@pavlor

You can change results a lot depending on versions of lame, parameters and so on

@OlafS25

Quote:

Which parameter do you use?

None parameters. As simple as possible.

@pavlor

Quote:

Someone made the "lame" test comparing it with other platforms and 68k emulated was the same as 500 Mhz PPC That was me and that was on older computer of my brother (Core 2 Q6600). Core i5-2500K can now reach performance comparable to at least G3 800 MHz.

G3 or G4? Because it makes difference in lame benchmarks and G3 is quite, ehm, lame

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

Quote:

G3 or G4? Because it makes difference in lame benchmarks and G3 is quite, ehm, lame

As I wrote - G3. There is no AltiVec in WinUAE.

My point is to show that 68k emulation on modern PCs delivers comparable performance as PowerPC CPUs - like G3 1 GHz.

@pavlor

BTW I am just integrating lots of Tracker formats f.e. Digibooster, MED, Fast Tracker II and many others

@pavlor

Quote:

My point is to show that 68k emulation on modern PCs delivers comparable performance as PowerPC CPUs - like G3 1 GHz.

Sure. If your PC was faster it could beat G5, too.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

Quote:

Sure. If your PC was faster it could beat G5, too.

Amithlon would be fast enough for such task even on my hardware. Sad it is not developed anymore.

@pavlor

Yes it is a pity propably companies wanted to sell new hardware instead of competition based on X86...

Pascal has created AMINUX based on my distribution that starts FS-UAE on a Linux distribution. It is of course not the same (half the speed of Amithlon running on same hardware) but at least it is supporting new hardware and thus will beat Amithlon sooner or later.

@OlafS25

Quote:

propably companies wanted to sell new hardware instead of competition based on X86...

There was some IP dispute between Amiga.Inc, Haage/Partner and Bernie Meyer (author of Amithlon). That is why HaP left AmigaOS4 project.

Some sort of Amithlon-like emulation for OS4 (or MorphOS) would be interesting. OS3.x is too dated for today standards.

@pavlor

that would need something similar to UAE emulating PPC. I know there was something (Opensource) but if it was good enough for MorphOS or AmigaOS is out of my scope. Or something like VMWare emulating a PPC hardware (there was a discussion recently). But then is the problem that both "commercial" teams see hardware as a kind of copy protection so I doubt they would be supportive. They would propably put something in the OS to block that use.

@OlafS25

Quote:

I know there was something (Opensource) but if it was good enough for MorphOS or AmigaOS is out of my scope.

QEMU is one candidate. MorphOS was already partialy ported to QEMU, but performance was not satisfying. QEMU can reach only 1/40 of native CPU performance (that would be still faster than Efika on Core i5-2500) and doesn´t support advanced 2D accelleration (not even talking about 3D).

@pavlor

I read about it at the time MacOS was still on PPC. But I think the project was stopped when Mac changed hardware to X86 and no need for PPC emulation anymore. But anyway it will not happen.

@OlafS25

That was PearPC.

QEMU is another emulator able to emulate many different CPU architectures (x86, ARM, MIPS and even 68k).

@pavlor

You are right... PearPC

QEMU is slow

@OlafS25

Quote:

QEMU is slow

I use it for AROS x86 and of course for some games that don´t run on Windows7 (Dungeon Keeper for Windows - much better GFX quality than DOS version from GoG ).

@pavlor

If there was someone with skills and devotion Amithlon could be recreated. Unfortunately it has almost disappeared from Amiga and the train was missed long ago, railway station burned and tracks removed.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

Intel will soon release a 15 core Xeon e7 v2 with over 4.3 billion transistors :)

_________________

@KimmoK

Sorry, hadn't missed your post, but hadn't looked at the T1022 info. I'm a little confused by the info, which states 4 PCIe controllers, which can be configured in different lane configurations. Does that mean (up to) 4 lanes, or does it mean 4 buses each with multiple lanes?

Quote:

So... for example with T1022 one could use one SATA2 port and the second SATA2/PCIe#4 port perhaps could be used as miniPCIe ... especially if there exist miniPCIe SATA card with a chip that we already have drivers for. Also.... one could bridge 3xPCI out of one of the PCIe, then put same PCI audio, sata etc stuff that exist on SAM440 and is well supported.

Can the second SATA port be reconfigured to provide a PCIe lane? I can't see any mention of that option.

