AmigaONE X5000 (A-Eon Technology Ltd on Facebook)

[Kremlar]

OK, one example scenario: A-EON buys 10,000 units of off the shelf x86 motherboards without CPU at about $1m cost, which will likely be discontinued stock in 3-6 months in the PC world.

 
 $100 per motherboard at qty. 10,000???  What are you buying?  There are solid board from reputable vendors for less than $60 in qty. 1 that run circles around the X1000 board technology-wise.  I'm sure you could get them for $50 or less at qty. 10,000.

Then Hyperion takes at least 18 months to port the OS to it.

 
 They might want to get started on a base x86 port ahead of time to avoid stock sitting for 18 months.  Your scenario exists in the custom PPC world, not in the x86 world.

What will happen? Will customers want an x86 board in 2 years time branded as an "Amiga" that has long since expired its shelf life in PC world terms? Or will they accept that it? Will be be able to sell 10,000 units? These are the business risks that would have to be taken.

 
 Well, some users are apparently willing to pay $3000+ for technology 6+ years old - so why not!  But again, that scenario doesn't exist.  
 
 Buy 5 boards today, start the base x86 port.  Once the base port is done the only problem is drivers for video, audio, etc... the same problems that existing with PPC hardware!  The only difference now is you no longer need to source custom hardware any longer.  You buy an off-the-shelf x86 board, develop drivers, and that's your new AmigaOne model for 2 years.  Rinse and repeat.  

In addition to the hardware, every third party software that has been written for PowerPC AmigaOS needs to be recompiled for the new x86 ISA under AmigaOS.

 
 Most Amiga users don't care about existing PPC software.  Bring classic software over with integrated emulation and let developers port what they want from PPC.

[Kremlar]

Whatever we do, we will never be able to take advantage of economies of scale like Commodore did, it's just impossible. Comparing Commodore to A-Eon is pointless unless you also compare the meerkat that A-Eon and Commodore are in.

 
 Funny, Raspberry Pi had no market when they started and they introduced incredibly cheap hardware and how many have they sold?

[Kremlar]

The RPi has a huge established market (Linux and/or geek users) and is really just a cheap embedded chip on a board with a few sockets - orders of magnitude less complex than an X1000.

They knew that because the unit cost was going to be SO low, they could produce millions and they would sell, and recoup costs.
Design costs of RPi: Much, much lower.
Market size of RPi: Much, much higher.
Risk of producing 1,000,000 Rpi and taking advantage of economies of scale: Much, much lower.

Making 1,000,000 RPi = make a fortune. Making 1,000,000 Amiga boards of similar spec to X1000 = suicide.

The two cannot be compared.

 You're making my point.  They should be making low-cost hardware, not "high-end" outdated hardware.  If the FPGA Arcade Relay ever makes it out it will sell tons more than the X1000 or Cyrus.
 
 The Amiga market is limited, no doubt, but there's a huge untapped market of classic owners and prior Amiga users who would be tempted to buy something if it was relatively inexpensive and interesting.
 
 To me, the X1000 is neither.

[Kremlar]

If you do not choose a product based on one specific x86 motherboard, A-EON or Hyperion will need to commission a wide ranging development of drivers and firmware for the any supported x86 motherboard because no two x86 motherboards are alike.

Of course, you start by supporting a select few boards - or even one board! Like you do with the X1000.

The developer resources (and financial cost) to keep pace with current x86 motherboard hardware developments would be significant, due to the short shelf life of x86 hardware. Then you are paying to support motherboards that are discontinued within a few months of release.

You don't have to support every board.

It would be better from a practical and anti-piracy point of view to concentrate on a limited set of x86 motherboards. Apple do exactly this. However, an investment in securing these boards for the 2 years+ product life would be required.

Right. How big is the Amiga market? Say you start by picking 1 board to support and buy 2000 of them. That's $100,000 for 2000 boards. How much was spent to develop the X1000? And how many boards were produced?

Today's Linux community can support a wide range of x86 boards due to the huge pool of developers willing to write drivers and provide support for different hardware configurations. AmigaOS developer community in contrast is very much smaller.

