bit pointers) would be compatible in that data structures could be shared and calls/branches between emulated and native code could be done.

Disadvantage? Most RMW and RM instructions would have to be done in a RISC style with separate loads and stores. The core of the operating system that is translated to native code could do accesses to internal data with the default little endian memory operations.

How is that worse than running on PPC?

@matthey

Quote:

1) standard hardware with an FPU (preferably a standard SIMD unit too)

True.

Quote:

2) retro compatibility with 68k CPU and Amiga custom chips

RunInUAE offers such compatibility.

Quote:

If Tabor was successful then there would be multiple production runs.

I don´t think A-Eon intends more production runs of Tabor - 1000 boards and no more.

@cdimauro

Quote:

"If you repeat a lie often enough, people will believe it, and you will even come to believe it yourself." - Adolf Hitler (Mein Kampf)

There it is!

https://en.wikipedia.org/wiki/Godwin%27s_Law

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

@pavlor

Quote:

I don´t think A-Eon intends more production runs of Tabor - 1000 boards and no more.

I think they may have trouble selling the 1000 they've already made, especially if the release is delayed far into 2016.

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

@bison

But, Godwin's law is on temporary suspend till the US presidential election in 2016 is over. At least.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC