page@ulowell.cs.ulowell.edu (Bob Page) (08/19/87)
Henry_Burdett_Messenger@cup.portal.com wrote: >These memory systems are *all* compatible with the new blitter chip, and, Has Atari started shipping any of these blitter chips? >unlike the Amiga, all memory on the ST is created equal (no "fast" and >"slow") Using the Amiga's definition that the 680x0 has unlimited access to the memory bus *all the time* in 'fast' memory, and has to share it with the 3 coprocessors in 'chip' [not 'slow'] memory, I'd say that the ST is *all* 'chip' memory, and no 'fast' memory ... unless there's something in the ST architecture that Atari hasn't been saying. >>while the ST seems to be the machine of choice for synth stuff, >>the Amiga seems better overall. > >This is still the case, and will continue to be the case. Unlike the Amiga, >the ST was *designed* for MIDI. Yamaha made a machine that was *designed* for MIDI also. They didn't sell too many of them, since they couldn't do much else very well. I'm not saying the ST follows the same trend, just that your argument isn't very convincing. You can't say the ST will always be the machine of choice for synth stuff, because you don't know that. If/When Atari puts in an MMU, blitter or a new OS (Multitasking TOS, UNIX, who knows what else), etc, this status could change quickly. Or another manufacturer could build a machine more attractive to musicians. Who knows? In my opinion, the Mac is the machine of choice for synth stuff, because the software support is wonderful. >And if you don't take my word for it, ask Tangerine Dream. They use STs NOT because it's an ST, but because the software they want/need is available on STs. Unlike a lot of us, they're more interested in getting a job done (composing/editing music) than worrying if they need to defend their purchase to an international audience. ..Bob -- Bob Page, U of Lowell CS Dept. page@ulowell.{uucp,edu,csnet}
daveh@cbmvax.UUCP (Dave Haynie) (08/21/87)
in article <1637@ulowell.cs.ulowell.edu>, page@ulowell.cs.ulowell.edu (Bob Page) says: > Xref: cbmvax comp.sys.amiga:7417 rec.music.synth:1269 > >>unlike the Amiga, all memory on the ST is created equal (no "fast" and >>"slow") > > Using the Amiga's definition that the 680x0 has unlimited access to > the memory bus *all the time* in 'fast' memory, and has to share it > with the 3 coprocessors in 'chip' [not 'slow'] memory, I'd say that > the ST is *all* 'chip' memory, and no 'fast' memory ... unless there's > something in the ST architecture that Atari hasn't been saying. I'm not a MIDI-head, but I am a hardware engineer. Based on the info I've heard about the Atari blitter, the entire Atari memory range is "chip" memory under this chip, but it's not necessarily as bad or as good as an Amiga system. For the most part, the Amiga chip memory slowdown comes from high resolution graphic modes. If you go over two bitplanes in either a 640x200 or 640x400 display, the Denise chip will have to start eating into 68000 CPU time to achieve it's display. The ST solves this problem by supplying only a 640x400x1 non-interlaced display, which takes the same bandwith as the Amiga's 640x200x2 non-interlaced or 640x400x2 interlaced displays. As long as we're sticking to ST display rates, there's really no FAST or SLOW RAM on either machine. The slowdown kicks in when a Blitter or Bimmer is used at these resolutions. The Amiga Bimmer handles memory accesses twice as fast as the 68000, but once video, sprites, floppy, and audio DMA have taken place, there's not much room left for the Bimmer. So the Bimmer often steals cycles from the 68000. This stealing is dynamic; the Bimmer will only steal as much as it needs. As I understand it (I'm certainly no expert on the Atari blitter, and I'd appreciate any comments from Atari blitter knowledgable folks), the Atari blitter executes full 68000 cycles, and will in fact (as long as it has something to do) swap cycles with the 68000; something like, 68000 gets 64 cycles, blitter gets 64 cycles. I guess in a standard Atari application, the result is that either the blitter is running full time (the 68000 has nothing to do, being that it's waiting for the blitter to finish), or the 68000 is running full time, having nothing for the blitter to do. On the Amiga, the slowdown has more meaning, in that as long as the Bimmer can use alternate cycles, the 68000 can run at full speed on another task, even if one task is waiting for the Bimmer to finish. So, plainly put, in a single tasking environment, any blitter slowdown is moot, since it's doing something faster than a 68000 can do it. And at Atari resolutions, no video slowdown is possible, though an Amiga application CAN choose to trade processor speed for increased video resolution. After all of this, I should say something about FAST memory. FAST memory is merely a name for memory on an Amiga that's not directly addressable by the custom chips (i.e. Agnus, Paula, Denise). The CHIP memory in an Amiga is on a completely different buss than FAST memory and the 68000. So if the 68000 is talking to FAST memory, I/O devices, or ROM when the Bimmer is talking to CHIP memory, there's no slowdown; they can talk at the same time. Again, this is really only useful for multitasking. Since the Atari blitter basically "clips" onto the 68000, there's no choice here; whenever the Atari blitter wants the buss, the 68000 has to stop; whenever the 68000 wants the buss, the Atari blitter must stop. > >>>while the ST seems to be the machine of choice for synth stuff, >>>the Amiga seems better overall. >> >>This is still the case, and will continue to be the case. Unlike the Amiga, >>the ST was *designed* for MIDI. If you look really close, you'll find that both machines were specifically designed for MIDI. Atari chose to put about $6.00 worth of MIDI jacks and optoisolators into the ST, instead of offering them as an add-on, but other than that, neither machine is better suited for MIDI than the other. My reasoning behind this is probably obvious to communications folks, and that centers around available BAUD rates. The standard baud rates go in power-of-two progressions: 1200-2400-4800-9600-19,200-38,400-ETC. If your ACIA is capable of producing the 31,800 or whatever that MIDI requires, you've obviously designed your ACIA for MIDI; there's no other reason to support that frequency. Since the Atari, Amiga, and MacIntosh produce these frequencies, you can assume that the were all designed to handle MIDI. The C64/C128, on the other hand, won't produce these data rates, so you can assume that these machines weren't designed for MIDI. Yet you can buy a C64/C128 MIDI interface card (contains one MIDI rate ACIA) for less than most Atari MIDI software packages. MIDI's cheap enough so it really doesn't matter. Though you really can't complain about the degree of software support the ST's achieved for MIDI, just based on the built-in interface logic. -- Dave Haynie Commodore-Amiga Usenet: {ihnp4|caip|rutgers}!cbmvax!daveh "The A2000 Guy" PLINK : D-DAVE H BIX : hazy "God, I wish I was sailing again" -Jimmy Buffett, Dave Haynie