Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!zaphod.mps.ohio-state.edu!think.com!masscomp!peora!tarpit!bilver!bill From: bill@bilver.uucp (Bill Vermillion) Newsgroups: comp.unix.sysv386 Subject: Re: Help with ESIX ESDI No-translation Mode Message-ID: <1990Dec26.082310.3737@bilver.uucp> Date: 26 Dec 90 08:23:10 GMT References: <94408976@bfmny0.BFM.COM> <1990Dec21.055405.1053@bilver.uucp> <2733@sixhub.UUCP> Distribution: usa Organization: W. J. Vermillion - Winter Park, FL Lines: 82 In article <2733@sixhub.UUCP> davidsen@sixhub.UUCP (bill davidsen) writes: >In article <1990Dec21.055405.1053@bilver.uucp> bill@bilver.uucp (Bill Vermillion) writes: > >| I have been running in translation mode from day one. The reason it was >| put there is so that you can run bigger drives and not give anything up. >| Why do you feel you have to run it WITHOUT translation. > >| Am I missing something here? Why not use the translation? > > The WD will not run a large drive in translation mode. Even with >translation set to, say, 64spt, you get 1024*64*16, or 524288k, (512MB). >Some of us lucky people are running drive larger than that. You also >lose performance (at least by my measurements) by using translation. Now that I go back and look at my notes, I must not have been running in translation mode, but running in normal mode. (Sorry for that, my mind has been wandering off and leaving me behind :-) . But what is the problem in non-translation mode that space is being thrown away (according to Tom Neff he can't use more thatn 512 megs). Here is a df -v on my drive as it is running while typing in this message. Mount Dir Filesystem blocks used free %used / /dev/dsk/0s1 167070 45658 121412 27% /usr /dev/dsk/0s3 409500 281184 128316 68% /usr2 /dev/dsk/0s4 571970 197162 374808 34% _______ Total Block 1148540 * 512 bytes/ block = 588 megs. Add in my 20 meg swap space, and 30 meg dos partition, and I come up with 638 megs. This is also a refurbed drive and I had about 350 bad tracks, and I used alternate sector translation, so I threw away one sector per track per head so that got rid of some more. The drive is a Maxtor 8760E, ostensibly 660 megs, running a WD1007V-SE2 (I think that's the correct number, the one with the floppy controller on it). Jumpers 11/12 are set for alternate sector mapping. That throws away about 12 Megs of drive space, but its sure give me much more than the 512 Meg Tom said he was limited to. (The Subject line asked for help in ESDI non-tranlation mode under ESIX). Tom emailed me on this last August when I brought the system up, and I didn't know there were problems. OS is Esix Sys V.3.2. AMI bios, C&T chipset. 25Mhz '386. > It's not obvious unless you are running a large fast disk (and take >the time to benchmark) that this is the case, so I can appreciate the >question, hope the answer is clear. I see it now on translation. But what is there to benchmark. I'm more confused than ever. The system is running a full news feed, and /usr2 is where news lives. (The great amount of space on that is because the main feed has been sick the last couple of days. I got 23 news articles on Sunday, and 730 on Monday. Saturday was full, and Friday was light. That partition normally runs about 65% full. > For the original poster, I don't remember doing anything fancy, I just >turned off the translation mode, formatted the drive, and when unix >asked me if the parameters were correct I told it no and put in the real >values. This seems to work for all SCO, ESIX D, and Dell or Intel V.4. >We were unable to make it run with Dell or ISC 3.2, although both of >them assurred us that we could just ignore the error messages about BIOS >limitations, phase of the moon, etc. The only problem I had was the jumpers on the drive were not right, and Maxtor gave us the proper jumper settings, but did not tell us what the jumpers did. (In the heat of getting things going I forgot to write down the settings before I installed the drive, and they can't be seen unless the drive comes out, and that is a pain in this cabinet). > My experience: after two weeks of work by people who have been doing >UNIX since V7, we could not make badtrack check anything past 1023, and >dropped ISC/Dell from evaluation. Thanks for the warning on that one. -- Bill Vermillion - UUCP: uunet!tarpit!bilver!bill : bill@bilver.UUCP