I really wonder if it makes sense to make an ASIC.
In the V600 FPGA we see an average performance of ~ 200 MHz 68020
Our expectations for the next card is a performance about 3 times higher
e.g ~ 600 MHz 68020
And the card will still be very affortable compared to 68060 cards.
Now if we would spend money on a fast FPGA - like people spend money on 68060 cards.
Then we could again double the value and reach a performance on a ballpark of over a Gigaherz 68020.
The Phoenix core offers excellent performance/price for Amiga hardware. This can help reduce the 68k market deterioration and probably even bring back some users who recently left or have NG Amigas. The performance/price of the Phoenix CPU compared to any hard CPU is only going to attract geeks who realize how awesome a customizable CPU is. An FPGA is too expensive to make it cheap enough for mass produced hardware like the Raspberry Pi and never high enough performance for the affordable high end CPU market and thus remains niche. I think there is still a smaller retro and embedded market as an FPGA if the CPU is debugged to maturity, development and OS software is improved and standards are adopted with good documentation. It would probably help proliferation to either create a business or open source the core. The current Amiga and Hyperion situation is bad for anyone wanting to do Amiga related business though. It might be worthwhile to at least look into obtaining ownership rights to the AmigaOS if the situation became investable (AROS is of course the other option). Some level of investment would probably be required to move the Amiga out of it's niche into something interesting and marketable to the rest of the world. I believe an ASIC would be a necessary part of that equation for a 68k Amiga but it would require the sacrifice and cooperation of the Amiga community which has never been seen.
Do we really need an ASIC?
No, but did necessity keep Apple from becoming the most valuable business in the world? An ASIC is not necessary to have fun with Phoenix but if you are going to the trouble of making a fast FPGA CPU then you have already done most of the work for a faster ASIC CPU. It makes me wonder, if you can do that with an FPGA, what could an ASIC do? It's like sending an unmanned rocket to the moon and then deciding we are done because we proved our point and that is good enough.
Most of us here have no interest in spending the heaps of dollars to get an ASIC developed.
There is such a lack of modern software, you would not get new customers.
Maybe at a later date when there are more potential buyers it might be nice to do.
It's the chicken and the egg syndrome. There is not enough new Amiga software because there is not enough affordable Amiga hardware and there is not enough affordable Amiga hardware because there is not enough new Amiga software. Without interfering in this tailspin, the Amiga will crash and burn at some point. IMO, the hardware situation is better to attack because it expands the number of users who can buy and develop new software. Trevor@A-Eon is working on the software approach by buying old software and using a web store which is a good idea to maximize sales to existing Amiga owners and should help sell some NG Amigas by increasing demand but I wonder if the gains are bottlenecked by the high cost of the NG Amiga hardware. Also, we may be in danger of running out of good developers as they are assigned to update his software products
.
If the amiga on a chip was turned into an asic, could ram be put on board too?
Yes, there is generally high speed memory on a hard CPU or ASIC CPU but it is commonly configured as caches because that is where it provides the most advantage. There is sometimes CPU addressable SRAM especially for embedded applications. This reduces the chip count of boards but SRAM is generally not practical in large sizes (>8MB?). Modern processors like the i7 have *huge* caches which would be more than enough to run the 68k AmigaOS if it was addressable by the CPU instead of caches. Would you rather have less memory that is faster or more memory that is slower? The latter is probably more useful with modern software as most of the data could be in the 68k caches (The 68k is a cache miser so 32kB ICache and 32kB DCache should be a lot).
It would be cool to see an asic get made. Do asic's run faster and use less power?
ASICs are generally considerably faster than an FPGA but an FPGA can hold it's own in parallel tasks. ASIC and hard processors usually use less power but may require some work to improve power efficiency. FPGAs are low power already, especially for retro use. For embedded markets, the lower the power use and/or the better the performance for the power use, the better.
