home *** CD-ROM | disk | FTP | other *** search
- Path: grafix.xs4all.nl!john.hendrikx
- Date: Wed, 07 Feb 96 22:44:34 GMT+1
- Newsgroups: comp.sys.amiga.programmer
- Distribution: world
- Subject: Re: Amiga doesn`t need Planar!
- MIME-Version: 1.0
- Content-Type: text/plain; charset=iso-8859-1
- Content-Transfer-Encoding: 8bit
- From: john.hendrikx@grafix.xs4all.nl (John Hendrikx)
- Message-ID: <john.hendrikx.4coj@grafix.xs4all.nl>
- Organization: Grafix Attack BBS Holland
-
- In a message of 06 Feb 96 Michael Van Elst wrote to All:
-
- >> Not really. I hardly call a write-mask 'hardware' anyway,
-
- MVE> Ah. Every extra hardware that chunky needs is "hardly hardware" at
- MVE> all.
-
- So isn't it true that Chunky is quite useable with a lot less hardware?
-
- MVE> Do you see a biased view here ?
-
- How could I possibly deny this, I mean seeing how Chunky has taken over almost
- every computer platform surely would have its effect.
-
- >> Of course? What kind of platform do you have in mind anyway? I'm
- >> talking Pentium's, P6, 68060's, PPC's, etcetera... memory is slow for
- >> these fast processors.
-
- MVE> You mean the x * 100MByte/s of memory bandwidth on the graphics card
- MVE> is slow ? It is just the CPU that is slow compared to the graphics
- MVE> hardware.
-
- So what is 'x'? Without it this statement could mean anything.
-
- However, the bus TO the gfx-card is slow, so doing a few extra tests'and
- branches for each pixel plotted would not make things any slower, and thus
- wouldn't be a big overhead, like you claimed in the first place.
-
- >> If you want multiple layers with planar you LOSE colors. Ie, 8 bitplanes
- >> = 256 colors. 2x 4 bitplanes = 32 colors.
-
- >> Not with Chunky,
-
- MVE> You are comparing apples and bananas. You say: "if you can use the
- MVE> harware to produce a layered display you lose colors if you use the
- MVE> same amount of memory" but "if you render the whole display you can
- MVE> simulate layering with software".
-
- If the hardware limits you to some arbitrary number of colors when it comes to
- layering displays then you can be glad that the display is Chunky as this kind
- of display can do quite rapid layering of its own without losing any colors.
-
- MVE> Don't you see that this has nothing to do with chunky vs. planar ?
-
- I wouldn't say nothing, however if you want a specific end-result then you
- better choose carefully what hardware to use. I mean, if you require a fast
- 256 color 3d view, with a cockpit overlayed then you either chose a 8-bit
- Chunky card, or a 9-16 bit Planar card.
-
- >> For consoles yes, they don't have a 'big CPU' to handle this kind of
- >> stuff.
-
- MVE> And why not ? Because it is too expensive for consoles. The special
- MVE> hardware that is NOW available is much cheaper and can do the same.
-
- Special hardware can only do what is popular, the latest effects will still be
- CPU based at first before they become popular, IF they become popular that is.
-
- >> is a much better way to solve these kind of speed-problems. What use is
- >> that expensive DSP when it is just sitting there having nothing to do,
-
- MVE> You mean the $10 DSP that handles the audio ports ? Or the $100 DSP
- MVE> that decodes the network video streams ?
-
- Why not add a $80 PPC603/66 and get an overall speed boost, not just at
- handling video streams, or the things the OS uses the DSP for. For most users
- this will safe them more time on overall.
-
- >> or when apps
- >> simply not use it but use their own CPU based routines?
-
- MVE> C0d3r stuff ?
-
- No, lack of foresight in the OS, or lack of upgrades. If there is no OS
- supported way to do something a DSP can handle then you could hardly call
- writing your own CPU based MPEG decoder c0d3r stuff.
-
- >> Adding an extra CPU is more expensive, but it can do so much more.
-
- MVE> Usually it is weaker than special hardware and hardly used too.
-
- Depends on the OS. It would be hard to get a single-processor based OS to make
- good use of multiple processors, but if it is was there right from the
- beginning the benefits could be quite high. Tasks involving frequent Harddisk
- access for example divide their CPU power over 3 different tasks in AmigaOS
- already. And almost any task can be split over multiple CPU's with a bit of
- forethought. But it would be hard to add now, and make decent use of it, even
- though it is never too late to start as I think multiprocessor designs are
- going to move into the homes real soon now.
-
- >> Certainly
- >> it will not be as fast as a real MPEG or JPEG decoder, but then again,
- >> what use is a JPEG decoder if you want to do texturemapping, or doing
- >> complex sound effects?
-
- MVE> What if you want to do texturemapping _and_ real-time MPEG ?
-
- Let me interpret your questions in the wrong way: there is only one DSP which
- can only handle one of these tasks at the time. The other task will have to
- wait. With multiple CPU's you will atleast get something done, but slower of
- course, after all you're doing multiple things at the same time.
-
- Also, why would CPU + DSP be faster at TMap + MPEG than 2 CPU's? The CPU doing
- the TMapping might actually have time left to help with the MPEG decoding.
-
- >> Do I really need to explain this in detail for you? Do you really don't
- >> see WHY a tmapped or shaded polygon is drawn one pixel at the time (as in
- >> multiple calculations are performed per pixel to give each pixel the
- >> correct look)?
-
- MVE> Now, for me texture mapped polygons are drawn in strips of 1-32 pixels
- MVE> because that's the length of the pattern buffer. It doesn't matter that
- ^^^^^^^^^^^^^^
- They will be stored one by one in the pattern buffer, maybe storing a few in
- registers before writing them to the pattern buffer, but the loop which
- calculates the pixels will clearly be working at one pixel at the time.
-
- MVE> pixels might be computed individually but these are hardly _drawn_
- MVE> individually.
-
- "Drawn" to me means writing them to memory. For me the C2P step is not the
- thing which draws the scene to the screen, instead the TMapping routine which
- builds the Chunky buffer is the thing that draws them.
-
- >> There is no need for a bandwidth increase. With a faster CPU you can
- >> just display more CPU intensive effects, the bandwidth could remain the
- >> same. In other words, with a 486/25 I could play 'DOOM' on my clone at a
- >> decent speed, now with my Pentium/90 I can play 'Magic Carpet' at a decent
- >> speed. The gfx-card is just as fast, still I gained a lot by upgrading my
- >> CPU.
-
- MVE> And that's an effect of chunky hardware if "number of memory accesses"
- MVE> is all that matters as you said ?
-
- For Chunky the number of memory accesses is what determines the maximum speed
- in FPS of a certain screentype. The CPU
- -and- Hardware determine how cool the effects can be.
-
- For Planar the hardware determines how cool the effects can be, and determines
- the maximum speed as well. The CPU is not involved.
-
- The situation with Amiga is amusing however: We now have the situation that
- the CPU actually IS involved in how cool the effects can be with Planar. This
- is because our Planar is actually so damned slow and so old that the steady
- upgrade of CPU speed has actually closed the ENORMOUS gap between Planar
- hardware speeds and CPU speeds of the same age. The speed difference is so big
- that the CPU can afford to do a complete C2P conversion to get new cool effects
- to be used with the old planar system.
-
- Grtz John
-
- -----------------------------------------------------------------------
- John.Hendrikx@grafix.xs4all.nl TextDemo/FastView/Etc... development
- -----------------------------------------------------------------------
- -- Via Xenolink 1.981, XenolinkUUCP 1.1
-