@matthey

Quote:

matthey wrote: @agami Surely by "target numbers" you mean minimum target numbers. I expect the hope was for new production runs as long as there was demand. There was way too much effort in hardware and software development for only targeting ~8,000 computers which would still be considered a complete failure in all but the highest profit margin niches which the Amiga1 was not.

No, I mean "circa'.
A cursory market analysis in 2009/10 revealed an addressable market of about 100k users for a an "NG" product such as an A1X1k and potential subsequent SKUs. With A-EON's approach, even without MorphOS and AROS x86 in that same market, it would be extremely challenging to get any more than 10% buy-in.

The only way to get more that 10% of that market, and potentially expand into a market which includes the demographic of once-upon-a-time Amiga users, is to lower the barriers to entry.
I even outlined this alternative approach in an article I wrote on the then Amiga Round-Table web site, back when Rich still maintained the site as a means for maintaining audience attention between the increasingly rare podcast episodes.

Quote:

I expect A-Eon intended to sell tens of thousands of each Amiga1 model. Starting with a low production high end and building toward a high(er) production low end sounds about right though. High end computers require more capital reducing the number that can be built per production run making them lower production and pushing up the cost and price. It's much better to start with the low end especially considering the low end was the primary 68k Amiga market.

Expectations not based in reality are fantasies. As per what I wrote above, it is unreasonable to expect more than 10k units in total across all SKUs with their approach. Even when counting on Linux users to cover the spread.

_________________
All the way, with 68k

agami Quote:

No, I mean "circa'. A cursory market analysis in 2009/10 revealed an addressable market of about 100k users for a an "NG" product such as an A1X1k and potential subsequent SKUs. With A-EON's approach, even without MorphOS and AROS x86 in that same market, it would be extremely challenging to get any more than 10% buy-in. The only way to get more that 10% of that market, and potentially expand into a market which includes the demographic of once-upon-a-time Amiga users, is to lower the barriers to entry. I even outlined this alternative approach in an article I wrote on the then Amiga Round-Table web site, back when Rich still maintained the site as a means for maintaining audience attention between the increasingly rare podcast episodes.

Who payed for and performed the market analysis? Was there data for variations in target price, features, markets and/or brand name? Was the compatibility using 68k emulation factored in or the whole retro 68k Amiga market ignored? Was there any analysis on how much improved 68k Amiga compatibility would enlarge the market?

By 2009/2010, market analysis wasn't really needed to determine the PPC Amiga1 was a noncompetitive failure on the desktop and compatibility wasn't good enough for the much larger Amiga retro and hobby markets. By 2015/2016, the Natami was attracting huge retro interest with the "MX Bringup Thread" at 761487 views when I recorded with it going higher after. Then there was the 2022 THEA500 Mini not so cheap Amiga toy without the AmigaOS that has likely exceeded the 100k NG projection. Meanwhile, the A1222 blunder production is only in an embarrassing hundreds of units. The Amiga IP squatters just keep on going intent on blocking and destroying any Amiga success. At best, the PPC Amiga1 is seen as an expensive niche hobby computer for the classes and at worst it is seen as an Amiga abomination blocking affordable retro 68k Amiga hardware.

agami Quote:

Expectations not based in reality are fantasies. As per what I wrote above, it is unreasonable to expect more than 10k units in total across all SKUs with their approach. Even when counting on Linux users to cover the spread.

If Amiga1 was already in fantasy land in 2010, what would you call it today?

@matthey

Quote:

matthey wrote: Who payed for and performed the market analysis? Was there data for variations in target price, features, markets and/or brand name? Was the compatibility using 68k emulation factored in or the whole retro 68k Amiga market ignored? Was there any analysis on how much improved 68k Amiga compatibility would enlarge the market?

These questions would typically be answered by an in-depth market analysis. I myself performed the cursory market analysis at the time, based on a selection of key indicators. With an acceptable margin of error at ±10%.

The market research parameters were mainly informed by, and reverse engineered from the tent-pole features in the A-EON marketing material for their announced X1000.

It included:
- people preferring some level of ability to run 68k productivity software and games on the new solution.

It did not include:
- people demanding high levels of backward compatibility in running 68k software and games on the new solution, or
- people mostly/only interested in the ability to run 68k productivity software and games, or just games, on the new solution.

That last one is a very large demo (2M+), which is why RGL was able to sell over 100k units of the A500 Mini, into a market where other paid and open-source software solutions are readily available.

Quote:

By 2009/2010, market analysis wasn't really needed to determine the PPC Amiga1 was a noncompetitive failure on the desktop and compatibility wasn't good enough for the much larger Amiga retro and hobby markets. By 2015/2016, the Natami was attracting huge retro interest with the "MX Bringup Thread" at 761487 views when I recorded with it going higher after. Then there was the 2022 THEA500 Mini not so cheap Amiga toy without the AmigaOS that has likely exceeded the 100k NG projection. Meanwhile, the A1222 blunder production is only in an embarrassing hundreds of units.

While RGL hasn't released a detailed breakdown of numbers, my experience leads me to the following:

- 70-80% people who already have at least one other method for running 68k productivity software and games.
- 15-20% people who at some point in the past had at least one method for running 68k productivity software and games.
- 5-10% net new users running 68k productivity software and games.

The problem with a "toy" approach, is that ~80% of those A500 Minis are sitting idle or boxed up and stored away. While I doubt the inactive systems are spread evenly across the above breakdown, conservatively 5% net new users of the remaining 20%, is still ~1,000 net new users, of primarily Amiga 68k games and to a lesser degree of Amiga 68k productivity software.

