dave@kbvan.com (Dave Van Allen) (03/14/91)
I hope I'm posting to the right place for this, correct me if I'm wrong. We are trying to re-install unix on an att 3b1 and cannot find or figure out the hard drive type. The front plate on the system says it has a 67Meg drive, but there is no 67 meg drive in the type table. The nearest maych is 64meg but we are worried that this type may not have the correct number of heads cyl's etc. Any help would be greatly appreciated. If nobody else is interested in this please email me uunet!kbvan!dave Thanx! Dave Van Allen
thad@public.BTR.COM (Thaddeus P. Floryan) (03/14/91)
In article <107@kbvan.com> dave@kbvan.com (Dave Van Allen) writes: >I hope I'm posting to the right place for this, correct me if >I'm wrong. >[...] >The front plate on the system says it has a 67Meg drive, but there >is no 67 meg drive in the type table. The nearest maych is 64meg but >we are worried that this type may not have the correct number of heads >cyl's etc. comp.sys.3b1 is the correct newsgroup (and, still, for awhile, unix-pc.*) The 64MB drive selection is the correct one due to how spare sectors are allocated on the drive. The largest HD on an unmodified 3B1 has 8 heads and 1024 cylinders, uses MFM recording, and has 17 sectors per track; you should pick the entry with these numbers, or select "other" and simply enter these numbers at each prompt. If you're truly starting from scratch, you may wish to consider some mods that permit up to two 150MB HDs. Again, browse comp.sys.3b1 Thad Floryan [ thad@btr.com (OR) {decwrl, mips, fernwood}!btr!thad ]
ostroff@Oswego.EDU (Boyd Ostroff) (03/20/91)
In article <2130@public.BTR.COM> thad@public.BTR.COM (Thaddeus P. Floryan) writes: >In article <107@kbvan.com> dave@kbvan.com (Dave Van Allen) writes: >>The front plate on the system says it has a 67Meg drive, but there >>is no 67 meg drive in the type table. The nearest maych is 64meg but > >The 64MB drive selection is the correct one due to how spare sectors are >allocated on the drive. Perhaps this is nit-picking, but I think you're wrong. The 3B1 actually does only have a 64MB drive. The discrepancy comes from the definition of a "megabyte". Recall that one kilobyte (K) is actually 1024 bytes, not 1000. A megabyte is 1024 K = 1,048,576 bytes, NOT 1,000,000 bytes. Therefore, 64 megabytes actually equals 67,108,864 bytes - this is where the 67MB figure comes from. Disk and computer companies like the use the most "optimistic" way of computing the hard drive capacities :-) The spare sector concept is correct, but the numbers are wrong. Due to the fact that the 3B1's disk controller can only handle a certain number of sectors (I forget the details), there are "spares" on each track which are used for mapping bad blocks. The full, formatted capacity of the Miniscribe drive is actually 71MB, but the unix-pc can only utilize 67MB. Of course, you're going to lose another 4 or 5 MB of your disk space to a swap partition, so you'll probably only have 62 MB free once the drive is formatted. == Boyd Ostroff, Tech Director == Sys Admin, "The CallBoard" == |||| Dept of Theatre, SUNY Oswego |||| Serving the Performing Arts |||| |||| (315) 341-2987 |||| 315-947-6414 1200/2400 baud 8N1 |||| == ostroff@oswego.oswego.edu == cboard!sysop@oswego.oswego.edu ==
tom@afthree.as.arizona.edu (Thomas J. Trebisky) (03/21/91)
In article <1991Mar19.180851.27460@oswego.Oswego.EDU> ostroff@oswego.Oswego.EDU (Boyd Ostroff) writes: > >The spare sector concept is correct, but the numbers are wrong. Due to the >fact that the 3B1's disk controller can only handle a certain number of >sectors (I forget the details), there are "spares" on each track which are >used for mapping bad blocks. The full, formatted capacity of the Miniscribe >drive is actually 71MB, but the unix-pc can only utilize 67MB. Actually the disk controller (hardware) is perfectly happy with 17 sectors per track. The 16 sector per track with sector 17 reserved as a spare is purely a software convention handled (presumably/hopefully) in the kernel device driver. The main motivation is to avoid long (and thus slow) seeks going to fetch replacement blocks when bad sectors are encountered. I took exactly the opposite approach writing my device driver for the miniframe. I use the last cylinder of the disk for spare blocks and suffer with the long seeks (but my typical 40M disk only has maybe 6 or so bad blocks at this time). This gives me 2M of additional capacity -- but the other trade off is that 17 is an ugly number -- filesystem blocks have to wrap around from one track to the next (I am using a 4K/512 BSD filesystem). You pay your money and you take your lumps -- I opted to loose some speed and gain capacity. However in the swap area, I will treat the disk as if it had 16 s/t and ignore all the last sectors on each track -- speed is of the essence there. -- Tom Trebisky ttrebisky@as.arizona.edu (Internet) Steward Observatory University of Arizona Tucson, Arizona