Coldfire - Binary Compatible

[SamuraiCrow]

lou_dias wrote:

Yes that's what I meant.
Sort of like a JIT, but not actually executing code, just analyzing it's execution, marking offsets and translating into CF compatible version.  If one 68K instruction has to be emulated by 4 CF instructions, then all jump/branch offsets have to be moved up 3 bytes/words thereafter.

The "obsession" with 68K/CF is that if Commodore were to make an A5000, it would be running a Coldfire cpu.  People are looking for a true upgrade path along the "classic" hardware lines.

Have you seen the NatAmi board with the AGA+ chipset?

If we have to write something that translantes native binaries into another instruction set, then why not use the LLVM Bitcode format?  It has been suggested several times on AROS-Exec.org since it would work on the x86, x86_64, and PowerPC chips as well as others (assuming you don't need the JIT compiler for anything).

[Crumb]

@Hans

I don't care much if 68k binaries could be written, because 90% of amiga software won't be rewritten.

BTW Biggun demonstrated *NOTHING* because there's no AmigaOS that runs on Coldfire. He just spreaded his theories about recompiling your binaries and making them slower on 68k just to be compatible with a handful of accelerators that run at 266Mhz and that are *slower* than a pitiful and cheap Efika/Pegasos (that at least has native PPC binaries and also can run 68k code at decent speeds).

IMHO a Coldfire would simply be a waste of time because it would not be 68k compatible. Recompiling would simply be stupid because 68k->PPC and 68k->x86 JITs are already faster than any Coldfire running native code (and since 1997-98 you have native PPC software). There are two OSes for PPC that run native PPC binaries.

What if current software is ported to Coldfire? Will it run faster than native PPC software running on Efika/pegasos/A1?-->NO

E.g.: would it be able to run Quake3 natively faster than Efika/Peg/A1?

Will it be faster running legacy 68k software? -> NO

[Piru]

@lou_dias

Sort of like a JIT, but not actually executing code, just analyzing it's execution, marking offsets and translating into CF compatible version. If one 68K instruction has to be emulated by 4 CF instructions, then all jump/branch offsets have to be moved up 3 bytes/words thereafter.

Even if you saved the generated code on disk, you could no longer read it back (since everything would need to be located to exact same addresses as while saving).

It just doesn't work, sorry.

[bloodline]

-Edit- this post isn't meant to sound as aggressive in tone as it does... :-)

Crumb wrote:

Don't sprout this 20 year old rhetoric... the performance of a modern architecture is now very dependant upon implementation. The ARM Cortext A8 offers the same MIPS/MHz ratio as the 3 core IBM Xenon, and uses less power.

Well, ARM performance usually sucks compared to normal cpus, just like VIA C7 sucks compared to AMD and Intel cpus. I don't care much about tweaked benchmarks.

ARM performance sucks... Ok... you get any of the current PPC implementations to run UAE at in full A500 compatibility mode on a 1000mAh battery for 2 hours... That wasn't a problem for my old Xscale 3years ago.

My old EPIA 800MHz can run the latest WinUAE perfectly!!

The big difference between the PPC and the ARM... is the number of companies supplying compatible parts and the amount money being poured into development... i.e. the ARM has vastly more on both counts.

PPCs are cheap, specially SOC models. Just because Apple charged you a lot of money it doesn't mean PPC is expensive. Just look at the Price of Efikas.

Because SOC systems are known for their performance :roll:

A quick look at wikipedia will show you companies that have licensed PowerPC:

So what... companies buy the IP core from IBM/Freescale/Whoever... there are far more using ARM... I'm not trying to big up ARM here... it's jsut a fact... The PPC is a well proven/documented architecture with plenty of dev software. It's quite attractive... but I don't see as much of a future as the ARM.

There are plenty of lower power, simple x86 variants that one could use... produced by a range of companies... and in vastly more configurations and with much more support hardware.

And these usually suck running 68k JITs because are little endian and lack big L2 cache. These also suck compared to normal x86 chips.

Really... I think you need to read up on the new 45nm chips from intel...

They run the UAE JIT fine!

One would need a "Northbridge" with a PPC too... the PPC offers nothing from a hardware point of view over the x86 in terms of 68k compatiblity...