On the other hand, I would be surprised if more that 50% of AAA's A1222+ systems were sold to:
- people who have never owned Amiga OS 4 computing hardware.

Making it maybe 100 net new users of AmigaOS 4.


Quote:

agami Quote: Expectations not based in reality are fantasies. If Amiga1 was already in fantasy land in 2010, what would you call it today?

Delusion, comes to mind. But only if those at the top still harbored any kernel of hope for "A new Amiga for a new age".

I saw a noticeable shift in enthusiasm from Trevor after the launch of the X5000/20. It stared well before COVID had us all in lock-down and before the major A1222 derailments.
I get the feeling now that there is no follow-up act for the X5000, i.e. some new non-PC big-box "Amiga", he's mostly going to make it so he can enjoy the X5000/40 for as long as possible.

All the IP drama can't do much to improve his mood, but it is a bed he had a major hand in making.

Last edited by agami on 10-Oct-2024 at 04:31 AM.
Last edited by agami on 10-Oct-2024 at 04:30 AM.

_________________
All the way, with 68k

agami Quote:

These questions would typically be answered by an in-depth market analysis. I myself performed the cursory market analysis at the time, based on a selection of key indicators. With an acceptable margin of error at ±10%. The market research parameters were mainly informed by, and reverse engineered from the tent-pole features in the A-EON marketing material for their announced X1000. It included: - people preferring some level of ability to run 68k productivity software and games on the new solution. It did not include: - people demanding high levels of backward compatibility in running 68k software and games on the new solution, or - people mostly/only interested in the ability to run 68k productivity software and games, or just games, on the new solution. That last one is a very large demo (2M+), which is why RGL was able to sell over 100k units of the A500 Mini, into a market where other paid and open-source software solutions are readily available.

Ok. So your "cursory market analysis" used fixed parameters (data inputs) and came up with one output for market size (±10%). What did you use for the Amiga1 hardware price since it is so important? Wouldn't the sales at least double at half the price?

price (USD) | SBC with PCIe
$500 Amiga1 (likely not possible without mass production)
$1000 Amiga1 (maybe possible with newer highly integrated SoCs)
$2000 Amiga1 (closer to Amiga1 micro production price)

Low end SBCs with PCIe are very competitive due to high performance x86-64 CPUs and Windows compatibility which pretty much locks up the market. This hardware is more likely to be used for low end desktop use and PC gaming than low power embedded use. An old PCIe graphics card with the standard x86-64 hardware is good enough for most games. Plug a PCIe card in most ARM, RISC-V or PPC hardware and it won't even work. PPC Amiga1 hardware with AmigaOS 4 3D support may be better off than ARM and RISC-V hardware except Nvidia cards are the most popular, drivers are not included and cost extra, there is no SMP support, Amiga1 has few games and most 68k Amiga games aren't compatible. There has been attempts to bring ARM hardware with PCIe to the market and it has largely failed. What remains has reduced specs to lower the cost. The RPi 5 with PCIe may have RPi users excited but it would be a joke if the integrated GPU was better. SiFive made a credible attempt to create low end RISC-V desktop hardware (x16 PCIe Gen 3 Expansion Slot with 8-lanes and M.2 M-Key Slot using 4-lanes for NVME 2280 SSD Module).

https://www.sifive.com/boards/hifive-unmatched-revb

https://linuxgizmos.com/updated-hifive-unmatched-sbc-showcases-new-fu740-risc-v-soc/ Quote:

The HiFive Unmatched is available for $665 pre-order, with shipments due in Q4 2020. More information may be found in SiFive’s announcement and product page, which has a pre-order button that says “contact sales.� The Freedom U-SDK may be found on this GitHub page. (Update 6/28/2024): Originally launched in 2020, this board has now dropped in price to $299.00. According to Mouser, 108 units are currently on order, with an expected restock date of July 17, 2024.

Price dropped from $665 to $299 likely because of lack of demand. Mouser has the SBCs in-stock at least.

https://www.mouser.com/ProductDetail/SiFive/HF105-001?qs=Imq1NPwxi75JBw6ulD0quQ%3D%3D&mgh=1&utm_id=17222215321&gad_source=1&gclid=EAIaIQobChMIqZnTl76EiQMV5TjUAR28gA4rEAYYAiABEgIilfD_BwE

The in-order SiFive U74 CPU cores are lower performance, lower power and lower cost than modern x86-64 cores which are practically all OoO but this doesn't help so much for the desktop where performance is more important. I expect the StarFive's VisionFive 2 SBC with integrated GPU at around $100 is a much better seller but still struggles to be competitive because of RISC-V lack of mature software. In many ways, Amiga software is more mature than RISC-V software. It is also possible to retain 68k Amiga game compatibility with more modern 3D support and I/O (the SiFive cores are licensable for the I/O). Integrating the 3D GPU is the way to go as it is better to reduced the price and compete with ARM hardware than x86-64 desktop hardware. The big Amiga advantage is 68k Amiga compatibility for a retro market which is 2 million plus strong. Real 68k CPU cores could attract 68k fans and retro 68k hardware users/fans that are tired of crap emulation on ARM and expensive CPU+chipset FPGA hardware (support small FPGA for chipsets). Half of RPi sales are for embedded use which could expand the market with appealing embedded features added. Do you think the market is large enough for mass production of a toy priced 68k Amiga SoC based SBC with integrated GPU?

