@matthey

You are familiar with the concept of Occam's Razor? The reason nobody has tried the ASIC approach is because it's just not a sensible proposition.

As for emulated being a fraction of native, that's quite subjective. I will leave it to the Emu68 author to quantify that fraction. It's not a small as you seem to think.

If you care so badly about native, just run AROS on the ARM itself.

_________________
Doing stupid things for fun...

@Matt3k

Absolutely, for the end user software is (marginally) more important than hardware. Software is after all the 'right tool' that makes half the work.
The OS is the software that connects to the user software to the hardware, and should make it easy for developers to develop and support user software.
The hardware is the ticking clock of the system. The faster the clock, the greater potential for labour-saving from user software.

Check your privilege. You are fortunate enough to have a 2GHz+ G5 PCIe Power Mac to get the most out of MorphOS and the admittedly excellent collection of user software.
That same software has recently been showing that the 1.5GHz G4 Mac mini is no longer up to the task.
For the past year, here in Australia, I have had an Ebay Saved Search with email alerts to notify me as soon as there is a Power Mac G5 2GHz+ with PCIe listed. And I have been looking for them on Facebook Marketplace. And the result has been: Zero. Nill. None.

So all that cool MorphOS user software means nothing because I can't buy the old Power Mac G5 hardware. And before you say anything, these things are heavy and large, and I am certainly not paying the exorbitant shipping and handling costs to have it delivered from the northern hemisphere.

So you should believe me when I say that I look forward to a MorphOS port to either x64 or ARM64. We need the new hardware so that we can run the MorphOS software.

_________________
All the way, with 68k

@bhabbott

Quote:

bhabbott wrote: @Hammer Quote: Hammer wrote: Curiously, Gerard Bucas did not trust CSG to produce the Gary chip on time. “MOS Technology, I would say, eventually became a millstone around Commodore’s neck,� he says. “Jeff Porter and I, but especially me, decided, ‘Listen, I'm not going to do the gate array with them. I don't believe they can meet the timeline and I don't believe they can make it in volume at the right price.’� This was not politically popular at Commodore, but Bucas felt it was the correct move. “We designed it, but the manufacturing of that, we actually outsourced to someone else,� he says. “We outsourced the chip to VLSI Technology, which was a third-party chip company.� A smart move. Gary was a consolidation of TTL logic that suited a 'standard' gate array. I'm not sure about the VLSI Tech reference though. Early A500's had a Toshiba Gary chip, which was buggy. Later Gary chips are marked MOS or CSG. Perhaps Bucas or Bagnall got mixed up here. On page 358 of "Commodore the Final Years" Bagnall says:- Quote: "On March 19, Robbins finished the the design... for Budgie (a budget version of the Bridgette chip... By May 18 1992, Robbins had all the schematics completed for the gate array chips and his motherboard. Unlike previous Amiga computers, the gate arrays would not be manufactured by Commodore. Ted Lenthe began looking for a company to manufacture the Budgie chip, offering bids to... VLSI Technology, with the latter winning the bid". Quote: Hammer wrote: Amiga's Gary chip was part of the cost reduction improvements. Yes, along with putting more stuff in Agnus. Jay Miner was skeptical, but 'fat' Agnus worked as intended.

There's sub-contracting i.e. VLSI Technology and Toshiba as a subcontractor.

Commodore - The Final Years

A2000-CR US Launch

In August 1987, following on the heels of the Amiga 500, the A2000 revision B was ready for launch in the US and a relaunch in Europe.

Haynie had successfully knocked $65 off the cost of the motherboard alone, which could translate to a savings of $200 at retail. He named his motherboard The Boss, in homage to fellow New Jersey native Bruce Springsteen, whose music fueled his late night engineering.

Unfortunately “The Boss� was never made public because the two-layer motherboard was so packed that there was not enough free real estate to stencil the name.

This time, the Gary chip would be manufactured by CSG rather than Toshiba, further adding to the savings. Although the price of the A2000-CR was initially hinted at $1495, it was increased to $2000 prior to launch.

Paul Higginbottom in sales began using the phrase, “It’s 2000 for a 2000.�


Paul Higginbottom from sales has pushed A2000-CR's price to $2000.

