ewhac@well.UUCP (Leo 'Bols Ewhac' Schwab) (08/24/88)
In article <2@orbit.UUCP> swordfis@pnet51.cts.com (Tim Mitchell) writes: >Leo, as I understand it you've talked to Don Sidoroff (who wrote Turbo >Silver.) Impulse and Don have parted ways, and Don seems to have lost his >taste for the Amiga. When did this happen? I spoke to him not more than a month ago, and he seemed decidedly pleased with himself at what he had accomplished on the Amiga, and justifiably so. Silver is a *great* renderer. Impulse hasn't sent me the upgrade yet, though.... >Considering IBM's market saturation, a >cheap 3D animation system for them could spell trouble for us. [ ... ] I don't think it'll be cheap. Consider: A VGA bitmap is bigger than 64K. Intel architectures tend to lose real big on data chunks bigger than 64K, so you'll need a very fast CPU to bash all the bits at a respectable speed, which won't be cheap. The '386 doesn't lose too bad on >64K, but they're not cheap either right now. Even if you can find a cheap fast CPU that can throw eight bitplanes around without bogging down, you still don't have NTSC output. We're safe. For a while.... _-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_ Leo L. Schwab -- The Guy in The Cape INET: well!ewhac@ucbvax.Berkeley.EDU \_ -_ Recumbent Bikes: UUCP: pacbell > !{well,unicom}!ewhac O----^o The Only Way To Fly. hplabs / (pronounced "AE-wack") "Work FOR? I don't work FOR anybody! I'm just having fun." -- The Doctor
scott@applix.UUCP (Scott Evernden) (08/25/88)
In article <6908@well.UUCP> ewhac@well.UUCP (Leo 'Bols Ewhac' Schwab) writes: > > A VGA bitmap is bigger than 64K. Intel architectures tend to lose >real big on data chunks bigger than 64K, so you'll need a very fast CPU to >bash all the bits at a respectable speed, which won't be cheap. The '386 >doesn't lose too bad on >64K, but they're not cheap either right now. Not to be a stickler, but The standard VGA measures 640 x 480, or less than 40k. It is true that there are multiple bit planes, but they all live at the same address. The hardware takes care of writing to the planes in parallel under control of mask, rotate, enable, select, and a assortment of other baroque registers. To read from an individual plane, you tell a register which one you want and read. So, at least for this display mode, you generally don't have to muck with segment registers too much. Anyways, the killer with EGAs/VGAs, is that you _have_ to get down and dirty with the IO registers to accomplish any real work. This is quite a painful process on the non-memory-mapped-IO 80x8x's. You repeatedly play games like: mov dx,SEQ_ADDR; mov al,WPL_REG; out dx,al; inc dx; mov al,1; out dx,al). Registers are pretty much lost, like DX, for this- on an architecture that is already register-starved. Also, these graphic adapters and their IO ports are byte oriented, and since the intel chips like to load bytes into registers backwards, ops like rotate (eg., for blitting) have to be done one-byte-at-a-time; 32 bit registers don't do a thing for you. Finally, it would seem that hardware capable of multi-plane writes in parallel would be a win. This might be true if not for a generally unpublished feature of these graphic boards which is the number of WAIT states they require to be accessed. I understand that some EGA boards introduce up to 12 wait states if written to. -scott
daveh@cbmvax.UUCP (Dave Haynie) (08/26/88)
in article <6908@well.UUCP>, ewhac@well.UUCP (Leo 'Bols Ewhac' Schwab) says: > A VGA bitmap is bigger than 64K. Intel architectures tend to lose > real big on data chunks bigger than 64K, so you'll need a very fast CPU to > bash all the bits at a respectable speed, which won't be cheap. The '386 > doesn't lose too bad on >64K, but they're not cheap either right now. Unless you're running '386 UNIX or something else that uses native mode, the '386 will loose just as bad on >64K chunks as an 80286 or 8088. It does have a 32 bit bus, but even that's not a critical issue in the current systems. About all a '386 gives you for MS-DOS computing is a clock speed you can't get with a '286 chip yet. > We're safe. For a while.... It would be nice. Don't know if anyone's ever safe in this industry... > Leo L. Schwab -- The Guy in The Cape INET: well!ewhac@ucbvax.Berkeley.EDU -- Dave Haynie "The 32 Bit Guy" Commodore-Amiga "The Crew That Never Rests" {ihnp4|uunet|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy "I can't relax, 'cause I'm a Boinger!"
ewhac@well.UUCP (Leo 'Bols Ewhac' Schwab) (08/26/88)
In article <6908@well.UUCP>, I write: > Impulse hasn't sent me the upgrade yet, though.... > Well, wouldn't you know it. I posted this comment yesterday, and today in the mail arrives the Silver upgrade. "Never mind." Hmmmmm. I wonder if the phenomenon is consistent? Let's try it: Commodore hasn't sent me the latest version of 1.3 yet, though.... _-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_ Leo L. Schwab -- The Guy in The Cape INET: well!ewhac@ucbvax.Berkeley.EDU \_ -_ Recumbent Bikes: UUCP: pacbell > !{well,unicom}!ewhac O----^o The Only Way To Fly. hplabs / (pronounced "AE-wack") "Work FOR? I don't work FOR anybody! I'm just having fun." -- The Doctor
peter@sugar.uu.net (Peter da Silva) (08/28/88)
In article <786@applix.UUCP>, scott@applix.UUCP (Scott Evernden) writes: > The standard VGA measures 640 x 480, or less than 40k. It is true that > there are multiple bit planes, but they all live at the same address. > To read from an individual plane, you tell a register which one you want and > read. This is, of course, bank selecting. Which is about the only thing worse than an intel segmented architecture (which amounts to bank selecting on chip). You'd do better with segments. -- Peter da Silva `-_-' peter@sugar.uu.net Have you hugged U your wolf today?
jimm@amiga.UUCP (Jim Mackraz) (08/29/88)
In article <2572@sugar.uu.net> peter@sugar.uu.net (Peter da Silva) writes: )In article <786@applix.UUCP>, scott@applix.UUCP (Scott Evernden) writes: )> The standard VGA measures 640 x 480, or less than 40k. It is true that )> there are multiple bit planes, but they all live at the same address. )> To read from an individual plane, you tell a register which one you want and )> read. ) )This is, of course, bank selecting. Which is about the only thing worse )than an intel segmented architecture (which amounts to bank selecting on )chip). It's a little more than that. You select things akin to "pen colors" and a single write can affect ALL planes, setting 0's in some, 1's in others. EGA, at least, come up an inch short of being able to set arbitrary foreground and background pens for a dashed line, for example, due to bad thinking in the register conventions (I got a long letter about this printed in Programmers' Journal, mostly rehashing an observation of Neil Katin). But it's better than bank selecting, and smells kind of like the kind of thing you'd want to do for a parallel Amiga blitter, if you were building one: separate minterms per bitplane. )You'd do better with segments. But, Peter, you get to be burned by BOTH at once!! ) Peter da Silva `-_-' peter@sugar.uu.net jimm -- Jim Mackraz, I and I Computing amiga!jimm BIX:jmackraz Opinions are my own. Comments regarding the Amiga operating system, and all others, are not to be taken as Commodore official policy.