agami Quote:

While RGL hasn't released a detailed breakdown of numbers, my experience leads me to the following: - 70-80% people who already have at least one other method for running 68k productivity software and games. - 15-20% people who at some point in the past had at least one method for running 68k productivity software and games. - 5-10% net new users running 68k productivity software and games. The problem with a "toy" approach, is that ~80% of those A500 Minis are sitting idle or boxed up and stored away. While I doubt the inactive systems are spread evenly across the above breakdown, conservatively 5% net new users of the remaining 20%, is still ~1,000 net new users, of primarily Amiga 68k games and to a lesser degree of Amiga 68k productivity software. On the other hand, I would be surprised if more that 50% of AAA's A1222+ systems were sold to: - people who have never owned Amiga OS 4 computing hardware. Making it maybe 100 net new users of AmigaOS 4.

Some people thought the RPi was a toy when it came out because of the unbelievably low price. It actually was pretty limited but not as limited as THEA500 Mini using emulation, no AmigaOS and no ethernet/WiFi. It is possible to make real 68k Amiga hardware cheaper, more powerful and a better value than THEA500 Mini because a 68k Amiga ASIC SoC is the ultimate integration and cost reduction possible. Sell a toy and deliver a powerful general purpose computer unlike the single purpose THEA500 Mini gaming toy. Retro opportunities to grow the Amiga user base won't be around forever.

agami Quote:

Delusion, comes to mind. But only if those at the top still harbored any kernel of hope for "A new Amiga for a new age". I saw a noticeable shift in enthusiasm from Trevor after the launch of the X5000/20. It stared well before COVID had us all in lock-down and before the major A1222 derailments. I get the feeling now that there is no follow-up act for the X5000, i.e. some new non-PC big-box "Amiga", he's mostly going to make it so he can enjoy the X5000/40 for as long as possible. All the IP drama can't do much to improve his mood, but it is a bed he had a major hand in making.

Maybe the problem is that Trevor is treated like a celebrity instead of the captain that ran into another iceberg that he ignored. It's similar to captain Irving Gould who ignored good advise and charted his own course through icebergs. He was warned early too. Did you read the Commodore Strategic Plan 1985-1987 marketing advise I posted in another thread?

https://archive.org/download/commodorestrategicplan19851987/Commodore_Strategic_Plan_1985-1987.pdf

There were visionaries who saw the icebergs and tried to plot a course around them but then there are perpetually oblivious leaders who ignore them.

@matthey

Quote:

not as limited as THEA500 Mini using emulation, no AmigaOS and no ethernet/WiFi.

It’s not a limitation, it’s freedom, it makes the THEA500 sell in a different category than if it had networking out of the box. Better and cheaper for RetroGames. Anyways, networking is only a usb dongle and some kernel modules away, not a big hinderence for anyone who really want it.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@matthey

Quote:

matthey wrote: Ok. So your "cursory market analysis" used fixed parameters (data inputs) and came up with one output for market size (±10%). What did you use for the Amiga1 hardware price since it is so important? Wouldn't the sales at least double at half the price? price (USD) | SBC with PCIe $500 Amiga1 (likely not possible without mass production) $1000 Amiga1 (maybe possible with newer highly integrated SoCs) $2000 Amiga1 (closer to Amiga1 micro production price)

For context, I did the analysis in 2011, before the X1000 was released.
Based on the available data, I would've projected with a moderate degree of confidence that sales would double at the $1,999 price tag ~20% lower).

In 2012, once the X1000 was shipping and being reviewed, despite not having AmigaOS drivers for some of the hardware, I would almost guarantee that doubling would be achieved at that price.
At some point in 2012 I asked Trevor why they didn't have a board-only option. I'm in Australia, and I'd rather not be paying high shipping and handling costs, plus unnecessary import GST (VAT) for included components I can source locally.
If they had a board-only option in 2012, and the price could be $1,600, I'd go as far to say that board-only SKUs would match or exceed sales numbers of the full system.

Trevor responded that the next machine will have a board-only option. At this point the next machine didn't have a release date nor a name. You might recall that A-EON ran a poll for the community to choose the name.

The X1000 moment was critical, and was a missed opportunity.
The marketing campaign actually worked, and it created the largest interest in a PowerPC Amiga ever. Prior and since. Having different options at more favourable prices would've moved a lot more inventory.

Were there a $1,600 board-only SKU, I know that would've tipped the scales for me. I'd probably end up using it more as a Linux machine, but that has more to do with the lackluster AmigaOS 4.1 and software library than the hardware.

By the time the X5000 was out, and even having a board-only option, the overall interest dropped. There was little-to-know marketing, and changes to the board made the X5000 look less like an upgrade, and more like a sidegrade.

Quote:

There has been attempts to bring ARM hardware with PCIe to the market and it has largely failed. What remains has reduced specs to lower the cost. The RPi 5 with PCIe may have RPi users excited but it would be a joke if the integrated GPU was better. SiFive made a credible attempt to create low end RISC-V desktop hardware (x16 PCIe Gen 3 Expansion Slot with 8-lanes and M.2 M-Key Slot using 4-lanes for NVME 2280 SSD Module).

There are some good developments in the RISC-V hardware space, but as you point out, the software has some catching-up to do. It might be a few years yet before all the elements of the RISC-V ecosystem line up to the point where it becomes a legitimate alternative to ARM in most use cases.