Last edited by Hammer on 26-Dec-2024 at 01:50 AM.
Last edited by Hammer on 26-Dec-2024 at 01:45 AM.
Last edited by Hammer on 26-Dec-2024 at 01:39 AM.
Last edited by Hammer on 26-Dec-2024 at 01:38 AM.

_________________

@Gebrochen

Quote:

Gebrochen wrote: @agami Maybe I misread your post?

You had stated that maybe Hyperion could surprise us with an AmigaOS 4 port to a new ISA this coming 2025, because they kind of surprised us with OS 3.1.x.

In my post I was trying to tell you that back-porting a handful of AmigaOS 4 systems to AmigaOS 3.x is a small task and can be done in relevant secrecy.
Whereas the porting AmigaOS 4 to another ISA is such a large and complex task that Hyperion could not accomplish it with any level of secrecy. Because keeping secrets costs money. The more people that know a secret, the harder it is to keep the secret. Also, the longer a secret is kept, the harder it is to keep.

So the only way to counteract these dynamics is with systems for controlling secrets, all of which cost money. And our infamous Hyperion does not have any money, which means there will definitely be no surprise port of AmigaOS 4 to another ISA. Not in 2025, not ever.
If a such a port is ever undertaken, it will be public knowledge.

But what we will have in 2025 is more AROS, and derivatives. Which is a good thing, and should be the preferred thing. Never mind what ppcamiga1 has to say.
More ApolloOS on Apollo 080 + SAGA + AMMX + Maggie 3D. More AmiBench on AmigaKit ARM-based 68k emulation hardware with native ARM AmiBanch modules. More AROS ABI v1 on X64. More AxRT.

So when you see those bounties go live, please support them as much as you fiscally can.

Last edited by agami on 26-Dec-2024 at 01:47 AM.

_________________
All the way, with 68k

@matthey

One wonders who advised QorIQ P1022 as a desktop platform use case.

From 2009, https://linuxdevices.org/powerpc-soc-family-gains-multi-core-members/

The QorIQ P1022 is expected to begin sampling in early 2010, with a suggested resale price of $42.41 in 10K quantities.


For example, Synology NAS x13 DS413 and DS213+ have Freescale P1022.

Quote:

@matthey, Emulated 68k performance is a fraction of native performance, made worse by different endianness

Emu68 runs in a big-endian mode.

Quote:

@matthey and load-to-use stalls that do not exist on 68k CPUs.

68060 would have implied load issues when you scale 68060 to 1 Ghz with the front end not designed for modern memory architecture.

Stop picking on ARM Cortex A53 when there superior ARM cores. Rockchip RK3566/RK3568, RK3588 has Cortex A55.

C= Amiga hardware has access to ARM Cortex A72 (RPi 4B, CM4), not just Cortex A53.

Quote:

High performance FPGA CPU performance is even worse with clock speeds limited to about 100MHz in affordable FPGAs. Power and price are improved with an ASIC. Nothing else comes close to competing with the value of an ASIC. Memory requirements of native code are a fraction of JIT compilation allowing further reductions in price. There is no question of the huge ASIC value advantage but it does need mass production volumes which are less clear for the Amiga. ASIC production has risks while there is no risk in micro production for the desktop market which is guaranteed failure.

You're not factoring in the business costs.

_________________

@Karlos

Quote:

Karlos wrote: @matthey As for emulated being a fraction of native, that's quite subjective. I will leave it to the Emu68 author to quantify that fraction. It's not a small as you seem to think.

You do realize that the bigger the fraction, the smaller the result

As for matthey's ASIC pipe dream:

That chip would need to be significantly better at a given price/TDP point over the competition to overcome the the lack of an established and well maintained eco system.
All while having to recoup money spend during development (which other alternative may have done years if not decades ago).

Or in short if there had been real demand for new 68k, someone would have done it 20+ years ago.

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

@matthey

Quote:

matthey wrote: Memory requirements of native code are a fraction of JIT compilation allowing further reductions in price.

Sure if you are considering either 32 or 64MB SKU JIT cache might be an issue

Once you moved into the 21st century a few MB used or not becomes irrelevant.

With today's use cases most of the data shuffled into a CPU is just that data with any half decent compiler being able to generate code that will live in L1 cache most of the times.

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

@Kronos

Quote:

You do realize that the bigger the fraction, the smaller the result

Erm, no.

