Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!seismo!ut-sally!husc6!think!ames!ucbcad!ucbvax!jade!eris!mwm From: mwm@eris.BERKELEY.EDU (Mike (My watch has windows) Meyer) Newsgroups: comp.sys.amiga Subject: Report on the Xebec Hard Disk Message-ID: <3447@jade.BERKELEY.EDU> Date: Mon, 4-May-87 16:31:18 EDT Article-I.D.: jade.3447 Posted: Mon May 4 16:31:18 1987 Date-Received: Wed, 6-May-87 04:47:13 EDT References: <3424@jade.BERKELEY.EDU> Sender: usenet@jade.BERKELEY.EDU Reply-To: mwm@eris.BERKELEY.EDU (Mike (My watch has windows) Meyer) Organization: Missionaria Phonibalonica Lines: 64 In article <3424@jade.BERKELEY.EDU> spencer@eris.BERKELEY.EDU (Randy Spencer) writes: >used in Mike's Xebec. It is also on the Microbotics hard disk (this That's Mike's ex-Xebec... The tale follows... I took one home. Plugged it into the Amiga without the memory expansion. Worked like a charm, if I booted off of their disk. After wading around some, I figured out what was going on to boot, at least enough that I didn't mind re-arranging things some. At this stage, it became obvious that the documentation was inadequate. It says "you can either autoconfig the drive, or make it work out of your startup-sequence." No words on how to do either. By examing their startup-sequence, I figured out what should happen. Step two. Plug my memory expansion back in, then plug in the Xebec controller. It no boot. In fact, it doesn't even do the "tweak every device to let it know we're here" stuff. Fiddle with the thing some, and finally convince it to boot. Gee, the neat little program "scsimount" that you run to read in the parms file off of SCSI: doesn't run. Stare at the wall, reboot a couple of times and try it again. Run GrabMem to turn off fast memory, then run it. Aha! It works. That Xebec would provide something _that_ important that doesn't work with expansion memory does _not_ bode well for the future. The system runs for a couple of hours, with little or no problem. Suddenly, it starts crashing. Regularly. Usually a 68000 trap in high memory. Give it up, unplug the Xebec, and go back to my stock system to work. A call the next day to Xebec produces "huh? We've never seen that before." They also weren't aware that scsimount wouldn't run out of fast memory. A call to RS Data (the people who built my memory expansion) produces: "You should get the C-Ltd system. The new one works like a charm with our hardware, and is _fast_." Also, "Well, it sounds like a power-supply problem. Do to our board, and it might fix it." The Xebec drive system has gone back to the store. I got full credit towards a C-Ltd system, again with the understanding that it _will_ be made to work. >The full SCSI spec supports buss arbitration and disconnect/reconnect. >This allows you to hook up more than one host to more than one drive. >This is actually what I am interested in doing. Apparently, the new C-Ltd system can do some of that. From dealing with SCSI on Suns, there are two things people tend not to do that needs to be done to make things go fast. One is disconnect/reconnect. The other is asynch I/O. The former gives you better throughput in multi-drive or multi-system environments. The latter gives you better throughput in any system (factor of 2!).