And I expect someone will create an ARM board larger than mini-ITX which will have multiple PCIe slots in about a year, and then it might take another 6-12 months to have a Linux distro running on it with support for a selection of AMD/Nvidia GPUs. I'm sure Pine64 would like to add a desktop ARM PC to its line-up.

Quote:

It is also possible to retain 68k Amiga game compatibility with more modern 3D support and I/O (the SiFive cores are licensable for the I/O). Integrating the 3D GPU is the way to go as it is better to reduced the price and compete with ARM hardware than x86-64 desktop hardware. The big Amiga advantage is 68k Amiga compatibility for a retro market which is 2 million plus strong. Real 68k CPU cores could attract 68k fans and retro 68k hardware users/fans that are tired of crap emulation on ARM and expensive CPU+chipset FPGA hardware (support small FPGA for chipsets). Half of RPi sales are for embedded use which could expand the market with appealing embedded features added. Do you think the market is large enough for mass production of a toy priced 68k Amiga SoC based SBC with integrated GPU?

An addressable market of 2M+ is a good enough size for a range of solutions, but by focusing on the Amiga Retro angle, only Amiga nostalgic investors would be interested. Irrespective of how good the numbers look in the business plan.

Most investors like opportunities which allow them to go back to the well and draw more water. And a new 68k-based SBC can not afford to work with an investor who has more of a "hit it and quit it" attitude.
We've discussed this previously: It's my view that the Amiga Retro angle should be positioned as the secondary market. The main value proposition has to be based on a market segment where a new 68k ASIC SoC/SiP based SBC is solving a problem marginally better than existing ARM-based solutions. That value comes more from the ecosystem surrounding the SBC, solid and interesting tools for coding and interacting. Creating these by leveraging existing open-source solutions increases delivery speed and reduces cost, and it creates a much longer tail for the new board, and its successors. It's easy to go back to that well. The Raspberry Pi Foundation has shown us all how.

A portion of the profits from the main market can then be used to help the new SBC have a running start in the Amiga Retro secondary market.

Quote:

It is possible to make real 68k Amiga hardware cheaper, more powerful and a better value than THEA500 Mini because a 68k Amiga ASIC SoC is the ultimate integration and cost reduction possible. Sell a toy and deliver a powerful general purpose computer unlike the single purpose THEA500 Mini gaming toy. Retro opportunities to grow the Amiga user base won't be around forever.

I agree.

Quote:

Maybe the problem is that Trevor is treated like a celebrity instead of the captain that ran into another iceberg that he ignored.

Definitely. Though I wouldn't say "captain", because ship captains tend to be more hands-on, and Trevor comes across more as the moneyed hands-off type that checks in every now and then to get an update on the ships path.

Quote:

It's similar to captain Irving Gould who ignored good advise and charted his own course through icebergs. He was warned early too. Did you read the Commodore Strategic Plan 1985-1987 marketing advise I posted in another thread? https://archive.org/download/commodorestrategicplan19851987/Commodore_Strategic_Plan_1985-1987.pdf There were visionaries who saw the icebergs and tried to plot a course around them but then there are perpetually oblivious leaders who ignore them.

No I didn't read this. I'd actually love to but the Internet Archive services are temporarily offline.

Last edited by agami on 11-Oct-2024 at 02:19 AM.
Last edited by agami on 11-Oct-2024 at 02:18 AM.

_________________
All the way, with 68k

@matthey

as usually you wrote bs about hardware you never have.
Amiga One around 2010 was nice home computer with everything what average home user need
like gnumeric, abiword, gimp, blender, mplayer decent web browser.
nice interesting home computer with different cpu and os.

today Amiga One is nice retro hardware that works like amiga 1200 only better because thousands times faster.

Last edited by ppcamiga1 on 11-Oct-2024 at 05:49 PM.

agami Quote:

For context, I did the analysis in 2011, before the X1000 was released. Based on the available data, I would've projected with a moderate degree of confidence that sales would double at the $1,999 price tag ~20% lower).

Ok. It roughly matches up with Jeff Porter's magic price points where he said dropping the price of the Amiga 500 by $100 about doubled the sales. Is the volume improvement based on price reductions usually linear or does it improve as the price drops, for example, into impulse buying territory?

agami Quote:

In 2012, once the X1000 was shipping and being reviewed, despite not having AmigaOS drivers for some of the hardware, I would almost guarantee that doubling would be achieved at that price. At some point in 2012 I asked Trevor why they didn't have a board-only option. I'm in Australia, and I'd rather not be paying high shipping and handling costs, plus unnecessary import GST (VAT) for included components I can source locally. If they had a board-only option in 2012, and the price could be $1,600, I'd go as far to say that board-only SKUs would match or exceed sales numbers of the full system. Trevor responded that the next machine will have a board-only option. At this point the next machine didn't have a release date nor a name. You might recall that A-EON ran a poll for the community to choose the name. The X1000 moment was critical, and was a missed opportunity. The marketing campaign actually worked, and it created the largest interest in a PowerPC Amiga ever. Prior and since. Having different options at more favourable prices would've moved a lot more inventory. Were there a $1,600 board-only SKU, I know that would've tipped the scales for me. I'd probably end up using it more as a Linux machine, but that has more to do with the lackluster AmigaOS 4.1 and software library than the hardware. By the time the X5000 was out, and even having a board-only option, the overall interest dropped. There was little-to-know marketing, and changes to the board made the X5000 look less like an upgrade, and more like a sidegrade.

