rosen@mtgzz.UUCP (t.rosenfeld) (08/05/87)
There has been some talk recently in this newsgroup about interleave
factors. Someone said a factor of 1:4 is recomended for the 6300, and is
much faster than 1:3 which may be some default.
My question is how can I tell what my current factor is? I have Norton
Utilities, will that help. I suppose I may also want to know how to change it
if I have to. I suppose this requires a low level format, but I'll worry about
that if and when I find out my factor is wrong.
Oh, by the way. Is 1:4 also recomended for the PLUS? Do I use the same
technique on the PLUS to find out what it is?
Thanks
--
Tom Rosenfeld @ AT&T Information Systems Labs, Middletown, NJ
(201) 957-5867
UUCP: {harpo,ihnp4,burl,akgua}!mtgzz!rosen
Disclaimer: I don't claim anything.kad@ttrdc.UUCP (Keith Drescher) (08/06/87)
In article <2928@mtgzz.UUCP> rosen@mtgzz.UUCP (t.rosenfeld) writes: >There has been some talk recently in this newsgroup about interleave >factors. Someone said a factor of 1:4 is recomended for the 6300, and is >much faster than 1:3 which may be some default. > >My question is how can I tell what my current factor is? I have Norton >Utilities, will that help. I suppose I may also want to know how to change it >if I have to. I suppose this requires a low level format, but I'll worry about >that if and when I find out my factor is wrong. I found that I had the wrong interleave factor on my pc6300 by running the Norton SI program. I got a hard disk performance ratio relative to the XT of 0.5, which was unreasonable. I was using a WD controller and a 10M hard disk. I kept low level formating (which allows choosing an interleave factor), running fdisk, and running the api format. Then running Norton's SI until I got 1.0 performance ration to the XT. At any rate, run SI, if your performance is < 1.0, your interleave is probably wrong (unless you've got a really slow hard disk/controller combination). 0.5 is definately wrong. Keep trying interleave factors until you get >=1. The reasonable values for most 10 and 20 Meg disks (that I've seen) are 3, 4, 5, 6. > >Oh, by the way. Is 1:4 also recomended for the PLUS? Do I use the same >technique on the PLUS to find out what it is? > > -- Keith Drescher (kad@ttrdc) | ... You can check out any AT&T | time you like - but you can Computer Systems Division, Skokie, Il. | never leave ... PATH: ...!ihnp4!ttrdc!kad | - Hotel California
crs@cpsc6b.cpsc6a.att.com (C. R. Seaman) (08/10/87)
In article <1844@ttrdc.UUCP>, kad@ttrdc.UUCP (Keith Drescher) writes: <In article <2928@mtgzz.UUCP> rosen@mtgzz.UUCP (t.rosenfeld) writes: <<There has been some talk recently in this newsgroup about interleave <<factors. Someone said a factor of 1:4 is recomended for the 6300, and is <<much faster than 1:3 which may be some default. << <<My question is how can I tell what my current factor is? The default interleave factor for hard disks varies from manufacturer to manufacturer, and I don't know of any sure way to find it other than to read the manual for the specific hard disk you are using, or perhaps to contact the manufacturer. Generally, they will set the interleave to a "safe" value, which is usually somewhat slow. <<I suppose I may also want to know how to change it <<if I have to. I suppose this requires a low level format, Right. <I found that I had the wrong interleave factor <on my pc6300 by running the Norton SI program. There is also a PD (or is it shareware?) program on most BBS's, called "coretest", which allows you to see the average seek time, track to track seek time, and data transfer rate for a given hard disk. This would be the best test. <Keep trying interleave factors until you get >=1. The reasonable values <for most 10 and 20 Meg disks (that I've seen) are 3, 4, 5, 6. <<Oh, by the way. Is 1:4 also recomended for the PLUS? Do I use the same <<technique on the PLUS to find out what it is? No. The interleave values for a hard disk are based on the performance of the disk, not the system. It basically determines the number of cylinders the head must pass after reading from/writing to a cylinder before it can read from/write to another cylinder. If the drive takes 4 cylinders to reset, and you set the interleave to 3, the head will wait for almost a complete revolution before attempting another read/write. Setting the value to 5 would mean that the drive is basically idle for one cylinder pass. I use a DTC controller with a Shugart 712 (10 Meg.) drive, and the optimum interleave factor is 5. <-- <Keith Drescher (kad@ttrdc) -- Chris Seaman | o\ /o crs@cpsc6a.att.com <or> | || See "Attack of the Killer Smiley"! ..!ihnp4!cpsc6a!crs | \vvvvvv/ Coming Soon to a newsgroup near you! | \____/
ching@amd.AMD.COM (Mike Ching) (08/11/87)
In article <300@cpsc6b.cpsc6a.att.com> crs@cpsc6b.cpsc6a.att.com (C. R. Seaman) writes: > The interleave values for a hard disk are based on the performance of >the disk, not the system. It basically determines the number of cylinders >the head must pass after reading from/writing to a cylinder before it can >read from/write to another cylinder. If the drive takes 4 cylinders to >reset, and you set the interleave to 3, the head will wait for almost a >complete revolution before attempting another read/write. Setting the value >to 5 would mean that the drive is basically idle for one cylinder pass. > NO. The interleave for a hard disk is based on the performance of the system and disk controller, not the disk. It is the number of SECTORS the head passes while the system processes the data (either creating another sector data buffer to write or moving the read buffer to system memory). If the head passes the *next* sector before the system is ready, it requires a whole revolution before the sector passes under the head again. Track-to-track interleave is the number of sectors that pass under the head as the head moves from one cylinder to the next (so that sector 0 is not adjacent to sector 0 of the next cylinder) and is based on disk performance but is rarely what is being discussed when interleave is the topic. mike ching