The fraction is the whole a/b division expresion. You're thinking of the divisor, the number on the bottom of a fraction. The larger the divisor for a fixed dividend (the number on top), the smaller the fraction. A fraction that is larger than some other fraction, is, well, just numerically larger.

Matthey said "a fraction of native" meaning to imply the solution is something terrible like 1/5. But 4/5 is still a fraction. A larger one.

Last edited by Karlos on 26-Dec-2024 at 10:19 AM.
Last edited by Karlos on 26-Dec-2024 at 10:14 AM.

_________________
Doing stupid things for fun...

@michalsc

szulc stop trolling and start working on mui on aros

no matter what some dumb people here wrote
repeating what phase5 done 28 years ago is extremelly stupid

amiga like solution for x86 or arm should be just mui port to unix
everything below is not worth effort is too outdated and should be cut off
first zune mui clone should be fixed on aros
then port to unix

Last edited by ppcamiga1 on 26-Dec-2024 at 10:50 AM.

@ppcamiga1

Honey, you cannot afford my services, but if you really want to I can prepare a quotation. Beware, considering all expenses and taxes it will be about 50-60 EUR per hour.

@Gebrochen

Quote:

Haha, interesting, so basically the Vampire standalone is more powerful than the gigantic X5000, yet not even half the price to obtain one versus the x5000.

It couldn't match the clock speed though. But some aspects are similar. Both would be 64 bit CPU cores running a 32 bit OS. AFAIK there's no Linux port to the Vampire so I'm not sure why they added 64 bit. No Amiga software knows about the Apollo 68080 either so perhaps having extra features is a moot point. It can be another optional extra for dedicated software.

IIRC the fast DDR memory on the Vampire 68080 core claimed to beat PowerPC would be the built in DDR3 on the Altera Cyclone 5 FPGA. If it has DDR3 since I don't know the exact model they use. If they are comparing against a 20 year old PPC with SDRAM then yeah I'd expect it would be faster. But the CPU core has to be simulated on the FPGA so would be a bottle neck there.

Quote:

Here is a curiosity question then, would that mean they could technically make Amiga OS4 run on it after porting it to it, enabling smoother run Amiga OS4.x?

In theory yes. But since they give estimates of 250Mhz speed for 68060 code and top average PPC was 233Mhz there might not be much point. Especially since the top PPC OS4 runs on is clocked at 2.2Ghz with native DDR3.

A: If you are using hardware that claims to be the next generation Amiga, which is Amiga compatible and beyond, then I expect a few users would want to use the official AmigaOS.

B: I don't see any Amiga support for Greaseweazle. Just Windows fully supported as usual and some Python code for OSX/Linux that may or may not work. It doesn't have any floppy controller I can see on the Vampire so no way to plug real Amiga disks in currently it looks.

C: Perhaps because it comes in a silver case and looks like an A600GS. It's more Amiga than an AmigaOne but it doesn't look like it. Also, it lacks a real Amiga keyboard, and the bundle includes some PC keyboard. With no floppy support and no modern Amiga keyboard (which I don't expect since no one has reproduced one yet in proper layout) it's no wonder if more people want to plug an Apollo board into their Amiga if they have one.

D: They suit different markets. MorphOS and OS4 suit people who were interested in bringing AmigaOS into the future. Vampire suits people who likes the past with RTG and 16 bit audio for some extra sugar on top.

E: I would say the best way is old hardware. Any new implementation like the Vampire and ApolloOS is not truly retro because they are modern designs. Even if they would look old to a newcomer. Except maybe OS3.2. Without the dock, unarc or internet stack of OS3.9 I don't see that catching on.

F: From what I know the software is really a hybrid. Parts are taken from AROS. They wrote AmiBench themselves which has real nice backdrops. They did their own ARM native graphics library. Not sure how much is AmigaOS based but they would need some compatibility for games.

Quote:

I remember that, when I first entered into the Modern generation Amiga OS, I was full of Hype and hope, only to then realise many things being purchased to use with it, some things, not mentioning names of software here, did not at all run smoothly to get working.

I recall Hexen was problematic. I think most of the problem was the Sam which had an embedded CPU. Where as a G4 was a full PPC and more compatible with a 603 that Hexen was compiled for. I suppose they couldn't be bothered doing a proper OS4 port. Or even releasing a patch so it was compatible.

Last edited by Hypex on 26-Dec-2024 at 02:47 PM.

@Hypex

Amusingly so I think you mean Heretic 2???