The RPi later showed how important it was to sell SBCs and how much of a mistake it was to not aggressively push the hardware price down.

agami Quote:

There are some good developments in the RISC-V hardware space, but as you point out, the software has some catching-up to do. It might be a few years yet before all the elements of the RISC-V ecosystem line up to the point where it becomes a legitimate alternative to ARM in most use cases. And I expect someone will create an ARM board larger than mini-ITX which will have multiple PCIe slots in about a year, and then it might take another 6-12 months to have a Linux distro running on it with support for a selection of AMD/Nvidia GPUs. I'm sure Pine64 would like to add a desktop ARM PC to its line-up.

As far as ARM goes, the RPi Foundation may have the best chance to successfully create an ARM desktop platform with PCIe support. They have built a loyal and growing customer base and invested in their fabless semi development. ARM IP cores/blocks are likely available for PCIe and they may receive preferential licensing/royalty treatment due to the ARM Holdings relationship and partial ownership. They could license SiFive IP and use it with ARM cores to avoid royalties and provide a cost advantage over the competition. They still need to come in under low end x86-64 SBC prices and that is going to be tough judging by SiFive's RISC-V attempt although RISC-V has no market for mass production and less mature software. They may be better off just integrating a better GPU but they may have made a mistake by moving up to OoO cores that use too much of the power and cost budget. A more balanced CPU+GPU would have made a better standard platform, especially for moving into gaming and building an ARM based low end desktop and gaming market. A better standard GPU could allow a micro-console standard but they will likely be under pressure to use integrated ARM GPUs (Mali/Immortalis) rather than Nvidia, AMD, Imagination Technologies, etc.

agami Quote:

An addressable market of 2M+ is a good enough size for a range of solutions, but by focusing on the Amiga Retro angle, only Amiga nostalgic investors would be interested. Irrespective of how good the numbers look in the business plan. Most investors like opportunities which allow them to go back to the well and draw more water. And a new 68k-based SBC can not afford to work with an investor who has more of a "hit it and quit it" attitude. We've discussed this previously: It's my view that the Amiga Retro angle should be positioned as the secondary market. The main value proposition has to be based on a market segment where a new 68k ASIC SoC/SiP based SBC is solving a problem marginally better than existing ARM-based solutions. That value comes more from the ecosystem surrounding the SBC, solid and interesting tools for coding and interacting. Creating these by leveraging existing open-source solutions increases delivery speed and reduces cost, and it creates a much longer tail for the new board, and its successors. It's easy to go back to that well. The Raspberry Pi Foundation has shown us all how. A portion of the profits from the main market can then be used to help the new SBC have a running start in the Amiga Retro secondary market.

The RPi Foundation example is a good template for how to enter and build low end hardware markets. They started with 5 investors and created a business that has exceeded CBM in many ways. I agree that the 68k Amiga market may be too narrow and risky alone but the 68k market combined is promising mainly do to the hot retro market. The Sega Genesis/Mega Drive market is likely larger than the Amiga market and the potential of the NeoGeo and X68000 markets may be as high. The other 68k retro and embedded markets combined are substantial (big endian hardware replacement is easier with big endian hardware and the 68k is still very competitive in code density). An ASIC 68k SoC is the ultimate cost reduction which likely would have higher CPU core development costs than a la carte ARM IP but may avoid royalties and have a performance advantage for in-order CPU cores which are perfect for a micro-console standard. Fabless semi development costs have dropped precipitously from CBM days when the Amiga required close to $49 million to bring to market according to RJ Mical, especially after adjusting for inflation. Amiga Corporation may have been able to stay independent if development costs were as cheap as today. The difficulty for them was mass producing hardware for a new market where an existing 68k market exists today of perhaps 5-15 million? Have you ever analyzed the 68k market?

agami Quote:

No I didn't read this. I'd actually love to but the Internet Archive services are temporarily offline.

It was still down last I checked. Send me a PM with your e-mail address if you want me to send you the Commodore_Strategic_Plan_1985-1987.pdf. I think you would enjoy it considering your marketing and CBM interest.

Last edited by matthey on 12-Oct-2024 at 07:21 PM.
Last edited by matthey on 12-Oct-2024 at 07:16 PM.

@matthey

Quote:

matthey wrote: Ok. It roughly matches up with Jeff Porter's magic price points where he said dropping the price of the Amiga 500 by $100 about doubled the sales. Is the volume improvement based on price reductions usually linear or does it improve as the price drops, for example, into impulse buying territory?

As in, for every 20% decrease in price there is a 100% increase in sales volumes? No, it's rarely linear. Reductions by 20% represent an asymptote, and so the inverse should be an exponential growth trajectory. Counterintuitive as it may sound, there is a point where the next 20% drop increases sales by 500% or more. This often happens when the product breaks out of its planned demographic. When an entirely new demographic starts using it for a scenario unforeseen by the product designers.

Getting to "impulse buying" territory is a tough balancing act. It's not just about a low price.

Quote:

The difficulty for them was mass producing hardware for a new market where an existing 68k market exists today of perhaps 5-15 million? Have you ever analyzed the 68k market?

It's on the to-do list.

Quote:

It was still down last I checked. Send me a PM with your e-mail address if you want me to send you the Commodore_Strategic_Plan_1985-1987.pdf. I think you would enjoy it considering your marketing and CBM interest.

According to The Verge it should be back in a few days, if not then we go with the above plan B.