I don't know if you didn't read what I wrote or you didn't want to understand it...

It's possible I didn't understand, I'm very tired at the moment.

PPC offers a good performance/consumption ratio, a very good price (just look at the price of Efikas). Your claims about low consumption embedded PPCs being expensive are simply ridiculous. Even desktop cpus like 970FX had decent price.

The PPC offers a mature platform for a company that wants to put a CPU core on their ASIC... but I don't see much advantage for a company wanting to build an accelerator for an Amiga.

A x86 or ARM will be cheaper and more long term solution...

Just for your information... making a board and a "northbridge" that supports buses of 1Ghz is more complex and expensive than making a board that uses an embedded PPC.

No it isn't. Just buy a NB off the shelf, and build a PCI-Amiga CPU slot bridge... not the easiest task, but you'd have to do that for either x86 or PPC...

68k->PPC JITs are faster than x86 ones. Just compare the speed of a CRAP board like BlizzardPPC, using 60ns SIMMs, no L2 cache etc and the speed of a much more powerful x86 with twice bus clock and faster memory bus.

That is a pointless paragraph... my £30 2.5Ghz Athlon64 can run a JIT sooooo much faster than my 240Mhz BlizzPPC... so what?  

If we are going to talk cost/performance... nothing even comes close to the x86... nothing can... the shear amount of development and scale of production, can't be matched.

though the PPC does offer Big Endian data format, but that is a software issue... and the ARM offers that also.

Just software? Come on! If I was doing an A1200 accelerator that would be the most important thing. It would be retarded to create an (expensive) accelerator for A1200 that didn't have good compatibility and performance. Ever wondered why phase5 included a real 060 chip on their CSPPC boards?

The biggest flaw with the PPC boards... they should never have had a 68k on them. I guess they needed to ensure good compatibility, and didn't have enough time/money to develop a 68k emualtor that would have allowed AmigaOS to run on the PPC and still be able to match the timings properly... it's not an easy task!

If you really don't care about classic software it would be quite stupid to create an incompatible accelerator for such an ancient board.

BTW, not all ARMs are bi-endian

No way!!! And certainly not for a decent price.

An embedded PPC would run 68k JITs way better than any x86 equivalent chip, it would be faster and it would have a similar price.

Not true, especially if you add in the cost of the support hardware.

Then I guess that Efika owners bought their efikas for 99$ and Genesi lost 600$ with each board. Yeah, sure.

Of couse not... but for $150 I could buy a x86 system that would wipe the floor with the Efika in terms of performance... but the efika isn't designed for performance it has other priorities.

Check out the ARM Cortex A8... have a look at the power consumption at 1100Mhz (~2000 MIPS)

Phone me when 68k->ARM JITs are available and when MorphOS and AmigaOS4 runs on ARM.

Ok... that's true, I don't know of any 68k->ARM JIT! But I will call you as soon as I find one :-)

[Hans_]

bloodline wrote:

Well... you are saying that a chip that can be programmed to imitate another chip is different to a peice of software that allows a chip to imitate another chip... while the approach is different they are conceptually the same thing.

Only one approach is cheap and easy to fix bugs... the other requires special hardware and bug fixing is a more involved process...

It's quite difficult to explain this clearly to people who don't know exactly what an FPGA is. People implementing circuits in an FPGA use some of the same tools that chip designers use to design fully custom chips (minus the IC transistor layout tools).

An FPGA has no predefined computer architecture; it is simply a huge array of logic gates with programmable interconnections. This is great, because it means that you can design and test computer circuits without having to manufacture a new chip every time you make a change.

It's the IC equivalent of having a huge prototyping board with all the discrete chips and wires required to implement any computer circuit you wish. If you wire up a circuit on a prototyping board that adds numbers together, you just built an adder; you didn't emulate it. Likewise, I could implement the same circuit on an FPGA. The only difference is that instead of manually plugging wires into a proto-board, I upload a bitfile to the FPGA that connects up the relevant wires between logic gates in the FPGA.