Also I had sold that version, and only recently was glad to buy the new offering.

The New offering works fine on the SAM fyi, not sure what they reprogrammed or coded, but it works...

Another side note, I think heretic 2 the older one not the recent new iteration on sale, was a tad buggy. Another note would be changing cards, the moment someone went away from a 9250, they started noticing not everything that used to ran would run anymore....

Oh well, see what the future brings.

_________________
Courtesy of SAM440Flex & Amiga OS4.1 only
Flex is 800mhz
A1000 with Classic 520 Amiga OS3.2.1
AmiKit 12
MorphOS PowerBook G4 (which can play youtube vids)

https://blitterwolf.blogspot.com

@Hypex

Some people want faster hardware. And on faster hardware a more powerful OS makes sense and is useble. If you only want to run old software, expecially old games, you do not need anything faster 68030 with some ram. Pistorm or Apollo are only waste then.

And certainly a X5000 is more poweful than a vampire or PiStorm. But you cannot compare both, different markets and different customers

Karlos Quote:

You are familiar with the concept of Occam's Razor? The reason nobody has tried the ASIC approach is because it's just not a sensible proposition.

My opinion is that your logic is flawed. Nobody tried a semi-mass produced standalone 68k Amiga Mini (THEA500 Mini) believing the market to be too small but now there is the A600GS copycat and new releases of the 68k AmigaOS as a result. THEA500 Mini sales were hurt by the competition like the RPi, WinUAE PCs, 68060 accelerators, MiST(er), AC/Vamp etc. Most early consoles and computers could not increase performance while retaining compatibility so emulation was adequate but several 68k computers and the CD32 can increase performance to modern multi GHz performance and at least 2GiB of memory with high compatibility. THEA500 Mini using emulation was not competitive enough to grab major market share like THEC64 Mini. Retro Games Limited (RGL) may have even realized this could be a problem as they contacted Jeri Ellsworth about making a 68k Amiga SoC ASIC. RGL is a micro business, likely smaller than A-Eon, and there was/is investment killing lawsuit uncertainty. RGL was not investing in the Amiga market long term or particularly seeking to improve performance, instead targeting the cost reduction and compatibility/accuracy advantages of an ASIC. Performance may be more important than they realized as RPi hardware with higher performance cores than the Cortex-A53 are eating their lunch.

Karlos Quote:

As for emulated being a fraction of native, that's quite subjective. I will leave it to the Emu68 author to quantify that fraction. It's not a small as you seem to think.

Emulated code performance is something like 1/5 to 1/3 of native performance which is very inefficient. Jitter, power and price are significantly increased too. Emulation is avoided, if possible, everywhere in computing because it is not competitive. Trying to proliferate what is best avoided is like sailing into the wind.

Karlos Quote:

If you care so badly about native, just run AROS on the ARM itself.

The standard Amiga is a 68k Amiga and we have the software. Other AmigaOS architectures will be split between ARM, x86-64 and RISC-V with I expect a smaller divided market than the retro 68k Amiga market with software including a large library of games. I want to improve 68k Amiga hardware so there is no reason to divide the Amiga market among other architectures.

Hammer Quote:

One wonders who advised QorIQ P1022 as a desktop platform use case.

Trevor claims the A1222 SoC was recommended to him. Commodity hardware is supposed to be easier but clueless leadership was still getting it wrong just like back in the Commodore days. At least Commodore used mass production instead of trying to lower niche micro production costs to try to compete with mass produced hardware.

Hammer Quote:

68060 would have implied load issues when you scale 68060 to 1 Ghz with the front end not designed for modern memory architecture.

The ColdFire V5, which is based on the 68060, increased the instruction fetch to 8 bytes per cycle and quadrupled the L1 caches from 8kiB-I/8kiB-D to 32kiB-I/32kiB-D. A modern memory controller and at least a L2 cache would be wanted for a 68k SoC. Integrated eDRAM or a MCU with SRAM memory are alternative options. I do not see any problems as caches and memory are more efficiently used by the 68k than equivalent architectures with the same caches and memory.

Hammer Quote:

Stop picking on ARM Cortex A53 when there superior ARM cores. Rockchip RK3566/RK3568, RK3588 has Cortex A55. C= Amiga hardware has access to ARM Cortex A72 (RPi 4B, CM4), not just Cortex A53.

