@TRIPOS

Quote:

But the existing Amiga binaries (PPC native and 68k Petunia translated) can not run on anything but Book-E/Power ISA.

I don´t inderstand you concern about Petunia, they can simply configure it as 68020 without FPU.

@Rob

Quote:

Rob wrote: @TRIPOS How much Amiga software actually uses an FPU. If the Tabor can't satisfactorily run Amiga 68k or PPC software that requires and FPU, how big a problem really is this. I'd still prefer to adopt a wait and see attitude towards this board, and I already have more powerful Amiga hardware anyway.

Recompiling with specific support to the e500v2 solves any problem, even with overlapping instructions, but with the prices of two different binaries used.

@OlafS25

Quote:

OlafS25 wrote: @TRIPOS the OS needs to be special compiled for it?

it may be perhaps possible with recompiling the kernel for most part, which must anyway be done to support further platform. how many applications need to be recompiled is not clear at this point even we trying to evaluate it, probably it remains to be seen when os4 has been released. definitely this is a caveat when considering the purchase, as apparently os4 has a history or losing compatibility even with its previous versions and between the platforms, which results even in essential tools working on one but failing on another. partition wizzard (or whatever it was called) comes to mind.

Last edited by wawa on 17-Oct-2015 at 02:54 PM.
Last edited by wawa on 17-Oct-2015 at 02:53 PM.

@cdimauro

Quote:

with the prices of two different binaries used

Thanks for pointing that out.
Because those are the choices, don't recompile for the e500v2 and trap standard fpu instructions and emulate them.
OR recompile and wind up with two binaries, which will only work on specific processors.

Of course there is the third, even more painful option, writing code that supports either with some sort of initial test to see which is present, but that really complicates the issue.

AND the forth option, which some have suggested, not using fpu instructions at all, which further limits us.

@iggy

Once you have a working solution its is a matter of the volume of e500v2 products in the market that will make it accepted.
It could allow products like the UltimatePPC to make sense.

Quote:

iggy wrote: @cdimauro Quote: with the prices of two different binaries used Thanks for pointing that out. Because those are the choices, don't recompile for the e500v2 and trap standard fpu instructions and emulate them. OR recompile and wind up with two binaries, which will only work on specific processors. Of course there is the third, even more painful option, writing code that supports either with some sort of initial test to see which is present, but that really complicates the issue. AND the forth option, which some have suggested, not using fpu instructions at all, which further limits us.

Last edited by Spectre660 on 17-Oct-2015 at 03:03 PM.

_________________
Sam460ex : Radeon Rx550 Single slot Video Card : SIL3112 SATA card

@pavlor

Emulation of FPU instructions will be handled transparently without significant impact on performance. So no need to either recompile the OS (the kernel is obviously always specific for a certain CPU) or any applications.

_________________

@Hyperionmp

Sounds good! If the price is right, I may be tempted to buy one.

@pavlor

Thread over :D

_________________

@Alloye

Quote:

One of the problems with x86 is that it's a constantly moving target. Just coming up with a reasonable set of drivers alone will be a massive effort!

If they try to support the entire array of Intel hardware, then it's effectively impossible. But if they select a single reference board that is widely available then it's not so bad -- all they have to do is support on-board networking and sound, and on-CPU graphics.

It wouldn't support every obscure piece of hardware that someone may want to use, but it would still be a usable system for a lot of us. Give me a WIFI connection, video, sound, a web browser and a C compiler and I'm set for most things.

Quote:

Only that x86 is not the panacea for attracting the "masses" back to Amiga-esque platforms that some think it is.

Yeah, that ship sailed a long time ago. Gateway was probably the last chance. But I'm fine with having a better (faster, cheaper and widely available) system for those of us who still value the Amiga virtues. If that groups expands from 5000 to 50,000, I would consider that to be a huge success, even if the rest of the world doesn't even know we exist. There are a lot of benefit to being overlooked.

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

@TRIPOS

Quote:

TRIPOS wrote: @bison Quote: The Tabor is only interesting to me because it will run AmigaOS 4 It will? I would actually be surprised if this board would reach the market at all. It simply doesn't make sense.

I assume it will run AOS4 (and MOS?), since it has no real value as a Linux or BSD system, and there no benefit in antagonizing their customer base with a product that doesn't deliver something of value.