Let's take this one step further. Suppose now that I design a circuit that performs the same function as the Paula chip. It has the same registers, and performs the same task. I've just designed Paula compatible hardware.* I could now wire up logic gates on my proto-board (the original Amiga prototype was built out of discrete ICs), or create an FPGA bitfile and implement it on my FPGA. If I'm completely happy with it, I could call an ASIC company and pay them to transfer my Paula compatible circuit onto an ASIC, and have them manufactured in medium quantities. Equally possible would be for me to hand my design over to an IC designer and have that person design a fully-custom chip containing my Paula implementation.

Now let's look at an emulator. It's written in software for hardware with a predefined architecture. Via software instructions that are executed serially, you emulate the behaviour of all the hardware elements in a completely different machine. Unlike Dennis' Paula implementation on an FPGA, there are no actual Paula registers, and the machine has to execute several instructions to emulate what happens simultaneously in the real thing. You can't take this code and hand it to an ASIC or IC engineer and say "hey, build this circuit" because it's not a circuit at all, it's a set of instructions for a CPU that emulate the behaviour of a hardware circuit in software.

I hope that this clarifies it a bit. Just because an FPGA is reprogrammable doesn't mean that it's running software. You're implementing hardware circuits on it.

Hans


* note: it's Paula compatible, not an actual Paula chip, because it's not gate-for-gate equivalent.

[Hans_]

Crumb wrote:
@Hans

I don't care much if 68k binaries could be written, because 90% of amiga software won't be rewritten.

Nor do I care.

BTW Biggun demonstrated *NOTHING* because there's no AmigaOS that runs on Coldfire. He just spreaded his theories about recompiling your binaries and making them slower on 68k just to be compatible with a handful of accelerators that run at 266Mhz and that are *slower* than a pitiful and cheap Efika/Pegasos (that at least has native PPC binaries and also can run 68k code at decent speeds).

I believe that Elbox has a working prototype Coldfire card (they demonstrated it somewhere IIRC), so I'd expect that it's actually been tried and tested. It's not a theory. Whether it's faster or slower than emulation on 1GHz+ machines is irrelevant really.

I'm personally not a big fan of using Coldfire as the next generation Amiga, but someone else might get a kick out of making it happen. The decision was made to go PowerPC over a decade ago. That might even change to x86 at some point.

None of this changes the fact that you could produce binaries that run on both 68K and Coldfire if you wanted to.

Hans

[bloodline]

Hans_ wrote:

bloodline wrote:

Well... you are saying that a chip that can be programmed to imitate another chip is different to a peice of software that allows a chip to imitate another chip... while the approach is different they are conceptually the same thing.

Only one approach is cheap and easy to fix bugs... the other requires special hardware and bug fixing is a more involved process...

It's quite difficult to explain this clearly to people who don't know exactly what an FPGA is. People implementing circuits in an FPGA use some of the same tools that chip designers use to design fully custom chips (minus the IC transistor layout tools).

;-)

Yes, I'm well aware of FPGAs... I think they are great... but my point is that A PC is essentially functionally exactly the same as an Amiga.

It has a Keyboard and a Mouse, for input.... Video and Audio for output... and a diskdrive for storage. THe PC and the Amgia re functioanlly identical devices, they just work differently internally. But a piece of software, the emualtor, can make one device work from an (external point of view) identical internally...

If you see what I mean...

[Crumb]

@hans

None of this changes the fact that you could produce binaries that run on both 68K and Coldfire if you wanted to.

Yes, but you have to avoid using normal instructions like MUL and "emulate" it, slowing down the execution on both cpus

[hbarcellos]

Bloodline... Finally someone expressed my feelings about FPGA in a perfect way. Thank you.

[Hans_]

bloodline wrote:
Yes, I'm well aware of FPGAs... I think they are great... but my point is that A PC is essentially functionally exactly the same as an Amiga.

It has a Keyboard and a Mouse, for input.... Video and Audio for output... and a diskdrive for storage. THe PC and the Amgia re functioanlly identical devices, they just work differently internally. But a piece of software, the emualtor, can make one device work from an (external point of view) identical internally...

If you see what I mean...

Sorry, I don't see what you mean. How does your point make an implementation of hardware in an FPGA an emulation? Whether it's an emulation or a clone depends on the internals not the external appearance.

