Path: utzoo!utgpu!attcan!uunet!lll-winken!csd4.milw.wisc.edu!mailrus!cornell!uw-beaver!rice!sun-spots-request From: mikem@xn.ll.mit.edu (Michael Maciolek) Newsgroups: comp.sys.sun Subject: Re: Need help with CDC Sabre 9720-850 and Xylogics 753 on Sun 4/280 Keywords: Hardware Message-ID: <1339@xn.LL.MIT.EDU> Date: 21 Apr 89 20:01:52 GMT References: <342@hydra.gatech.EDU> Sender: usenet@rice.edu Organization: MIT Lincoln Laboratory, Lexington, MA Lines: 119 Approved: Sun-Spots@rice.edu Original-Date: 1 Apr 89 00:56:34 GMT X-Sun-Spots-Digest: Volume 7, Issue 236, message 5 of 11 I offer this as a general posting so it can go on record in the archives in case anyone else needs the same information. Relevant to the original poster, who wants to run this on a 4/280 - we haven't tried it on a Sun-4, only on four Sun-3/280s. I see no reason why this should fail to work on a 4. Note that the controller settings and the drive settings are quite independent of each other. I was using the same drive settings when our Sabres were connected to xy451 controllers. The disk: 1. The CDC 9720 Sabre 850 megabyte drive has two (known to me) different control-logic boards, which you can tell apart by looking at the sector switces on top of the drive. The older style has three 6-position DIP switches, the newer style has two 10-position switches. The settings we use will configure the drive for 69 sectors of 594 bytes each. WARNING! The Xylogics 753 guide claims that a minimum of 600 bytes per sector are required. This configuration works for us, and we haven't crashed a disk or lost a file yet, but we assume no responsibility for others' use of these settings. Our sector switches are set like this: closed +------++------++------+ | x||x x ||xx | (3 x 6) o o o o |123456||123456||123456| |xxxxx || x xxx|| xxxx| ^^^^^^^ +------++------++------+ ||||||| open (status LEDs) closed +----------+ +----------+ |xxxx x x | |xxx xxxxx| (2 x 10) o o o o |123456789a| |123456789a| | x xx x| | xx | ^^^^^^^ +----------+ +----------+ ||||||| open (status LEDs) The controller: Oriented component-side up, P1 and P2 connectors at the bottom: xxxxxxxxxxxxx xxxxx xxxxx xxxxx xxxxx +---------------------------------------------------------+ | | | | | | | | | | | :jG | | | | | | jE ...... | | :: ...... | | :::::::::::: ...... | | jA ...... | | jBjCjD | | | +---------------------------------------------------------+ ##################### ##################### jA selects the address, which should be ee80 for xdc0. It should be set like this: . . . # . . . # . # # # . . . # . . . # . # # # jA jB thru jD determine the bus request and bus grant levels. They should be set like this: . . . ### . . . . ### . . . . ### . ### ### ### jB jC jD jE modifies the behavior of the controller's maintenance mode and self-test operation. It should be set like this: jE # . # . jG is "proprietary to Xylogics" and should be IN. When you run "format", tell it 594 bytes per sector, 68 sectors/track, 15 heads, 1379 data cylinders and 2 alternate cylinders. I think it will also prompt for "drive type", (if you are mixing different types of disk on the SAME controller, you will want to assign unique drive types to each drive before you format it) and "disk rpm" (3600). This configuration allows one slip-sector per track, which ought to be fine for any normal application. You will have about 687 megabytes of space after formatting, before creating filesystems. (my definition of megabyte is 1024x1024 bytes, by the way) Good luck! Michael Maciolek mikem@juliet.ll.mit.edu (preferred) G43 SysAdmin mikem@xn.ll.mit.edu MIT/Lincoln Laboratory