Just looking around
Posts: 15 from 2016/1/22
> Many project went to the trash, because of the lack of interest/toughness anymore.
> Not mention the time, what the development needs. Colud take years to have a single mainboard...
> I bet, it will be outdated, when it appears.
If we try, it may or may not work, if we don't try, it is garanteed to fail.
> 2: There should be at least two version of this machine: a portable (notebook) and a desktop.
If you read, the proposal, I am suggesting switch to another market as a tactic to reconqueer the market later.
As I have said times and times again, the MIPS suggestion **only** makes sense in the context of the previous proposal. If MorphOS is to remain in its current
> First find an investor with a plan. Then you might have a case.
I will see what I can do . I was suggesting that the team branch into hardware delvelopment in addition to software development and I wonder why they wouldn't do so, however, if the team really refuses to do so, I will try to see if I can setup something on the side for hardware.
>> the market needs a server processor which is power efficient
> Where is it? Which current MIPS core is suited for that?
The current Octeon includes 1 to 48 cores of the type i6400 (they only sell them as network server processors but they could be used for other types of servers). Now that Imagination has a 64 bit p series processor (something which was unavailable) it is only a matter of time before they make processors of with 1 to 48 cores or more of the type p6600 this will be the beginning of the return of the MIPS in the server.
> Due to its single CPU core, the SoC wouldn't be popular with anything other than single-core
> operating systems such as Amiga-like OS. I doubt IP licensors would go without fixed licensing
> fee in this case.
Most SoC are only used by a single operating system, I do not see where the difference is. Moreover, when everything is hardware accelerated, there is much less uses for several cores. Finally, as a last ressort measure, which would increase the cost per unit even more but allow a fee per unit from licensor (to lower the upfront cost), it is always possible to include 2 or more cores in the chip and destroy the extra ones before including the chips on the board for the first production run. Then once there is some money from the sale of the products, the second revision can be done truly properly.
>> For the cost of the hardware engineers, the key is to have only 3 engineers to
>> keep costs reasonnable. This cost can be brought even lower if we find some who
>> are former Amiga users and do it as a labour of love [...].
> The chip you described (multi-GHz, GPU, hardware overlay, SATA, USB3, GbE, IP hardware
> offloading, Wi-Fi, Bluetooth, NOR flash, DCT/IDCT/FFT/IFFT, layer-4 protocol checksum
> offloading, IR decoding, GDDR3-SGRAM and RLDRAM3 controllers) would be a very complex
> SoC. Three or less engineers developing this SoC in their spare time would need something
> like a decade to get it ready for the market.
Well, the work they would need to do would be integration, all the mentionned units should be licenced, not created from scratch, preexisting solutions are available. Only the hardware overlay needs to be adapted to our specific case, even the, it should only be adapted, not created.
> In conclusion, I still think you're a dreamer.
All worthwile projects started as a dream.
> Good luck convincing Apple to make this wonder-SOC you want. ;)
If you read carefully, Apple has NOT bought Imagination Technologies.
> BTW Spell checking is usually a good idea.
Sorry about that, I should have re-read the file for grammar and spelling an extra time or two.
> You use an argument about ARM server chips as a reason to use
> MIPS in a convergence box. These are unrelated markets so this
> argument doesn't make sense.
Maybe I wasn't clear but what I want to say with the ARM server question is that there **will** be a third architecture alongside the ARM/x86 duo. The processor market will not remain x86/ARM only. The fact that the MIPS platform is growing, being used in cheap smartphones and tablets, makes it a likely candidate. It will allow what people had hoped about the ARM, to use it in both mobile devices as well as, with many more cores, in servers, being cheap and power efficient, precisely what companies wanted to do with the ARM but which didn't supply the hoped power efficiency.
> You also have quite obviously not done your market research.
I disagree, see the next point.
> You can already buy boxes that already do most of the same
> functions for very low prices.
It only does a few of the functions at the time. None of the current devices, for example, allows to play games on par with a current generation tablet (except game consoles which have their own serious limitations). You only can have from 2 to 4 functions from my list at the time.
> you can't get a box that does *all* the functions, but thats only a matter of time.
You are literally proving my point, this is where the market is headed. While the competition will add one function at the time, we can provide a full convergence box with all functions at launch.
> OS wise, Android pretty much has this market sewn up.
I wonder what makes you say this, the most popular device is the Roku, followed by the AppleTV.
> The open source version is free, good luck competing with that.
If you read my file again you will see that I have a way to stand against that, even though I wonder if GoogleTV/AndroidTV would be real competition as the platform is considered bad (contrarely to the smartphone and tablet version, the TV version is considered bad). Please read again the first paragraph of section 4d.
> There's no need for a special SoC. There's a whole row of
> Chinese chip makers who will happily sell you an incredibly
> cheap 64-bit multi-core chip that does everything required.
Well, if we cannot put up an adequate team to make a SoC, licencing an existing SoC as a starting point can be a sort of solution; as an absolute last resort, using an existing SoC, destroying the unnecessary sub-units and adding the missing functionnality as external circuits can be workable.
> The whole argument for Fast-Mem made sense in the days when
> the display took a significant amount of memory bandwidth and
> the CPU had little or no cache. Those days have long gone.
A modern GPU has an extremely high bandwith requirement, the GDDR memory wasn't invented for nothing. Using it as a GPGPU/DSP only worsens the problem.
> RLDRAM might be useful for something really high end (and
> expensive) but that's not going to be a convergence box.
The CPU only needs a small bandwidth but is sensitive to latency. Think of the RLDRAM3 Fastmem as an external cache.
> The proposed box also includes interfaces that were obsolete
> years ago - PCMCIA ???
It is still used quite often, for example, CAM modules use PCMCIA.
> As for the building processor, expecting people to work on it
> for free is not only ridiculous, it's illegal.
The same thing goes for software, yet people in the Amiga world do it as a labour of love and only get paid a little bit here and there, nowhere near the true price.
> You need a big team of engineers
I would go with only 3 engineers.
> Then there's all the hardware and *extremely* expensive EDA software
Or you can try to get away with some basic VHDL tools.
> you need and a boat load of multi-million dollar licenses.
Or you can negociate an cost per unit produced to reduce upfront costs (but it increases total cost).
> Oh, and you'll need a supercomputer to simulate it.
Or you can bake some test chips as an alternative.
When you try something, you may or may not succeed, when you don't try it, you're garanteed not to succeed.
The worse which can happen is to fail which will be no worse than not having tried. Keep this in mind.