Hans

[bloodline]

Hans_ wrote:

bloodline wrote:
Yes, I'm well aware of FPGAs... I think they are great... but my point is that A PC is essentially functionally exactly the same as an Amiga.

It has a Keyboard and a Mouse, for input.... Video and Audio for output... and a diskdrive for storage. THe PC and the Amgia re functioanlly identical devices, they just work differently internally. But a piece of software, the emualtor, can make one device work from an (external point of view) identical internally...

If you see what I mean...

Sorry, I don't see what you mean. How does your point make an implementation of hardware in an FPGA an emulation? Whether it's an emulation or a clone depends on the internals not the external appearance.

Hans

Ok... We will use your Paula example, it is a good example.

Paula is a Chip, and FPGA is a chip.

They are both devices that send and receive electrical signals.

Paula responds to electrical signals in a set way, as defined by its internal design.

The FPGA can be programmed to respond to the electrical signals in an identical way.

Both chips now, from an external point of view are identical. Even though their internal implementation maybe different.

But the Amiga is a complete computer it's a device that receives human input and responds in a set way as defined by its internal design using outputs human compatible signals. The PC is the same.

The PC can be programmed so that it can respond to the human interaction (I include software as a very abstract form of human interaction here) identically using outputs human compatible signals.

As far as the human concerned external to the devices, they are identical.

So when we want an Amiga compatible solution to a problem, we can look at the lowest level of what can be called Amiga... that is a complete device made up from smaller chips, but a complete device none the less. So it is a sensible idea to take a similar device that can be programmed to work like the Amiga, rather than go one level down and do essentially the same thing at a chip level.

All I'm sugegsting is that you should look at a problem from the most practical level... not that FPGA's are bad... but the Emualtor and the FPGA are doing the same thing but at different levels of the problem.

Sorry to be so verbose.

[hbarcellos]

Ok. in theory its not emulation, but...
It's always made through reverse engineering, to have the same behaviour of the original chip. But it's not made exactly the same. It's not the real thing.
As it's much more difficult to debug, the accuracy is probably much worst.

[Crumb]

My old EPIA 800MHz can run the latest WinUAE perfectly!!

And an Efika running MorphOS would run rings around it when running Amiga m68k code.

The PPC is a well proven/documented architecture with plenty of dev software. It's quite attractive... but I don't see as much of a future as the ARM.

The 3 biggest console producers have jumped to PPC boat so I don't agree.

Really... I think you need to read up on the new 45nm chips from intel...

I doubt they offer the same performance per watt running emulated m68k code or running PPC code (OS4 and MorphOS are attractive to Amigans, don't you know?)

The PPC offers a mature platform for a company that wants to put a CPU core on their ASIC... but I don't see much advantage for a company wanting to build an accelerator for an Amiga.

Running AmigaOS3/OS4/MOS software perhaps?

A x86 or ARM will be cheaper and more long term solution...

I doubt it.

That is a pointless paragraph... my £30 2.5Ghz Athlon64 can run a JIT sooooo much faster than my 240Mhz BlizzPPC... so what?

That is a pointless reply as you would never put a power hungry Athlon64 in an A1200 accelerator.

If we are going to talk cost/performance... nothing even comes close to the x86... nothing can... the shear amount of development and scale of production, can't be matched.

I was talking about Amiga accelerators (that need to be low-power and run m68k and PPC software) and you have started  talking about x86 vs PPC.

The biggest flaw with the PPC boards...

The biggest flaw would have been investing 800€ on an accelerator with no software. No one would have bought it.

they should never have had a 68k on them. I guess they needed to ensure good compatibility, and didn't have enough time/money to develop a 68k emualtor that would have allowed AmigaOS to run on the PPC and still be able to match the timings properly... it's not an easy task!

Running the entire OS under emulation wouldn't have been fun

Not true, especially if you add in the cost of the support hardware.

VIA cpus or AMD low power cpus aren't exactly brilliant running m68k or PPC software

Of couse not... but for $150 I could buy a x86 system that would wipe the floor with the Efika in terms of performance... but the efika isn't designed for performance it has other priorities.

