[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference hydra::amiga_v1

Title:AMIGA NOTES
Notice:Join us in the *NEW* conference - HYDRA::AMIGA_V2
Moderator:HYDRA::MOORE
Created:Sat Apr 26 1986
Last Modified:Wed Feb 05 1992
Last Successful Update:Fri Jun 06 1997
Number of topics:5378
Total number of notes:38326

3159.0. "chip questions" by ALASKA::QUIRICI () Tue Nov 28 1989 12:15

    hi,
    
    is there any public information released by Commodore, or rumors,
    indicating when, if ever, the amiga will have an enhanced audio
    chip (closer to the performance of the chip on the NEXT)? a similar
    question can be asked about the video chip.
    
    as a related question, does the 7MHZ 68000 limit the amiga's
    audio/video to just about it's current levels?
    
    as another related question, does the amiga's memory speed (what
    is it, 120ns or something) provide similar limitations?
    
    finally, is jay miner still developing chips for the amiga?
    
    ken
T.RTitleUserPersonal
Name
DateLines
3159.1Memory bandwidth is the limitTLE::RMEYERSRandy MeyersTue Nov 28 1989 18:1952
Re: .0

>    is there any public information released by Commodore, or rumors,
>    indicating when, if ever, the amiga will have an enhanced audio
>    chip (closer to the performance of the chip on the NEXT)? a similar
>    question can be asked about the video chip.

The closest I've come to hearing something like that was when the
new president of Commodore, Copperman, spoke a month ago at a
Boston Computer Society meeting.  Someone said, "The Amiga needs
more colors and resolution."  Copperman said, "That is a known
requirement."

Personally, I hope the Amiga 3000 surfaces within 6 months
with a greatly enhanced chip set.  But, no one has promised
or hinted about that.

>    as a related question, does the 7MHZ 68000 limit the amiga's
>    audio/video to just about it's current levels?

No, the processor has very little to due with video or sound.  These
functions are off loaded to the custom chips.

>    as another related question, does the amiga's memory speed (what
>    is it, 120ns or something) provide similar limitations?

Yes, the memory speed is the bottle neck to speeding up the custom
chips.  Right now, some of the Amiga video modes use up so much
memory bandwidth that the processor has a hard time accessing memory.
At a even higher resolution with larger number of colors, you couldn't
fetch the memory fast enough to keep the video updated, even if
no other part of the system accessed memory!

Most Amiga memory boards will run at full speed with only 150ns
memory.

Note that in the IBM world, slow processors and memory are sometimes
coupled with really fancy graphics.  The way those folks get things
to work is to have a small amount of dedicated "video ram," high-speed,
dual-ported memory that runs much faster than the processor.  All-in-all,
it's a bit like the Amiga's "chip memory," except in the Amiga, chip memory
runs at the same speed as regular memory and it isn't dual ported (instead
access by the 68000 interleaves with access by the video system).

>    finally, is jay miner still developing chips for the amiga?

Nope, he's retired with a non-competition contract from Commodore.
In a recent interview, he did say that he'd be interested in helping
Commodore develop 32 bit enhanced versions of the custom chips.
But, Commodore hadn't contacted him, and he hasn't contacted Commodore.
Personally, I hope that Commodore is already pretty far along in
designing such chips.
3159.232-bit speedier?ALASKA::QUIRICIWed Nov 29 1989 10:378
    randy,
    
    thanks for your informative reply.
    
    as far as 'memory bandwidth', can we conclude that 32-bit memory, even
    if at 150 ns, would offer twice as much video-processing power?
    
    ken
3159.3ELWOOD::PETERSWed Nov 29 1989 10:5016
    
    
    	re .2
    
    	32 bit memory is not on the AMIGA bus. All the 68020 and 68030
    cards available have memory on board, daughter boards, or separate
    cables to run 32 bit transfers.
    
    	My GVP 68030 card uses 80 ns, 32 bit wide, burst mode transfers.
    This allows the processor to run with no wait states at 25 mhz. Wait
    states are then added when the processor goes out to the standard
    Amiga bus. The GVP card even has a built in disk controller for
    faster disk access.
    
    		Steve
    
3159.4Bigger buses and faster clocksTLE::RMEYERSRandy MeyersWed Nov 29 1989 13:4729
Re: .2

You are correct: An Amiga with 32 bit wide memory (or to be more precise
32 bit wide data bus) would have twice the memory bandwidth.

That's why a 68020 running at twice the clock speed can run up to four
times faster than a 68000.  It's getting twice the memory, twice as fast.

One standard technique to increase the speed of high end computers is
to increase the memory bandwidth my making the memory transfer size
larger.  A 4 byte request for memory (for a 32 bit CPU) is turned into
a request of 16 or 32 bytes from the memory system.  Once the data
shows up, it is stored in the cache, and the four bytes originally
requested is delivered to the CPU.  If the CPU then requests the next
contiguous four bytes of memory (as it frequently does), it is pulled
directly from the cache, with no access to the memory box at all.

As Steve points out in .3, none of the currently available 32 bit
memory for Amigas is used as chip memory.  Instead it usually is
private to a 68020 accelerator board.

An Amiga with 32 bit wide chip ram running at twice the clock speed
of current Amigas would have four times the memory bandwidth.  Four
times the memory could be accessed to produce a video image.  Imagine
a 320 pixel by 200 line by 20 bitplane display (one million colors)!
Or, a 1280 pixel by 400 line by 4 bitplane display (sixteen colors)!
Or, 640 pixel by 200 line by 16 bitplane display (sixty-five thousand
colors)!  All the above is an existing Amiga display mode enhanced
to use four times the memory bandwidth.