rosen@mtgzz.UUCP (t.rosenfeld) (08/11/87)
In article <300@cpsc6b.cpsc6a.att.com>, crs@cpsc6b.cpsc6a.att.com (C. R. Seaman) writes: [ .... ] > No. The interleave values for a hard disk are based on the performance of > the disk, not the system. It basically determines the number of cylinders > the head must pass after reading from/writing to a cylinder before it can > read from/write to another cylinder. If the drive takes 4 cylinders to > reset, and you set the interleave to 3, the head will wait for almost a > complete revolution before attempting another read/write. Setting the value > to 5 would mean that the drive is basically idle for one cylinder pass. > > Chris Seaman | o\ /o I take it Chris meant "sectors" not "cylinders". -- Tom Rosenfeld @ AT&T Information Systems Labs, Middletown, NJ (201) 957-5867 UUCP: {harpo,ihnp4,burl,akgua}!mtgzz!rosen Disclaimer: I don't claim anything.
timothym@tekigm2.TEK.COM (Timothy D Margeson) (08/11/87)
Hi,
OKAY FOLKS - Interleave of hard disk systems is to compensate for inefficient
drive controllers AND computer bus structures (primarily DMA channels).
Disks themselves DO NOT and CANNOT effect what interleave is best for a given
system (except in extreme cases where a disk does not meet some standards such
as ST506 or ST412 or SCSI, but is still used in systems using those standards).
So, forget about what disk drive you have when considering interleaves, more
important is the controller and bus structure and related timimgs.
For example, using the same disk drive, but two different controllers on a
Compaq Deskpro (8086), I have seen interleaves of 4:1 and 2:1 as optimums for
the different controllers.
Even on the same controller with a different BIOS, I have seen a difference
of 4:1 and 2:1 giving the fastest transfer rates.
Finally, for a given system, the best way to find what interleave is best is
trial and error. Start by formatting the disk with an interleave of 1:1, and
work your way up to 6:1. Somewhere in between you will find an interleave that
gives the best transfer rates.
--
Tim Margeson (206)253-5240
PO Box 3500 d/s C1-937 @@ 'Who said that?'
Vancouver, WA. 98668
{amd..hplabs}cae780!tektronix!tekigm2!timothym (this changes daily)enchant@oliveb.UUCP (Dan Crocker) (08/11/87)
> The default interleave factor for hard disks varies from manufacturer to > manufacturer, and I don't know of any sure way to find it other than to > read the manual for the specific hard disk you are using, or perhaps to > contact the manufacturer. For those programmers out there, it shouldn't be too hard to write a program to calculate interleave. The following should be the basic algorithm: 1. Issue an INT 13 to read sector #1 2. Start a timer 3. Issue an INT 13 to read sector #2 4. Stop timer The results of this sequence can be interpreted in the program or manually. Basically, there are two possible results: a. The number is less than 16.7 ms If this is the case, then the time is your interleave. This is due to the fact that each sector on a 17 sector track is about 1ms long (16.7/17). For an interleave of 3:1, it will take about 2ms to arrive at the beginning of sector 2 and 1ms to read it. There are extra delays due to BIOS overhead and data transfer to the system, but these cancel each other out when the timer is started and stopped. This idea can be expanded to handle other track sizes such as RLL (26 sec/trk) or ESDI (34,35,36), although the numbers are not as nice. b. The number is greater than 16.7ms This means that you missed a revolution between reading sector #1 and #2. The case above (a) can be arrived at by simply subtracting 16.7. Then you can proceed as before. This result is also an indication that your interleave is too small. You should probably increase it by one until you get a number that is less than 16.7ms. All of the above is based on the assumption that there are no problems with reading ID's or errors in the data field. Soft (random) errors in either case could give strange results. I hope this is helpful to someone. Dan -- Here comes the supernatural anesthetist If he wants you to snuff it All he has to do is puff it He's such a fine dancer
enchant@oliveb.UUCP (Dan Crocker) (08/11/87)
