home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.arch:11903 comp.arch.storage:884
- Path: sparky!uunet!crdgw1!rdsunx.crd.ge.com!ariel!davidsen
- From: davidsen@ariel.crd.GE.COM (william E Davidsen)
- Newsgroups: comp.arch,comp.arch.storage
- Subject: Re: ?Concurrent DMA possible on smarter PC buses (EISA/MCA/Localbus)
- Keywords: EISA,MCA,Localbus,VESA,PC,IBM,smartIO
- Message-ID: <1992Dec23.192154.17137@crd.ge.com>
- Date: 23 Dec 92 19:21:54 GMT
- References: <1gntdfINNu7@cbl.umd.edu> <1992Dec16.211712.13142@twisto.eng.hou.compaq.com> <1992Dec17.153141.3926@urbana.mcd.mot.com> <1992Dec17.191131.17701@twisto.eng.hou.compaq.com> <1gul5nINN7ln@cbl.umd.edu>
- Sender: usenet@crd.ge.com (Required for NNTP)
- Reply-To: davidsen@crd.ge.com (bill davidsen)
- Organization: GE Corporate R&D Center, Schenectady NY
- Lines: 19
- Nntp-Posting-Host: ariel.crd.ge.com
-
- In article <1gul5nINN7ln@cbl.umd.edu>, mike@cbl.umd.edu (Michael Santangelo) writes:
-
- | So this theoretical EISA SCSI controller and this theoretical EISA FDDI
- | controller could both be doing DMA writes to main memory (interleaving their
- | accesses I assume)? Since PC's do not have multiport memory, I assume
- | the EISA subsystem would itself (on behalf of BOTH of these controllers) have
- | full control over the memory during the dual transfers, starving out the CPU?
-
- a) these are shipping, not theoretical
-
- b) every memory system has a limit to bandwidth, after which you make
- either i/o or cpu wait. Most people find better total throughput when
- they make the cpu wait. A quick calculation of the bandwidth of the bus
- and any remotely plausible i/o system will tell you the cpu doesn't wait
- long.
-
- --
- bill davidsen, GE Corp. R&D Center; Box 8; Schenectady NY 12345
- Keyboard controller has been disabled, press F1 to continue.
-