Commodore Free 11

home *** CD-ROM | disk | FTP | other *** search

/ Commodore Free 11 / Commodore_Free_Issue_11_2007_Commodore_Computer_Club.d64 / t.iv jens 4 < prev next >

Wrap

Text File | 2007-01-01 | 9.9 KB | 310 lines

u Interview with Jens Schofield Individual Computers http://www.jschoenfeld.com/index.htm PART 4 of 4 Q-Are you currently working on any- thing new, as of yet undisclosed? And, would you care to share? I promise to keep it between us & a few hundred readers. A-I'm working on two flickerfixers, one for the Amiga that will output a 75Hz picture for PAL screens, and another for the C64, because I keep losing 15kHz C64-compatible monitors. Both products will most probably not be finished this year [this interview was made in 2006, the flickerfixers are put on hold because of the very successful contract work] Since we agreed that this interview is not published before October 2006, I can also share the biggest project with you: I'm working on the Amiga chipset, together with Oliver Achten, the developer of the MMC64 & now the project leader of "Clone-A". He's one of the two guys that I have hired since January of this year, & we started the plans & preparations for this project in late October of 2005. The goal is to create a cycle-exact Amiga chipset replica with today's technology. Oliver & I are going to demonstrate the proto- type at this year's [2006] AmiWest show. The demonstration will be some- thing that the visitors can influence: I'd like to encourage people to bring their favorite disks with them & try them in our computer. Any game, demo or program, just bring the disk -- the computer has 2MB chipmem & a 68000 processor. Let me emphasize again that we are not selecting the kind of soft- ware that will be demonstrated -- we will be completely unprepared for the kind of programs people bring. We just want to demonstrate that if it runs on an A500, it'll also run on our chipset. The current name of the project is "Clone-A". We're replacing each chip of an Amiga 500 bit by bit. At this point, Denise, Gary, Paula & the CIAs are completely removed & replaced by small FPGA boards. It's pretty certain that we won't have a finished Agnus chip at AmiWest, but we already have a very good idea of what's inside, because we have already reverse engineered the inter-chip communication, & to do this large parts of Agnus need to be known. [update: the Clone-A chipset is finished, including Agnus, & has been demonstrated on march 31st, 2007]. Our approach to a re-implementation of the chipset is surely one that takes the longest, compared to a full implementation "from scratch" that Dennis van Weeren did for his Minimig. There's also NatAmi that tries a full re-implementation based on documentation. Oliver & I are going the most complicated route, but being able to work with any combination of the real chipset & our FPGA-replace- ments leaves no room for mistakes. We have no other chance than to be 100% correct, & everytime we're fixing a bug, we discover that the fix actually produces a smaller design. It's really amazing what Jay Miner & his colleagues have crammed into about 20,000 transistors per chip! I really have no idea when I can turn this chipset into a product, but it gives a lot of possibilities. You can think of a new classic Amiga board, an extended chipset with more chipmem & a faster blitter, a daughter card for whatever mainboard that you can buy today & many other things. I'd also be open to a cooperation with someone who wants to turn this chipset into a toy like the C64DTV. A portable Amiga that runs for many hours on a cellphone battery is definitely within reach. [update: the first Clona-A machines are planned for Xmas 2007]. Q-Could you elaborate a bit about the different approaches between "Clone-A" & the Minimig? A-Dennis took the UAE source & the description of the chipset registers to make a forward- engineered Amiga- compatible chipset. This is not neces- sarily a bad move, but like I said, our approach leaves no room for mistakes, while he has much more freedom for a similar, but not exactly-the-same implementation. Before we take the next step of implementing a part of a chip, we do extensive reverse-engineering, for example by writing test programs & doing measurements on the chips with a logic analyzer while the programs are running. We also do really strange things to find out what the inner workings of the chips are, for example programming them in ways that you're not supposed to. Q-How many man-hours have gone into bringing it to its current state, & how much more time do you expect will be required before you have something finished? A-We never counted the hours, but the combined manpower is easily 2000 hours, given the fact that we're both not working full- time on the project, & that an average working-year has 2000 working hours. I'd say that we need another 300 hours to bring Agnus to a state where it can be considered cycle- exact. [update: it took a lot more than that...] It's really hard to tell how much work has gone into the project, because I have put a lot of knowledge into it that I gathered before the start of