Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!zaphod.mps.ohio-state.edu!sol.ctr.columbia.edu!ira.uka.de!smurf!urlichs From: urlichs@smurf.sub.org (Matthias Urlichs) Newsgroups: comp.sys.mac.programmer Subject: Re: async write to disk (MPW C) Message-ID: Date: 1 Feb 91 17:57:08 GMT References: <1016@ra.MsState.Edu> <48623@apple.Apple.COM> Organization: University of Karlsruhe, FRG Lines: 30 In comp.sys.mac.programmer, article <48623@apple.Apple.COM>, stevec@Apple.COM (Steve Christensen) writes: < > < >[ async PBWrite to hard disk ] < >The write is taking about 50 msec and returning with ioResult already 0. < > < >Any suggestions? < < You're operating from a misconception. The SCSI chip in the IIfx does indeed < provide DMA support, but the SCSI Manager doesn't use the feature. Therefore < the SCSI Manager itself is still synchronous, so the data collection and < writes to disk are happening in a serial fashion, not concurrently. < It gets worse -- the current SCSI Manager interface prevents any attempt at using the SCSI bus asynchronously, i.e. letting the target device disconnect if it's going to seek, multiple outstanding requests (one per device for SCSI-1, possibly more for SCSI-2), another initiator on the bus, etc. One can play some dirty tricks (some backup software does, if it knows that it has total control of the bus), but SCSI is intended to be target-driven (i.e. the hard disk on the other end tells you what it wants to do next). This doesn't match the SCSI Manager's idea of how to access the bus. Whatever happened to the new SCSI Manager, as described in the Alpha 7.0 documentation ?? Some people out there think it is (has been?) the most necessary feature of 7.0, at least for their applications... -- Matthias Urlichs -- urlichs@smurf.sub.org -- urlichs@smurf.ira.uka.de /(o\ Humboldtstrasse 7 - 7500 Karlsruhe 1 - FRG -- +49+721+621127(0700-2330) \o)/