The in-order Cortex-A53 is still being used in new Amiga market hardware to better compete against mass produced hardware with niche lower production hardware. The A600GS and A1200NG are still using the 2012 Cortex-A53. I would not be surprised if THEA500 Maxi continues to use the Cortex-A53 as well. Think of it as a replacement for the 2008 QorIQ P1022. These cores have known performance deficiencies but are used anyway to lower the cost to better compete in price with mass produced hardware. Customers understand the price but are less likely to understand the performance deficiency. Emulation of the 68k further lowers the already deficient performance. The 68060 is much smaller than these cores, needs fewer caches, does not suffer from load-to-use stalls, etc. A 68060 SoC ASIC could cost as little as a RP2040 SoC sold for under $1 USD. Nobody wants to invest in mass production and become a fabless semiconductor developer like RPi though. The RPi Holdings stock nearly doubled in the last month reaching a 1.16B GBP market cap.

https://www.google.com/finance/quote/RPI:LON

I expect Trevor keeps feeding the A-Eon PPC AmigaNOne black hole of disappearing investment instead. Eben Upton understands the importance of mass production while Trevor is clueless. Trevor is doing more harm to the Amiga than helping it. He should get out of the way and let everything go back to the Amiga Corporation owner. Clueless people must not understand how clueless they are. The worst part is that he pretends to be an "angel investor" but is looking more like a con man, especially with his tolerance of Ben who definitely is a con man. Legitimate business men build businesses instead of stealing them.

@matthey

Quote:

Emulated code performance is something like 1/5 to 1/3 of native performance which is very inefficient. Jitter, power and price are significantly increased too

You need to show your working/data behind this claim. You would need to compare a benchmark compiled for 68K on the PiStorm, versus a native ARM compiled version. You'd also have to work within like for like conditions, which would mean the ARM native version needs to be running 32-bit Big Endian and wouldn't be allowed to make use of vector extensions.

I assume you have done all of this and have the code where it can be validated and your results behind the claim reproduced.

Where can I get it?

_________________
Doing stupid things for fun...

@Karlos
Some things are obvious. A business can be started producing easier and cheaper to produce steam engine cars but nobody is going to seriously use them. The steam engines may be vastly better than old steam engines and old steam engine users may think they are a large upgrade. They will not be competitive with modern technology though.

@michalsc

Quote:

michalsc wrote: Beware, considering all expenses and taxes it will be about 50-60 EUR per hour.

So you value your time at about half of that of an average plumber?

mkay......

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

@agami

I will disagree with you slightly my friend, I think the margin is very heavy for user software in the Amiga Land we all live it. Most software is terribly old now. I run the 68k versions of FinalWriter (although most of my word processing is done is MorphOS PageStream.) for spell checking and not much more, TurboCalc as that is the really only viable spreadsheet to use (MicroExcel for MorphOS seems to be stalled atm), CheckItOut (the only decent accounting program I have ever found in any Amiga flavor), Outline, and a few other smaller 68k stuff. Most everything is native to MorphOS. There isn't really much to be done in Amiga 68k productively these day, and we seem to have very healthy options for hardware for 68k that is very welcome.

I am thankful PCIe Power Macs are easily and cheaply had on this side of the pond :).

Lastly, I can't understand what is even remotely taxing your Mini? I do productivity on the road all the time with my powerbook and it is mostly major overkill for everything I do.

The only issues I have ever had was Iris being sluggish with email and Wayfarer only being useful for most of my browsing needs.

Iris was corrected with Caching that Jaca brilliantly added a few versions ago and now it just screams even on the powerbook.

Wayfarer is going the route of other Amiga browsers where the computing will be done elsewhere so Wayfarer will be able to be run on a small foot print.

Everything else, PageStream, VPDF, PolyOrga and others all scream on the powerbook. I probably have the biggest PolyOrga database in existance and my Pagestream documents are muliple pages with graphics and even the powerbook is instant or close to it for everything.

So for me using it day do day for work almost exclusively, I just don't see any slow down or need for more computing power. I do see a need for more developement with the applications as I always am coming up with ideas to do work better.

YMMV of course...:)

Last edited by Matt3k on 26-Dec-2024 at 09:18 PM.

@Kronos

Sure, but decent plumbers rarely need more than an hour for most jobs they charge that rate for.

_________________
Doing stupid things for fun...