Quote:

And if the DIU of T1022 is proven useless somehow, see if a chip without DIU is cheaper.

Let's assume it'll be crap.

Assuming 4 controllers means 4 lanes:
In the case of the custom FPGA AGA framebuffer thing, I guess bandwidth wouldn't exceed a single PCIe lane, so you'd still have another 3 to play with. If you're configuring the 4 lanes for a 4x slot for a graphics card, you might be a bit stuck for further expansion though.
You could even consider reducing graphics down to a 2x lane width, freeing the other two up for another 1x and a PCI bus.

NB - I wonder if that would make a noticeable difference to performance: I'm guessing 2x bandwidth isn't going to be a bottleneck on AmigaOS even with the higher end cards - since no applications currently will be demanding enough to saturate it.

10 year longevity support sounds like a damn good starting point for a long term project though.

Quote:

But adding normal PCIe and PCI slots the board grows large and the price might start to rise up towards SAM460 prices and above... And even if the price would not rise, larger boards are more dificult to use in some purposes.

Depends entirely on your target market I suppose - are you aiming for a PPC RPi, or a general purpose AmigaOS desltop board at a low cost. You can't have two for the price of one.

Last edited by Boot_WB on 16-Feb-2014 at 03:53 AM.
Last edited by Boot_WB on 16-Feb-2014 at 03:52 AM.

_________________
Troll - n., A disenfranchised former potential customer who remains interested enough to stay informed and express critical opinions.
opp., the vast majority who voted silently with their feet.

Quote:

Boot_WB wrote: Now I'm not a computer engineer, just a hobbyist in hardware terms, so your experience in almost infinitely more than mine, but to me it sounds like you're proposing more your dream design of an Amigaone rather than discussing a pragmatic approach to designing a reasonably priced board for a small market where sales can't be guaranteed to reach 500 units. Nothing wrong with that, but we may be talking of a potential market of 1 person.

Indeed, I would be designing something for myself. And I'm OK with that.

Quote:

Quote: And for me, saving a little power is important, so long as I can source parts and maintain feature set. I want performance. I also want long battery life. What can reasonably be done is somewhere in the middle of those two goals. I won't go without an SD slot to increase battery life, I want a way to power down that slot when not in use. That happens anyway when not in use. Flash is static memory, it doesn't consume energy sending a strobing refresh signal.

You're forgetting about leakage currents, which are becoming more and more important as gate oxides shrink. Removing power to certain things is how you reduce wastage here.

Quote:

ACPI, NAP, Frequency-stepping are all fairly standard things. Turning off memory controllers and cpu cores though, is that even possible without reinitializing hardware?

Of course you would have to reinitialize when you power back on. I would wager that would look a lot like the initial power-on sequence for those particular bits. If it's powered off, then nothing is being used in there to put back, you'd start using it or perhaps take some time to rebalance currently used resources. Turning a used memory controller/bank off would require moving data out first. And all of that would require OS support. My philosophy is to make hardware capable. If you do not, then software will never make use of that later on. I like to keep my options open.

Quote:

Imo this is not a philosophy which results in a minimised cost

I'm not making the absolute minimal cost as my main priority. An as modern of a PC laptop feature set that I can get, with a PPC CPU, is not going to be cheap, and I'm not trying to make it cheap. Otherwise I'd put a 5200 on some tiny PCB and be done. But that's not what I want, and I'm not trying to be a sane businessman, so I won't. I'm not looking to get rich with my idea. I work on things when time and money allow. If they do not, then I end up with a 10 year thought problem. I'll spend my time and my money as I please.

Quote:

So you wouldn't allow your design to be limited by support for pre-existing peripherals, in terms of buses/standards used, simplifying design choices... but you would limit your design's layout around the existing physical holes in a re-used laptop shell? That's crazy talk!

I don't understand this last bit at all. I am of course use pre-existing buses and standards. What new things have I proposed to invent anywhere? I'm not going to invent a new SD format just to be different. I'm not going to insist we only go with with SRIO additions to the CPU, even though that is a standard.

I would use USB, PCIe, ethernet, SATA, etc. Those are all very standard things. But a t4240 laptop with no display is not very interesting, so a gfx component needs added to that somehow. MXM3 is quite a standard, and saves me from designing a graphics card into the motherboard. Hurray! You should be pleased... If you have several USB links, why not put a connector in a few laptop shell USB holes, and then connect the rest to USB things in the laptop shell? If you have a sound chip, and you should, why not put a headphone jack in the laptop headphone hole? If there is an SD hole, why not ponder how to accept an SD card into it?