But for that 150$ wouldn't be low power and run m68k and PPC software at decent speeds.

[Hans_]

@bloodline & hbarcellos

By that reasoning, an AMD CPU is an x86 emulator too. It's not identical to the original x86 chip, but it does the same thing. Multiple companies make different but compatible versions of the same thing all the time. That doesn't make them emulators.

We're fully agreed that a clone (be it in an FPGA or otherwise) is not the original. Where we differ is in the definition of emulation. I agree that the end goal is roughly the same. However, the approach is quite different. An emulator is one thing, an FPGA implementation is another.

This is why I call UAE an emulator, and the Minimig, an A500 clone. Neither are a genuine A500. To make things a little more complicated, the PIC micro on the Minimig that's connected to an SD-card slot is an Amiga floppy drive emulator, since there is no floppy drive at all, and it's making a file on the SD-card look like an Amiga floppy drive to the Minimig chipset.

BTW, has anyone thought of taking the floppy emulator on the Minimig and connecting it to their Amiga? That would mean no more searching for disks.

Hans

[bloodline]

Crumb wrote:

My old EPIA 800MHz can run the latest WinUAE perfectly!!

And an Efika running MorphOS would run rings around it when running Amiga m68k code.

Maybe it would... I don't have an Efika to test... but my EPIA is 6 years old and only cost just over £100 back then... it runs rings around my 25Mhz 040... but has full compatibility with my A1200...

The PPC is a well proven/documented architecture with plenty of dev software. It's quite attractive... but I don't see as much of a future as the ARM.

The 3 biggest console producers have jumped to PPC boat so I don't agree.

They are not going to see these CPU's and we are seeing more and more reliance on extra hardware (Stream processors, SPU, etc...)

Really... I think you need to read up on the new 45nm chips from intel...

I doubt they offer the same performance per watt running emulated m68k code or running PPC code (OS4 and MorphOS are attractive to Amigans, don't you know?)

They really will outperform any PPC Performance per watt, as for PPC code... well I don't really care about that :-)

The PPC offers a mature platform for a company that wants to put a CPU core on their ASIC... but I don't see much advantage for a company wanting to build an accelerator for an Amiga.

Running AmigaOS3/OS4/MOS software perhaps?

We don't have the resources to develop our own ASICs... let alone the HiSpeed devices that we would need.

A x86 or ARM will be cheaper and more long term solution...

I doubt it.

I don't

That is a pointless paragraph... my £30 2.5Ghz Athlon64 can run a JIT sooooo much faster than my 240Mhz BlizzPPC... so what?

That is a pointless reply as you would never put a power hungry Athlon64 in an A1200 accelerator.

But you would put a PPC 970  - G5?

If we are going to talk cost/performance... nothing even comes close to the x86... nothing can... the shear amount of development and scale of production, can't be matched.

I was talking about Amiga accelerators (that need to be low-power and run m68k and PPC software) and you have started  talking about x86 vs PPC.

Oops... I didn't mean to drag this off topic... I just think the PPC or the Coldfire are not best suited to the task...

The biggest flaw with the PPC boards...

The biggest flaw would have been investing 800€ on an accelerator with no software. No one would have bought it.

True.

they should never have had a 68k on them. I guess they needed to ensure good compatibility, and didn't have enough time/money to develop a 68k emualtor that would have allowed AmigaOS to run on the PPC and still be able to match the timings properly... it's not an easy task!

Running the entire OS under emulation wouldn't have been fun

No, exactly... I appreciate that... it's just a shame, that is all.

Not true, especially if you add in the cost of the support hardware.

VIA cpus or AMD low power cpus aren't exactly brilliant running m68k or PPC software

Of couse not... but for $150 I could buy a x86 system that would wipe the floor with the Efika in terms of performance... but the efika isn't designed for performance it has other priorities.

But for that 150$ wouldn't be low power and run m68k and PPC software at decent speeds.

But how much Amiga PPC only software, do you have that you couldn't do without? really?

[Crumb]

@Hans_

I think that comparing a reimplementation with an emulator is like claiming that modern x86 chips are just emulators because they internally work as RISC cpus and simply have x86 interpreters.