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

@bison
Quote:

If they try to support the entire array of Intel hardware, then it's effectively impossible. But if they select a single reference board that is widely available then it's not so bad -- all they have to do is support on-board networking and sound, and on-CPU graphics. It wouldn't support every obscure piece of hardware that someone may want to use, but it would still be a usable system for a lot of us. Give me a WIFI connection, video, sound, a web browser and a C compiler and I'm set for most things.

Trouble is... how long do x86 motherboards stay on the market before they are superceded by a new one with probably a new chip set to support?? It would be a constant struggle to write new drivers every year or more... if not also parts of the kernel too.

I've assembled quite a few x86 (all AMD, not Intel, admittedly) computers for my wife and for my home office where Windows is required to run the software we need. Usually the motherboard I choose has been replaced by the manufacturer with a new model with an updated chipset within about 4-6 months.

Paul

_________________
Builder of Frankenthousand, the monster A1000
The Young Frankenthousand A1-XE G4
X5000

@bison

And what "value" does it have for "Amiga" users, when realtime performance will be well below other offerings, even those at the cheaper end of the market ?

IF Amiga was any consideration during the design of this board than someone really need to think about a change in careers.....

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@Kronos

Well, from posted benchmarks, we know it is faster than SAM460EX. What we don´t know is end user price. There is great difference between 200 USD (p-cubed) and 1000 USD (A1-500) computers - this board will lie probably somewhere in between, I hope closer to p-cubed.

Last edited by pavlor on 17-Oct-2015 at 07:13 PM.

@pavlor

It is faster in those benchmarks with binaries compiled for the CPU.

Wether it will be faster running reallife code that might or might not include PPC-FPU instruction is to be seen.

_________________
- We don't need good ideas, we haven't run out on bad ones yet
- blame Canada

@Paul

Quote:

Trouble is... how long do x86 motherboards stay on the market before they are superceded by a new one with probably a new chip set to support??

Actually, there is a large variety of "long-life" and "long-availability" x86 / x64 mainboards that have been designed for embedded and industrial uses. These type of mainboards tend to use higher quality components, are available for 5+ years and often feature the same type of ports and slots that you would expect to see on consumer mainboards.

@ASiegel

I hope MorphOS port to new architecture (be it "x86") will show viability of such transition.

Quote:

ASiegel wrote: @Paul Quote: Trouble is... how long do x86 motherboards stay on the market before they are superceded by a new one with probably a new chip set to support?? Actually, there is a large variety of "long-life" and "long-availability" x86 / x64 mainboards that have been designed for embedded and industrial uses. These type of mainboards tend to use higher quality components, are available for 5+ years and often feature the same type of ports and slots that you would expect to see on consumer mainboards.

I assume these will come with a premium price?

Something like this I presume:

http://www.bcmcom.com/bcm_product_RX170Q.htm

Last edited by Everblue on 17-Oct-2015 at 07:55 PM.

_________________

@Paul

Yes, the hardware tends to iterate fast, but the changes are usually not that big. (The exception being when Intel or AMD come out with a new CPU family.) A hypothetical example: a new version of a motherboard may have a newer version of the sound chip used on the previous board, but it's usually not something entirely wdifferent. Sometime it is, but usually not. So the task becomes updating an existing driver, not writing a new driver from scratch.

And it's really not necessary to support every iteration of hardware -- a new reference board once a year is probably sufficient. Not many people are going to be concerned about using year-old technology. In a worse case situation someone may have to wait a few months to build a reference system if the existing supply channel dries up before a new reference board is supported. Which is a much better situation than what we have right now.

Last edited by bison on 17-Oct-2015 at 08:16 PM.

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

@Kronos

Quote:

And what "value" does it have for "Amiga" users, when realtime performance will be well below other offerings, even those at the cheaper end of the market ?

At this point neither performance or price are known, so I really can't say. But I remain cautiously optimistic.

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

@Spectre660

Quote:

It could allow products like the UltimatePPC to make sense

It would allow them to continue development with the processor they started with.

And, if it is Hyperion and A-eon's intent to trap and emulate the missing fpu instructions, that solution would allow for uniform program binaries.

So...good luck with it.
After all, it WILL work.

Its not the later 64bit core I would have preferred, OR an e500 core with a standard fpu, but it is an economical PPC with adequate performance.

My apologies for the thread title on MorphZone.
This doesn't leave quite that bad a taste in my mouth.

Last edited by iggy on 17-Oct-2015 at 08:15 PM.