I expect to have more connections than I typically see laptops having holes... The extras will go unused, like all of those extraneous ethernet ports on a t4240, any extra SATA, USB, PCIe, etc...

In attempting to reuse the host laptop's CPU heatsink, then the CPU should go in approximately the same place. Same for HD connector, CD connector, and whatever else there is.

If I would go with a laptop shell that does not have a VGA hole, then I'm not going to dremel one in. So yes, I am limited by what the best shell provides, whatever that would be. I have a list of requirements to shop around for, some that might be nice but not required, and some I'd prefer to avoid... One should not reach for the first PC laptop you come across and stop your search there.

Now, would you want a laptop that has no sound because it's not built into the CPU? Would you want a laptop that has a webcam from it's PC origins, but is not hooked up? Would you want to run it off wall outlet power only, because the CPU doesn't come with a built-in battery charging circuit? Going too minimal makes for an unlikable product. And that's not OK either.

Finally, understand that I see "my laptop" as my only computer. Having a desktop or a tower chained to a desk in some "computer room" somewhere doesn't give me much opportunity to use it these days. For several years it's been like that. If I want to use an Amiga, then I need a laptop, not a desktop, to accomplish that with. What's an Amiga to me without any sound? Without graphics? Without a hard drive? Without networking today without WIFI in the CPU? Without whatever else that Freescale insists on leaving out of their CPUs? It doesn't make sense. I'm not going to let Freescale or AMCC tell me what my design can or cannot do. If I believe it needs supplemented, then I'll look for a supplement.

And that's what I enjoy thinking about.

Last edited by billt on 16-Feb-2014 at 04:47 AM.

_________________
All glory to the Hypnotoad!

@billt

Quote:

billt wrote: [...] And that's what I enjoy thinking about.

No problem Bill, and wasn't meaning to be overly critical, I was just challenged to understand your design philosophy. Generally (aside from Human Factors) one would design the shell around the resultant motherboard, such that routing wouldn't be limited to pre-existing locations. It seemed an odd choice to start with the simplest bit (the case) as the template and design the most complicated bit (the motherboard) around that.

My last comment had been preying on my mind somewhat, and I'd almost edited it back to "seems like crazy talk to me."

Re: buses at al - sorry, for some reason I'd inferred that you were planning an expansion card using CAN bus or something - my misunderstanding, got lost in acronym land.

Re: SD card slot leakage current - what sort of magnitude are you talking here? Can this actually be a noticable saving, or is it hysteresis, or losses through decoupling capacitors (ie physically measurable, but in practise only likely to consume/save a second or two from a four-hour battery life).
What level of power loss can actually be saved here? Is it proportionate to prioritise this as a power-saving goal, or is the additional complication of control circuitry actually likely to increase power loss overall?

Re: integrated expansions - they are the only thing that makes sense in a laptop, I agree. You don't want to waste power on a discrete chip when you've got the peripheral already built into the SoC. For a desktop, focusing once again on cost, it's likely to be less of a price-driver than man-hours spent writing a new driver for an esoteric SoC peripheral, or revising a board design (more complexity makes this more likely).

Re: Reinitialisation of hardware without rebooting, adding/removing memory during runtime: That sounds like a very large can of worms - I wouldn't even know what was involved, how much is handled by firmware/BIOS, etc.

Re: existing laptop shell: One question: would it not be cheaper, and free you from arbitrary design limitations (and compromise) to just have a laptop shell made? Pressed aluminium, or injection moulded plastic isn't that expensive compared to revising prototype boards (if it saves you a single revision it'll be cheaper).
You could even just replace the base of an existing shell to free up your design whilst taking advantage of pre-manufactured subassemblies (eg screen panel, hinges, etc) - although as an computer/electrical designer, that probably sounds incredibly dull to you.

All the best Bill, I am interested to know where you go with it: with the Amiga I was more of a hardware tinkerer than software, so I am always quite interested in other people's hardware projects.

_________________
Troll - n., A disenfranchised former potential customer who remains interested enough to stay informed and express critical opinions.
opp., the vast majority who voted silently with their feet.