_________________
All the way, with 68k

agami Quote:

As in, for every 20% decrease in price there is a 100% increase in sales volumes? No, it's rarely linear. Reductions by 20% represent an asymptote, and so the inverse should be an exponential growth trajectory. Counterintuitive as it may sound, there is a point where the next 20% drop increases sales by 500% or more. This often happens when the product breaks out of its planned demographic. When an entirely new demographic starts using it for a scenario unforeseen by the product designers. Getting to "impulse buying" territory is a tough balancing act. It's not just about a low price.

I expected price reductions to improve sales volumes better than linearly but also expected various factors to be complex. Marketing is more of a soft science, not to demean its usefulness.

agami Quote:

It's on the to-do list.

The global 68k market would be very difficult to value. Asia is weird with Japan, South Korea and China being large and very different markets despite their proximity. I believe the retro 68k market in Japan could be much larger than the Sega Mega Drive, x68000 and Neo Geo sales numbers. The x68000 and Neo Geo were expensive and high quality products that many people couldn't afford or justify but liked or wanted. The Neo Geo was originally intended to be rental only and was practically a home arcade console. The x68000 computer was like a higher end version of the Amiga that was too expensive and large for most home users where internet "gaming" cafes are popular and further reduced the potential market. These systems came out during the 1980s-1990s golden age of Japanese games when Japan developed over 50% of global games including arcade games (Japan global game development share is less than 10% today). The games are a different style than cheaper Nintendo games but the 2D graphics hold up well and non-realistic games do well in Japan (anime, fantasy and cartoon artistic styles are popular). Sega Sammy has likely done the best job of monetizing retro IP which includes 68k IP with the Sega IP.

Sega Sammy (Sammy made slot machines and arcade cabinets before the merger)
2024 Revenue: ¥467.896 billion or $3,137,609,510 USD
2024 Net Income: ¥59.778 billion or $400,858,355 USD

Not bad for a business that abandoned the console market and likely makes a large percentage of its money on retro IP. More modern "realistic" 3D gaming is more valuable and was at least partially responsible for reducing the Japanese dominance of video games development. The South Korean developed PUBG game variants have a combined revenue of over $13 billion USD and the game has been downloaded over 1.1 billion times which is not bad considering the world population is ~8.2 billion. The big Asian markets are tricky but can be valuable. While modern "realistic" 3D gaming is popular, the originality and variability of high quality retro 2D games certainly have their appeal and retain their appeal well. Support for 3D can be added with the large retro 2D 68k game library to further increase the appeal with minimal increase in hardware cost. The added bonus is one of the easiest to program and best code densities ISAs ever. Affordable hardware could be appealing to not only older nostalgic fans but to younger generations as toys that are more than toys. The 68k Amiga was designed to be a toy with the hope that it could be integrated and cost reduced enough to become a toy. Now that it can be produced for a fraction of the cost and much more easily than the original, we ignore the opportunity. Sorry you wasted your time Jay Miner. You were better off developing pace makers.

agami Quote:

According to The Verge it should be back in a few days, if not then we go with the above plan B.

Ok.

Last edited by matthey on 13-Oct-2024 at 07:22 PM.

@agami

Quote:

I agree, but I think it ended up being more niche then originally hoped.

Overtime, that's what happened to the AmigaOne and OS4. Most people still active in the Amiga scene in the early 2000's knew about the AmigaOne, OS4 and followed the progress. A lot of the 68K people, since all of us were 68K people back then (aside expensive PPC expansions), were interested in OS4 and new hardware.

Fast forward 20 years and the market has completely changed. Since the outside computer technology has affected the Amiga market over time as well. Some 68K people disappeared into the woodwork and came back. A portion of these people didn't know about the AmigaOne or OS4 and some expressed anger about it. They just didn't like it and didn't see the point of it against their 68K Amiga. I wonder where these people went for decades? How could they not know? And then the common comparison of comparing the AmigaOne to a common PC, where it started off as reasonable enough to compare, then over time became worse. Once Amiga people were defending against the PC, then they needed to go from Intel Outside to Intel Inside, and became like defectors who now mount the charge against anything Amiga and point out how Windows and x86 was superior all along.

I think the X1000 was the last time AmigaOne hardware was reasonable enough in price/power. But I also bought mine as a beta board, so bought only the board to keep costs down, at a slight discount. It was powerful enough to replace an A1/G4@800Mhz, with a 1.8Ghz PPC64, but not much else. Mind you, PPC had dropped out of the computer market, so there was no Mac to compare it with any more and that left AMD64 where it could never meet for price/power. At the same time, by comparison, I don't think any later boards produced are as reasonably priced as the A1/XE. The later Sam boards, while featuring DDR and later PCIe, were just not more powerful enough. And the embedded CPU was worse than a G4. So, I say comparatively, later AmigaOne boards like the X5000 and Sam series, cannot even compare to an XE on power or price. To me the XE is still best for what it offered. Nothing else has scaled the power up enough to warrant the price.

So OS4 is in a real niche now. But this was carving from the start. 20 years ago I watched people get used to these AmigaOne machines and they struggled. Back then I wrote an article pointing out what criticisms I could find. Things like UBoot were too foreign. The Amiga had Kickstart, not firmware. UBoot looked like a primitive PC BIOS. The OS4 install needed to be held by hand and by default used the wrong filesystem settings. I'm loath to say it wasn't idiot proof, it simply wasn't intelligent enough! Other Amigaisms on the Workbench were left as is, like programs being named after developer terms, instead of known user terms. There's things that should have been made easier, that require too much manual intervention, but were left as is. Even by the end of the 90's, using AmigaOS was too technical. All that was dragged over and wasn't fixed. I won't go into detail, but with OS3.2 and OS4.1 combined, AmigaOS is just too hard to use!