Clone-A. I once did an add-on for a TV studio that wanted to have a special Genlock interface: They not only wanted to mask between the computer-picture & the TV picture, but they wanted to define a transparency for every color that the Amiga displays. I was only able to do that with an almost complete implementation of the Denise bitmap-logic, so I already had good knowledge about Denise back in 1995. Then Oliver did further research for his implementation of Denise, which has many features that haven't been discovered yet, even by emulators. The next thing is the part of Paula that controls the disk drive: There's a lot of Catweasel knowledge that I explained to Oliver, & he implemented the floppy part in record time. That's another big advantage we have: We are a team, while the other projects are made by single people. Q-So, let's just get this straight, the Clone-A will basically be a complete miniature-A500, 100% hardware compatible, that can be offered in any configuration you might want an A500 in today? A500-in-a-joystick, A500-in-a desktop-case, A500-in-a-handheld, etc? A-Anything that involves an Amiga-on- a-chip (such as a joystick or handheld) requires huge investments that I can't do alone. However, our approach compared to the forward-engineering approach that Dennis does - produces the smaller design, so in the end, ours is better-suited to be produced as a chip. If there is an investor who wants an Amiga on a chip, Oliver & I can offer the smallest & therefore cheapest design with the positive side-effect that it is cycle-accurate. Q-Does it come with an IDE interface & RF-modulator? Would there, by any chance, be Catweasel technology doing the floppy controlling? A-I'd say that a mini-ITX Clone-A should have everything that a classic Amiga user has today: Halfway decent CPU (at least an 030), IDE, floppy that works with normal 1.44Mb drives & a PC keyboard & mouse connector. The monitor should be VGA, not an RF modulator. There's a good demand for A1200 boards, but supply is limited & the prices are high. However, anything that I'm dreaming of here should be taken as what it is: Speculation. I don't know what the final specs will be, nothing is written in stone yet. Q-How do you anticipate products incorporating Clone-A being marketed? What will the first revision be able to do, & what can be expected from future revisions? Will there some day be a "Clone-B" emulating the AGA chipset? A-AGA does not add very much to the design size. Since the blitter & the complete Paula chip stayed the same, we're only talking about 27 instead of 25 DMA channels & a local bus performance upgrade by a factor of 4. That's something we can easily accomplish with the type of memory & the type of logic chips that we're using. Remember that we're talking about a machine that was up-to-date in 1985. According to Moore's law, AGA performance should have been reached 36 months later in 1988. We are using 2007 FPGA technology, where talking to 133MHz SD-Ram is no trouble at all. Q-If a new revision would at some point be released, would the original Clone-A be upgradable by a quick reflash of the FPGAs? A-Yes, as I said earlier, I'll try to keep all my future hardware recon- figurable. Q-Will it come with an "Amiga OS & ROM" licence, making it a complete A500/2000 once you add the basic peripherals? Or could it be bundled with Amiga Forever? A-That's what I tried to discuss with Bill McEwen, but he failed to prove that Ainc. is the rightful owner of the Amiga IP over almost two years now. There are more options, because there are multiple authors who made Kickstart replacements for the Amiga1000 back in the days, & I managed to dig up two of those authors. However, a true name license from the big brand of the 80s is the most desirable option, & that's work in progress; nothing to repport at the moment. Q-Have you been in discussion with any- one else about turning this into any specific type of mass market product? A-Sure, but no written agreement has been signed yet, so I would not like to give their names here.I don't expect that it's reasonable for all the good boys & girls out there to expect Santa to bring them one of these this year? [2006 answer] No chance, sorry. As I said, Agnus still needs a lot of work, & after that, we still have to implement a 68000 processor into an FPGA before an Amiga-on-a-chip can be made. I even doubt that Santa will make it in 2007, but you never know. The worst thing that can happen to a hard- ware design is that the news about the prototype is spread too soon. If Minimig & NatAmi would not exist, I would not be demonstrating Clone-A at AmiWest this year, I would rather wait until the first product with the results of the work is ready to ship. Most people don't realize that a working prototype only makes perhaps 30% of a finished product. [2007 update] Oliver & I are working hard on Clone-A so the first version can be shipped this year. However, Clone-A is still no finished product & the only thing I can ask for is patience. Q-Lastly, anything else you would like to add? A-This has gotten really extensive. I hope not boring. Thanks for everybody's time!