home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!sun-barr!west.West.Sun.COM!news2me.ebay.sun.com!jethro.Corp.Sun.COM!smli.Eng.Sun.COM!nonskid!bobeson
- From: bobeson@nonskid.Eng.Sun.COM (Robert Ellefson)
- Newsgroups: comp.arch
- Subject: Re: burst mode
- Date: 30 Jul 92 20:06:30 GMT
- Organization: Sun Microsystems
- Lines: 26
- Distribution: world
- Message-ID: <l7giu6INN41u@smli.Eng.Sun.COM>
- References: <Bs7vLK.Ln0@acsu.buffalo.edu>
- Reply-To: bobeson@nonskid.Eng.Sun.COM
- NNTP-Posting-Host: nonskid.eng.sun.com
-
- In article Ln0@acsu.buffalo.edu, chen-m@acsu.buffalo.edu (Mingteh Michael Chen) writes:
-
- >What is the burst mode of the Micro channel architecture?
- >and what makes it to be superior than other approaches? (maybe DMA?)
- >I have some vague idea that it transfers a whole memory "block"
- >instead of a memory word individually. But as long as the bus bandwidth is
- >fixed, each memory "word" still needs to be transferred sequentially...
- >
-
- As I remember, the Streaming Data Mode (which I am assuming is
- what you mean by 'burst mode') allows for block transfers to occur
- without needing to setup the address for each word, so that the clocking
- for each data word after the start of the block is much faster.
- (Normally, you setup the address before you clock the data, because different
- portions of the address space have different access times, which requires different
- wait state assertions.) Also, there is now (with the RS/6000, at least)
- a 64 bit streaming data mode, which uses both the address and data busses
- for data transfer.
-
- Other than the 64 bit transfer option, I'm not sure that the microchannel
- approach is significantly better than others. (Nubus at least has a comparable
- block transfer mode (not 64 bit)). Maybe someone more recently familiar with
- the other existing bus protocols can give a better comparison, though.
-
-
-
-