Last edited by Hypex on 26-Oct-2024 at 02:28 AM.

@Hypex

Quote:

Hypex wrote: @agami So, I say comparatively, later AmigaOne boards like the X5000 and Sam series, cannot even compare to an XE on power or price. To me the XE is still best for what it offered. Nothing else has scaled the power up enough to warrant the price. Even by the end of the 90's, using AmigaOS was too technical. All that was dragged over and wasn't fixed. I won't go into detail, but with OS3.2 and OS4.1 combined, AmigaOS is just too hard to use!

Context is everything.

I agree that humans often determine value by comparing to other things. What we know is that Amiga OS didn't change enough over the decades, which then by comparison to the broader ecosystem results in Amiga OS going from "user friendly" to "is just too hard to use".

All because a few elitist laggards "do not want Amiga to become a PC".

_________________
All the way, with 68k

@matthey

Quote:

The in-order SiFive U74 CPU cores are lower performance, lower power and lower cost than modern x86-64 cores which are practically all OoO but this doesn't help so much for the desktop where performance is more important. I expect the StarFive's VisionFive 2 SBC with integrated GPU at around $100 is a much better seller but still struggles to be competitive because of RISC-V lack of mature software. In many ways, Amiga software is more mature than RISC-V software. It is also possible to retain 68k Amiga game compatibility with more modern 3D support and I/O (the SiFive cores are licensable for the I/O). Integrating the 3D GPU is the way to go as it is better to reduced the price and compete with ARM hardware than x86-64 desktop hardware. The big Amiga advantage is 68k Amiga compatibility for a retro market which is 2 million plus strong. Real 68k CPU cores could attract 68k fans and retro 68k hardware users/fans that are tired of crap emulation on ARM and expensive CPU+chipset FPGA hardware (support small FPGA for chipsets). Half of RPi sales are for embedded use which could expand the market with appealing embedded features added. Do you think the market is large enough for mass production of a toy priced 68k Amiga SoC based SBC with integrated GPU?

AMD64 (X86-64 v1) ISA has been out of US patent since 2023, making RISC-V pointless.

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

@agami

Quote:

agami wrote: @BigD Quote: BigD wrote: ... the AmigaOne project was always a niche thing. I agree, but I think it ended up being more niche then originally hoped. I'm sure the planned projections went a little something like this: 1. Recapture the Amiga mantle and high-ground with a new series of AmigaOne computers based on PowerPC running AmigaOS 4, plus a bit of "je ne sais quoi" in Xena/Xorro, starting with retail system aimed at re-igniting software development, i.e. X1000 (target numbers = ~1,000 units) 2. Once the beachhead has been established with step 1, create an even better high margin SKU, aimed at those who were on the fence about the X1000, providing CPU options and a board-only option, and MorphOS compatibility, i.e. X5000 (target numbers ~2,000) 3. Now that most of the people with deeper pockets have helped offset the initial R&D costs, we expand with a low cost option to create a critical mass of users, i.e. the extremely poorly named A1222 (target numbers 5,000+) If they delivered on a more reasonable time-scale, and if the A1222 were priced more like an Apollo V4 Standalone, they would've gotten pretty close to those numbers.

A1222's P1022 CPU wasn't a standard PowerPC e.g. non-standard FPU.


"P1022 device is "Not recommended for new designs", please use the replacement families Power Architecture (T1013, T1023), Arm Architecture (LS1012A, LS1023A)."
- NXP

_________________
Amiga 1200 (rev 1D1, KS 3.2, PiStorm32/RPi CM4/Emu68)
Amiga 500 (rev 6A, ECS, KS 3.2, PiStorm/RPi 4B/Emu68)
Ryzen 9 7950X, DDR5-6000 64 GB RAM, GeForce RTX 4080 16 GB

Hammer Quote:

AMD64 (X86-64 v1) ISA has been out of US patent since 2023, making RISC-V pointless.

RISC-V can scale much lower than AMD64 which the SiFive U74 in-order core design demonstrates. The early in-order Atom Bonnell design tried to scale back all the way to 32-bit x86 to avoid baggage.

https://en.wikichip.org/wiki/Bonnell Quote:

Overview Bonnell's architecture shares very little in common with other Intel designs. To achieve the strict ultra-low power objects, Bonnell features a very slimmed down design discarding many high-performance techniques used by Intel's high-performance architectures such as aggressive speculative execution, out-of-order execution, and µop transformation. Part of the design requirement was that Bonnell retain full x86 compatibility, up to the latest extension - at one tenth of the power consumption of the Pentium M. This meant any software is now 100% compatible but it forced engineers to deal with all the baggage the architecture brought along. The decision to offer full compatibility brought its own set of benefits such as access to the largest software code base in the world, including the ability to run any other x86 operating system unmodified. At the same time it forced the design team to resort to other means of reducing power. Up to Bonnell, all of Intel's existing architectures put very low priority on power efficiency (note that this has significantly changed since the introduction of Sandy Bridge). High-performance, high-throughput, complex designs are simply inadequate for the kind of power goals required out of Bonnell, even if they were trimmed down. It was decided that Bonnell would be designed from the scratch with power goals in mind. For those reasons Bonnell resembles the P5 microarchitecture.