That's why you start small.

Why do so many in the Amiga community think that it's all or nothing? Custom PPC or support EVERY x86 board? It doesn't have to be that way. Smart people don't want x86 support because they want to run AmigaOS on EVERY available motherboard - they want it so they don't have to spend $3000 on outdated hardware just to run an underdeveloped OS!

[Kremlar]

I think a "unique" feature would be running faster on a hardware than competition. All Amiga-successors are simpler than Linux/Mac/Windows, that is sometimes a disadvantage but that can be a advantage in regards of performance. Even if some devs (including members of the MorphOS team) claim otherwise, performance is still very important and defines what can be done with a computer.

Irrelevant while AmigaOS is tied to PPC.  Who cares if AmigaOS is more "efficient" when you can buy a FAR more powerful x86 that will run circles around an X1000 for sub-$500 and consume 16 watts?

[Kremlar]

Despite the long time AROS has been available since late 1990's, I do not see a plethora of AROS retailers out there and a lot of AROS machines sold. I expect the AROS team continue development from a love of the original Amiga spirit, as we all do. The big business question is: if you are confident that x86 Amiga is the commercial future of our platform, over the years why has AROS not attracted a significant market share yet? Is it because of the pace of keeping up with x86 hardware changes?

Perhaps, but tied to that it's because it doesn't have the Amiga name. AROS is also not as developed as AmigaOS or MorphOS. They also don't create systems, they create an OS.

What if someone like Trever had decided to put his money somewhere else? Pay for a portion of the AmigaOS port to x86. Choose and secure the necessary x86 boards and create a select line of "AmigaOne" x86 systems. Perhaps choose 3 boards - MiniITX, MicroATX and full ATX variations of the same board. Have 3 systems all requiring the same drivers - something small form factor, a slim tower, and a mid-tower.

Instead of AmigaKit building the same hardware over and over again you could be building a variety of systems at a variety of price ranges - all for probably lower development costs.

In the above scenario, would he have sold more systems or fewer? I bet more. Much more. And the x86 port would have been done and future systems would have been much less costly.

If a port is not feasible, then go the ARIX route - it makes sense to me. Going down the path the Amiga market is currently going does not make sense - it leads to certain and inevitable death.

[Kremlar]

It would be financially suicide to build a variety of systems for a variety of price ranges in a small market.

Besides, IMHO ppc makes sense, there is no need for amiga os to become yet another obscure x86 OS. And stepping in others footstep, means you can never pass them.

Not when the variety comes at no additional costs.  It's not like they're engineering multiple motherboards.

So, becoming an obscure x86 OS is somehow worse than being an obscure PPC OS?  Certainly not for the end users shelling out the cash for the hardware.

[Kremlar]

230 GBP (incl. VAT) - This price is for single unit purchases! And this is certainly not the lowest price I have seen.

So, your argument appears to be that because some people might perhaps object to pay 50% more for extra durable and long-term available hardware, it is better to continue to sell hardware that costs 1000% more (and is notably slower).

 
 I wouldn't even pay that.  For the extra money you'd pay for these boards I'd just buy more stock of standard boards to hold in inventory.

[Kremlar]

A port to X86 is not going to happen. We all know Hyperions position on that subject, which IMHO is good. If people want X86 then go for AROS or UAE.
New hardware is good news for most of us.

This is true, my arguments are for the community in general.  AROS does not have the Amiga name and with the community so fractured there are not enough people in any of the segments to have any traction.  It's too bad.

[Kremlar]

Yes, I'm blue, and I support A-eon.

And I am not sure why Andre would assume I'd paying that much for a rather unremarkable X64/X86 motherboard.
I hope that strategy is not what AROS or MorphOS developers consider in the future.
Its one thing to pay a premium for a board built to market the requirements of our tiny markets.
Its quite another thing to pay a premium for a board that competes in the commodity computing market.

 There's no need to pay a premium for an x86 board if you go with a common board from a reputable vendor.

[Kremlar]

I am afraid you have not thought this through. I would recommend that you talk to business owners about the immense difficulty of accurately predicting consumer demand several years in advance.

