Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!ucbvax!bloom-beacon!eru!kth.se!cyklop.nada.kth.se!news From: d88-pfo@dront.nada.kth.se (Peter Forsberg) Newsgroups: comp.os.os2.programmer Subject: Re: HPFS block size Message-ID: Date: 29 May 91 20:11:08 GMT References: <716@cronos.metaphor.com> Sender: news@nada.kth.se (Mr News) Organization: Royal Institute of Technology, Stockholm, Sweden Lines: 26 In-reply-to: plee@cronos.metaphor.com's message of 28 May 91 22:52:02 GMT In article <716@cronos.metaphor.com> plee@cronos.metaphor.com (Ping-Dar Lee) writes: I am trying to write a performance related application on the top of the HPFS. The block orientied application might gain the performance benifit if both application and HPFS have the same block size. Does anybody know the block size of the HPFS. Thanks in advance! The HPFS uses a sector size of 512 bytes. I don't think, though, that it matters that much how much you read/write at a time, as the HPFS will adapt itself after *how* you read/write (i.e. how much at a time, sequentially/ randomly, etc.). It will do its prereading and write cacheing tricks depending on the info you give it by means of the way you access your files. It also uses history information stored about each files previous usage for optimization purpouses (this became like an advertise for HPFS :-). One tip for helping HPFS to give you optimal performance is this; if you know how large a file will eventually be when you create it, you should give that info to HPFS at that time. This helps HPFS minimize file frag- mentation. /Peter -- Peter A. Forsberg Royal Institute of Technology, Stockholm, Sweden. IBM Sweden, Banking & Finance Applications. E-mail: d88-pfo@nada.kth.se or peterf@stovm1.vnet.ibm.com #include