carl@p4tustin.UUCP (Carl W. Bergerson) (05/02/91)
garygm@leland.Stanford.EDU (Gary Brainin) writes: > > I have a no-name ("IPC") 386 which currently has DOS 3.3 on it. > Feeling constrained by the 35MB limit on my hard drive . . . Why do you feel constrained by the 33 MByte limit? Do you have any one file that is greater than 33 MBytes? If not, then you ought to use DOS 3.30 to partition your disk into smaller chunks. Why? For efficiency. Let me explain a little. When you access a file, DOS looks at the directory entry and the FAT. The FAT is stored on the first cylinder of the partition and the directory entry and the file itself are stored SOMEWHERE. Now, if you partition your disk into two equal sized partitions and if you spread your applications and files equally between the partitions, the distance from the FAT to SOMEWHERE will be about 1/2 what it would be if you had one humungous partition. Looking at it slightly differently, if you have a disk with a track to track access of 4 ms and an average access of 28 ms, partitioning it into 2 halves would change the effective average access to 16 ms ((28-4)/2)+4 ms. Of course to achieve some substantial portion of this theoretical improvement you need to give some thought as to which partition you place individual applications and files. For example, I make sure my C compiler, libraries, include files, sources, and the tmp directory are all on the same partition, but if I have two physical disks I move the tmp to the second disk. I don't worry about where my editor is, because it is only read from the disk when I start a session and it is never referenced (from disk) again (unless it's some gigantic program with overlays, but I wouldn't use an editor like that for programming). BTW, using the various performance testing programs, such as Norton or Coretest usually will not show a performance improvement because they seem to access the physical information DOS has about the disk not the logical information stored in the partition table. flame off really big FLAME ON type file | more - but that's another story really big FLAME OFF Carl -- Carl Bergerson uunet!p4tustin!carl Point 4 Data Corporation carl@point4.com 15442 Del Amo Avenue Voice: (714) 259 0777 Tustin, CA 92680-6445 Fax: (714) 259 0921
john@jwt.UUCP (John Temples) (05/05/91)
In article <3367@p4tustin.UUCP> carl@p4tustin.UUCP (Carl W. Bergerson) writes: >Now, if you partition your disk into two equal sized partitions and if >you spread your applications and files equally between the partitions, >the distance from the FAT to SOMEWHERE will be about 1/2 what it would be >if you had one humungous partition. >Of course to achieve some substantial portion of this theoretical >improvement you need to give some thought as to which partition >you place individual applications and files. I think that last point should be emphasized. If your file accesses are randomly distrubuted across the two partitions, you'll end up spending more time seeking, not less. You need to organize things so that the file accesses in a given session are mostly from one partition or the other. Otherwise, you waste time seeking across the empty space between the two partitions. I'm not sure that I could break my files up into two logical "chunks," of which I'd only likely to be using one or the other. -- John W. Temples -- john@jwt.UUCP (uunet!jwt!john)
tporczyk@na.excelan.com (Tony Porczyk) (05/08/91)
The News Manager) Nntp-Posting-Host: na Reply-To: tporczyk@na.excelan.com (Tony Porczyk) Organization: Standard Disclaimer References: <1991Apr20.061909.24406@leland.Stanford.EDU> <3367@p4tustin.UUCP> Distribution: na Date: Sat, 4 May 1991 18:14:11 GMT In article <3367@p4tustin.UUCP> carl@p4tustin.UUCP (Carl W. Bergerson) writes: >garygm@leland.Stanford.EDU (Gary Brainin) writes: >> >> I have a no-name ("IPC") 386 which currently has DOS 3.3 on it. >> Feeling constrained by the 35MB limit on my hard drive . . . > >Why do you feel constrained by the 33 MByte limit? Do you have any one file >that is greater than 33 MBytes? If not, then you ought to use DOS 3.30 >to partition your disk into smaller chunks. Why? For efficiency. Since the original poster indicated he has a 386, there is a 100 to 1 possibility that he has some extended memory and has a disk cache installed. At this point your highly theoretical calculations about micro-seconds saved in head movement with 33 Mb partitions are totally irrelevant. Having a 100 or 200 Mb drive with 9 partitions on it is about as efficient as having all your teeth pulled so you will never get a toothache. 33 Mb partition sucks, plain and simple. My advice is, if you can wait for DOS 5.0, do so, if not, go with 4.01 and then upgrade with 5.0 (I hear you will be able to "upgrade" without the need to reformat). Good luck, Tony
sigma@obee.ipl.rpi.edu (Kevin Martin) (05/09/91)
tporczyk@na.excelan.com (Tony Porczyk) writes: >Since the original poster indicated he has a 386, there is a 100 to 1 >possibility that he has some extended memory and has a disk cache >installed. At this point your highly theoretical calculations about >micro-seconds saved in head movement with 33 Mb partitions are totally >irrelevant. Having a 100 or 200 Mb drive with 9 partitions on it is >about as efficient as having all your teeth pulled so you will never >get a toothache. 33 Mb partition sucks, plain and simple. My advice >is, if you can wait for DOS 5.0, do so, if not, go with 4.01 and then >upgrade with 5.0 (I hear you will be able to "upgrade" without the >need to reformat). I don't think you fully understand the benefits of staying with DOS 3.3 until a decent MS-DOS comes out. The 33Mb partitions are IMHO quite handy, both for improved efficiency and organization. I have a 3Mb EMS disk cache, and believe me, there's a world of difference in speed between running PicLab on C: with temporary files on I: and running it on C: with temporary files on C:. It follows logically that if I have a large partition where my data and programs might be widely separated (even though not fragmented), I'll suffer in performance. I have dozens and dozens of packages installed on my machine (a big reason for having a big drive). I obviously can't put everything I might use into my path. But I can run programs from one partition while staying in the current directory of another partition like so: E:\TMP> c:pl286 And that's a fair bit easier than 'c:\graphics\piclab\pl286' I'd say. Staying with DOS 3.3 also renders me virtually immune to a host of glitches introduced with DOS 4.x, such as the mouse problem and the extended memory "support." I also have nearly 600K free with little effort. I know of at least one program which will NOT run on a stripped machine under DOS 4.x. I also don't have to pay to upgrade, not until DOS 5.0, and even when that comes out, it'll have to prove itself after the 4.x fiasco. Newer (not necessarily) == better. -- Kevin Martin sigma@ipl.rpi.edu "Can I kiss one of the bridesmaids instead?"
leonard@qiclab.scn.rain.com (Leonard Erickson) (05/17/91)
Pick up a copy of Compaq MS-DOS 3.31 (or any other 3.31). It supports large partitions. And Utilities support it. If it doesn't do it the same as 4.x it is apparently at least compatible. In any case, I'm running a 40 meg drive with one partition at home (the boot drive is a 40 meg with two partitions, but only because most XT controllers can't boot from a BIGDOS partition). At work, we set have numerous machines with 40, 80, even 120 meg partitions. And we haven't had any trouble with software not liking them. Not even things like Norton uitlities, various disk optimizers, etc. So we have 3.3x *and* we have large partitions. -- Leonard Erickson leonard@qiclab.uucp personal: CIS: [70465,203] 70465.203@compuserve.com business: CIS: [76376,1107] 76376.1107@compuserve.com