Actually I am a business owner. There's no need to predict several years in advance other than, perhaps, gambling that the x86 market will remain relevant enough for years to come and warrant a port. It will.

There are huge financial risks involved if you overestimate demand which will need to be considered when you choose a price. If you cannot sell 30% of your inventory or you need to sell for 30% below your initial investment just to get rid of all units, which would not be an unusual percentage, you would have to choose a roughly 50% higher price just to cover that.

You don't need to overestimate demand. You don't need to buy thousands of units upfront, perhaps just a hundred. That's the beauty of not custom building your own boards. You pick a reasonably priced board from top vendor that's at the beginning of it's lifecycle.

Also, financing inventory and storage cost money as well. The bigger your inventory, the higher your storage costs, etc.

Why would you need huge inventory when there are literally thousands and thousands of vendors in the x86 market doing it for you? Perhaps you keep tabs on motherboard availability - but that's it. We're talking about a market that would likely initially support the sale of HUNDREDS of units over a year or 2, probably not thousands.

In addition to all of this, there are also huge issues with warranty replacements once a product has been out of production for two years which is not the case if you deal with long-term available hardware.

You buy 5% above what you sell for warranty replacements and out of warranty repairs. That's far more reasonable a price to pay versus paying quadruple+ the price for the oddball boards you are recommending, especially because you can always resell what you don't need or use. Most top tier board manufacturers warranty their motherboards for 3+ years.

From a business perspective, going the consumer mainboard route would most likely be more expensive as well as much riskier.

I disagree. It's far less expensive and much less risky.

[Kremlar]

exactly, and since what you say is true, it remains nothing else to do as to wish the effort success and that both customers and developers will be satisfied with the outcome.

I disagree.  If you don't like something you try to change it.  Put your money where your mouth is, support the vendors you believe in, and try to take back your community.  You don't just sit and accept things - that's a poor attitude.

[Kremlar]

You might want to read again what I actually wrote. I have not been recommending any mainboards that are four times as expensive as comparable consumer mainboards.

 
 Oh.  Perhaps someone hacked your account:
 
 

Prices in the UK tend to be a bit higher than on the European mainland but you can definitely find much better pricing for long-term available mainboards.

Since I mentioned Kontron, here is an example:
http://www.dpieshop.com/kontron-ktq87mitx-industrial-miniitx-intel-core-i3-i5-i7-socket-lga1150-motherboard-p-1281.html

230 GBP (incl. VAT) - This price is for single unit purchases! And this is certainly not the lowest price I have seen.

 
 
 

If the goal is to minimize development work by supporting as few mainboards as possible for as long as possible, which is what Amikit hinted at, you can either port your OS to a long-term available mainboard or you can build up a relatively huge inventory of a particular consumer mainboard with a shorter product life. I have stated my preference regarding the former which was in no way an endorsement of any specific mainboard.

And again the former makes no sense when the price is so high.  If it was $10 more?  Sure.  Quadruple+ the price?  No.
 
 It makes much more business sense to keep some reasonably priced boards on-hand at EOL rather than pay extremely high prices for some vertical market mainboard manufactured by a relatively unknown vendor.

[Kremlar]

Kontron has been around for years. However Zotac would be preferable since they seem to have crossed over into the consumer market.

Not saying they haven't been, but relative to other vendors in the top tier consumer board market they are unknown.
 
 They are not an ideal vendor to source a motherboard from when building a consumer system for a highly cost-conscious market.

[Kremlar]

Once again, you might want to read what I wrote a fourth and fifth time.

Out of curiosity, spirantho looked up prices for Kontron mainboards in his home country and shared his findings regarding the cheapest offer he could find. I found an offer for a Kontron mainboard in his country that was 80 USD cheaper and specifically mentioned that prices in the UK tend to be higher and that I have seen lower prices elsewhere.

In what strange universe does this constitute a "recommendation" for any specific mainboard?



For the second time, Kontron were mentioned because of the unusual 7 years long product life cycle which was in response to claims that x86 mainboards have extremely short product lifes.

If you need me to repeat it a third time, please let me know and I will gladly do so.

 Are you a politician by chance?