Path: utzoo!utgpu!news-server.csri.toronto.edu!turing.toronto.edu!funke Newsgroups: comp.periphs.scsi From: funke@turing.toronto.edu (Mark Funkenhauser) Subject: Re: Installing Wren VII on Sun Message-ID: <1990Jun13.015427.7051@jarvis.csri.toronto.edu> Keywords: Wren VII, Sun Organization: CSRI, University of Toronto References: <1990Jun6.211328.13096@utu.fi> <1990Jun8.223303.440@com50.c2s.mn.org> Date: 13 Jun 90 05:54:27 GMT Lines: 77 In article kusumoto@chsun1.uchicago.edu (Bob Kusumoto) writes: >chris@com50.c2s.mn.org (Chris Johnson) writes: >>disk_type = "Imprimis Wren VII 94601-15" \ >> : ctlr = MD21 : fmt_time = 4 \ >> : cache = 0x11 : trks_zone = 15 : asect = 0 \ >> : ncyl = 1947 : acyl = 2 : pcyl = 1949 : nhead = 13 : nsect = 80 \ >> : rpm = 3597 : bpt = 41301 > >I strongly advise against using this format.dat entry. The MD21 controller MD21 controller? What's this? This sounds like an Emulex product number. Wren scsi disk drives do not need external (Emulex) scsi controller boards. The scsi controller board is a builtin on the disk drive. Unless you are implying that the Wren VII uses the MD21 controller as its builtin controller board ?? >needs 4 cylinders (I don't remember the reason). But the cyl parameters >should be: > > ncyl + acyl + 4 = pcyl > It is possible that the Wren VII is formatted with 4 alternate cylinders, but on the Wren V and Wren VI I've only seen it formatted with 2 alternate cylinders. Two alternates is plenty. >I was told from my distributor that the total number of physical cylinders >was 1931, rather than the 1949 listed here, probably because he has a 1.5GB >version rather than my 1.2GB drive (the model number on my drive is 94601-12G). The only number I've ever seen for the Wren VII is 94601-12G. Seagate, which took over the Imprimis products, now calls it ST41200N. >Call your dealer and ask for some of the parameters, such has the number of >physical cylinders, the number of heads, the number of bytes per track, the >rpm of the drive. Go figure from there. This doesn't quite work for Wren SCSI disk drives as I will explain. The thing to remember about SCSI Wren drives is that come with a built in scsi controller. Most (if not ALL) of these scsi controllers format the disk with VARIABLE number of sectors per track. Thus, when you take the number of cylinders, number of heads and number of average sectors per track that Imprimis (or Seagate now) quotes in their product info and multiply them together, the numbers may not seem match exactly the quoted capacities. What you need to know is how the disk is *actually* formatted at the factory. Here at U of T we have machines (ICM3216) with which we can connect up scsi drives and determine how the disk was formatted including determining the maximum disk capacity. I admit that we have only purchased Wren V and Wren VI drives but they have always been pre-formatted with 512 byte sectors, 1 alternate sector per track and 2 alternate cylinders (e.g 32 alt tracks). I assume that when we order a Wren VII it will be pre-formatted similarly. (someone can correct me if I'm wrong). The trick to creating a format.dat entry for Wren scsi drives is to maximize the use of the available (e.g user accessible or logical) sectors on the disk. I've been told that this number for the Wren VII is 2026964 512-byte sectors. This number does NOT include the sectors of the alternate tracks (e.g. this number is the total logical data blocks on the disk). Thus once you have determined the ncyl number, you add 2 to get the pcyl number. Remember: scsi provides an interface where data is accessed by its logical block (sector) number, NOT the physical sector number NOR by (cylinder, head, sector). Therefore, we are free to choose any values for ncyl, nhead and nsect so that when they are multiplied out you get <= 2026964. Caveat: I am assuming that you are using format.dat to label the scsi disk, NOT re-format it. I'm not sure if this arbitrary use of ncyl, nhead and nsect can be used for re-format since I don't have the Wren product spec in front of me to check.