home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.atari.st
- Path: sparky!uunet!elroy.jpl.nasa.gov!hanauma.jpl.nasa.gov!hyc
- From: hyc@hanauma.jpl.nasa.gov (Howard Chu)
- Subject: Re: WHAT'S GOING ON ?
- Message-ID: <1992Nov8.000013.24978@elroy.jpl.nasa.gov>
- Sender: news@elroy.jpl.nasa.gov (Usenet)
- Nntp-Posting-Host: hanauma.jpl.nasa.gov
- Organization: SAR Systems Development & Processing, JPL
- References: <cdf1tb+@rpi.edu> <1992Nov3.101001.12943@elroy.jpl.nasa.gov> <1deq7qINNp0d@phakt.usc.edu>
- Date: Sun, 8 Nov 1992 00:00:13 GMT
- Lines: 57
-
- In article <1deq7qINNp0d@phakt.usc.edu> baffoni@phakt.usc.edu (Juxtaposer) writes:
- > I have to disagree. A PDS is the only way to go for CPU upgrades or
- >other products that require a 0waitstate operation within the machine. What
- >good would it be to have a 50MHz '040 if _every_ access to main memory incurs
- >one or more waits? I agree that VME is a good bus architecture - 1WS incurred
- >very fast, and with the larger bus (6U) it is accesses about everything the CPU
- >can. The point is that not everything is appropriate for an external bus like
- >that. True, you don't want to put your graphics adapter in a PDS (unless, like
- >the Falcon it is the only one you have :( ) but you don't want the '040 in the
- >VME either.
-
- I certainly wouldn't put a processor upgrade out on the VMEbus. Nor would I
- put it in a PDS; I'd pull out the old processor and put the new one in the
- vacated socket... The PDS is completely useless for this application anyway
- (if we're talking about the Falcon) since it only provides a 68000-style
- A24/D16 interface. (Of course, no matter what you had, it would be pretty
- tough to upgrade a <=68030 system to 68040 since the '040 doesn't support
- an asynchronous bus...)
- >
- >>VMEbus is not very different from what you'd have to cram into a processor-
- >>direct slot. The difference in bus control/arbitration additions is negligible
- >>now since you can get single-chip VME bus interface controllers and such.
- >
- > One waitstate (80ns or less) doesn't sound like much...until you try
- >to access memory. It is questionable whether a VME based DSP/sound sytem would
- >do well, besides lacking the ability to bus master (on the Atari, not the spec)
- >so that sound DMA to the DSP or other chips (like DAC/ADC) would be impossible.
- >I am not saying that you couldn't transfer to those chips very quickly -
- >bandwidth to the 16-bit/8chnl ADC/DAC would only be max 800k/s, but that is not
- >insignificant even on the TTbus, let alone the MegaSTe bus. However, a PDS
- >based DSP/sound system would definately work (if the PDS supported bus
- >mastering, all data/address/interrupts ). Too bad the MegaSTe and the TT
- >don't have a (decent) PDS.
-
- Like you say, nothing rules out putting a decent DMA engine on a VME device.
- Also, you seem to be overlooking the possibility of block-mode transfers
- when you keep referring to that 1 wait-state.
- >
- >>Of course, you don't need to put *everything* on the VMEbus. It still makes
- >>sense for system RAM and ROM to live on a private bus, just like the way STs
- >>and TTs already work. But you could stick myriad enhancements on the VME -
- >>graphics boards, custom coprocessors, ethernet network controllers, serial
- >>port multiplexors, etc. etc. etc... VME is definitely the best way to go.
- >
- > Agreed. For those (relatively) low bandwidth applications.
- >
- >> -- Howard Chu @ Jet Propulsion Laboratory, Pasadena, CA
- >>
- >>All true wisdom is conveyed in one-line witticisms.
- >
- >-Mike
-
-
- --
- -- Howard Chu @ Jet Propulsion Laboratory, Pasadena, CA
-
- All true wisdom is conveyed in one-line witticisms.
-