No x86(-64) in-order core design has ever been as good as the in-order 68060 design. An additional problem was that software had already moved on to the baggage ridden AMD64 ISA. Bonnell power reductions were lackluster enough that it was not worthwhile to scale so low considering the performance lost and how important x86(-64) compatibility is so Intel went back to an OoO design. The failure was not for lack of trying as Bonnell is a very sophisticated core with many high tech power reduction techniques.

The in-order SiFive U74 core design is relatively crude and simple in comparison to the Bonnell design. The U74 core resembles a CISC design which eliminates most load-to-use stalls but uses much less hardware. There is only one AGU (single load/store per cycle) which is weaker than the old 68060 and less symmetric dual execution pipelines to reduce hardware requirements. The U74 design has common 32kiB L1 I+D but large 2MiB L2 where performance would be improved with an intermediate cache between such a large differential in cache sizes. RISC-V is a relatively weak performance RISC ISA and can't execute the equivalent of 2 RISC instructions from each execution pipeline like a CISC ISA. With all the simplicity and handicaps, the U74 core performance is surprisingly good.

https://en.wikichip.org/wiki/sifive/microarchitectures/7_series Quote:

0-cycle load-to-use latency (down from 1 cycle)

Why do high performance in-order core designs matter when there are higher performance OoO core designs?



The first set of benchmarks in the pic show the SiFive U74 core outperforming older and very popular in-order ARM cores using more traditional RISC designs with load-to-use stalls. The Cortex-A53 and Cortex-A55 are more sophisticated and likely larger than a U74 core. More interesting is that the in-order U74x4 cores match the performance of OoO Cortex-A57x4 cores further below.

year | CPU | transistors
1975 6502 3,500
1979 68000 68,000
1984 68020 190,000
1985 ARM1 25,000
1985 80386 275,000
1986 ARM2 30,000
1987 68030 273,000
1990 68040 1,170,000
1993 Pentium 3,100,000 superscalar in-order 2-way
1994 68060 2,530,000 superscalar in-order 2-way
1994 ARM7 250,000
1995 PentiumPro 5,500,000 OoO uop
2002 ARM11 7,500,000
2008 Nehalem 731,000,000 (1st gen Core i7, 4 cores 8 threads) 64 bit OoO uop
2008 Bonnell 47,212,207 (1st gen Atom: Silverthorne, 1 core 2 threads) superscalar in-order 2-way
2011 Cortex-A7 10,000,000 superscalar in-order 2-way
2012 Cortex-A53 12,500,000 64-bit superscalar in-order 2-way
2012 Cortex-A57 75,000,000 64-bit OoO 3-way big.LITTLE companion of Cortex-A53

A 2-way in-order SiFive U74 core has similar performance to a 3-way OoO Cortex-A57 that is likely using roughly six times the number of transistors and very likely much more power. The SiFive U74 core is likely winning in power, tying in performance and winning in area compared to the Cortex-A57 leaving little reason to use the OoO core design. ARM has more modern OoO core designs like the successor Cortex-A72 used in the RPi 4 and four successors newer Cortex-A76 used in the RPi 5 but these cores grow roughly exponentially in transistors and power used. The SiFive U74 core can be significantly scaled up in performance closing in on the Cortex-A72. I don't think the U74 core with a RISC ISA will get there but it could come close enough that the large transistor/area and power advantage make it the better value. A CISC ISA like the 68k with an efficient proven in-order design like the 68060 may be able to get there. I doubt a modern and improved 68060 design would ever reach the Cortex-A76 level of performance but the high power requirements and more expensive cores are less practical for relatively small embedded, retro and hobby devices. The old in-order Cortex-A53 design remains popular because it is smaller, lower power and cheaper than more modern ARM designs. This can be seen with the RPi 3, THEA500 Mini and the A600GS despite the large 3 cycle load-to-use penalty making the Cortex-A53 a poor choice for JIT emulation due to lack of instruction scheduling.

Last edited by matthey on 27-Oct-2024 at 02:43 PM.
Last edited by matthey on 27-Oct-2024 at 02:37 PM.
Last edited by matthey on 27-Oct-2024 at 02:18 PM.

@agami

you have been told many times what you should do
if you want people switch to pc give them os that will be on win/lnx/osx level
nobody sane will waste time os amiga like os
when on exactly the same computer everything may be done
hundred times faster on win/lnx/osx level
stop trolling start working on mui clone for aros
then port it to unix

@ppcamiga1

Quote:

stop trolling start working on mui clone for aros then port it to unix

Since all this is pretty much already in place, we just need to know exactly which UNIX you want to port to. I presume AIX, since you’re such a PowerPC zealot?

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@agami

It wouldn't make more sense to provide a low spec platform cheap so that the user base could be incremented easily so this way it would actually make sense to create software and afterwards get more powerful devices? at first sight you may even have the same revenue, but on long term having more software would make it more interesting

_________________
Indigo 3D Lounge, my second home.
The Illusion of Choice | Am*ga

@pixie

Quote:

pixie wrote: @agami It wouldn't make more sense to provide a low spec platform cheap so that the user base could be incremented easily so this way it would actually make sense to create software and afterwards get more powerful devices? at first sight you may even have the same revenue, but on long term having more software would make it more interesting

Yes. That is what I argued and outlined in 2011.

As I mentioned before: While I acknowledged that my proposed plan was late for the then delayed X1000, it could still be implemented in part or whole to get more users, more developers, and more software.

More software brings in more users. That was true in 1985 and is still true today.

_________________
All the way, with 68k

@agami

the A1222 was planned to be a "cheap" entry model. We all know that it was not very successful.