> No. The interleave values for a hard disk are based on the performance of > the disk, not the system. It basically determines the number of cylinders > the head must pass after reading from/writing to a cylinder before it can > read from/write to another cylinder. If the drive takes 4 cylinders to > reset, and you set the interleave to 3, the head will wait for almost a > complete revolution before attempting another read/write. Setting the value > to 5 would mean that the drive is basically idle for one cylinder pass. I suspect you meant sector and not cylinder in the above explanation. To clarify, the interleave factor is dependant on both the system and controller performance. A proper interleave will allow enough time for a sector to be read from the disk and its data to be transferred to the system BEFORE the next logical sector passes by the head. The time it takes to read the data from the disk is solely dependant on the rotation speed of the disk. The interleave allows for: delay of controller in notifying system of data available, time required by system to transfer data, and time required by controller to become ready to look for the next sector or receive a new command. Therefore, moving the same disk/controller combination from an 8088 based system to a 386 based system could have a drastic effect on the allowable interleave. Obviously, other system performance factors can have a similar impact. Two additional points: -The required interleave for single-sector accesses can be greater than that for multi-sector accesses. It depends on how much time is taken between sector requests. -Obviously, interleave is only important when you are reading logically consecutive sectors. In other words, if most of your files on your disk are fragmented, then changing the interleave will not show such a drastic improvment. Dan -- Here comes the supernatural anesthetist If he wants you to snuff it All he has to do is puff it He's such a fine dancer
tes@whuts.UUCP (STERKEL) (08/12/87)
In article <2999@oliveb.UUCP>, enchant@oliveb.UUCP writes: > > The default interleave factor for hard disks varies from manufacturer to > > manufacturer, and I don't know of any sure way to find it other than to > > read the manual for the specific hard disk you are using, or perhaps to > > contact the manufacturer. > > > For those programmers out there, it shouldn't be too hard to write a program > to calculate interleave. The following should be the basic algorithm: OR... you could simply buy the HFORMAT *set* of utilities from Mace Inc. Included is interleaving timers, a program for telling you the *optimum* interleaving factor given your disk/head/controller/buffers/DMA/CPU/clock combination. And low and high level formatters to implement this optimum factor.
crs@cpsc6b.cpsc6a.att.com (C. R. Seaman) (08/12/87)
In my previous posting about interleave factors, I somehow typed in
'cylinders' when I *REALLY* meant to say 'sectors'. Well, what can I say...
I uh... The uh... The Sun was in my eyes! Yeah, that's the ticket!
--
Chris Seaman | o\ /o
crs@cpsc6a.att.com <or> | || See "Attack of the Killer Smiley"!
..!ihnp4!cpsc6a!crs | \vvvvvv/ Coming Soon to a newsgroup near you!
| \____/ pgf@mtung.ATT.COM (Paul Fox) (08/13/87)
In article <300@cpsc6b.cpsc6a.att.com> (C. R. Seaman) writes: >No. The interleave values for a hard disk are based on the performance of >the disk, not the system. It basically determines the number of cylinders >the head must pass after reading from/writing to a cylinder before it can >read from/write to another cylinder. If the drive takes 4 cylinders to >reset, and you set the interleave to 3, the head will wait for almost a >complete revolution before attempting another read/write. Setting the value >to 5 would mean that the drive is basically idle for one cylinder pass. [ You should substitute "sector" for "cylinder" in the above, but that's not my main point. ] Are you sure about that? I think you've just defined the minimum interleave, all right, but don't forget the software overhead of setting up for DMA transfers, managing buffers, application overhead in non- read-ahead systems, etc. Those things all add to interleave needs as well. It's actually a very tough thing to get right on a given system -- most UNIX boxes have low-level formatted interleaves, as well as an interleave built into the ordering of the [initial] freelist... the two interleaves can actually work against one another if you're not careful. -- Paul Fox, AT&T Information Systems, Middletown NJ. [ihnp4|vax135]!mtung!pgf (201)957-2698
tes@whuts.UUCP (STERKEL) (08/14/87)
In article <300@cpsc6b.cpsc6a.att.com>, crs@cpsc6b.cpsc6a.att.com (C. R. Seaman) writes: > In article <1844@ttrdc.UUCP>, kad@ttrdc.UUCP (Keith Drescher) writes: > <In article <2928@mtgzz.UUCP> rosen@mtgzz.UUCP (t.rosenfeld) writes: > <<There has been some talk recently in this newsgroup about interleave > <<factors. Someone said a factor of 1:4 is recomended for the 6300, and is > <<much faster than 1:3 which may be some default. > << > <<My question is how can I tell what my current factor is? > > The default interleave factor for hard disks varies from manufacturer to > manufacturer, and I don't know of any sure way to find it other than to > read the manual for the specific hard disk you are using, or perhaps to > There is also a PD (or is it shareware?) program on most BBS's, called > "coretest", which allows you to see the average seek time, track to track > seek time, and data transfer rate for a given hard disk. This would be > the best test. "coretest" is a proprietary Package of Core Technologies, Inc. It is rather exciting to watch, and hear, because it makes your hard drive sound like a washing machine, all while painting "real-time" performance bar charts. It does not work on all controller/drive combinations, please do not ask which ones, just an observation, never did a well-constructed test. (I never saw any restrictions on the use of coretest, nor shareware-type messages, but then I was not interested enough in the program to look that hard) Two side notes: 1. No, I do not have the program any longer, do not ask for it. 2. When you download, use a *very* responsible source, the ease of turning this program into a trojan is self-evident.
epb19@uxe.cso.uiuc.edu (08/15/87)
When the interleave factor on a drive is default 5, use 7 on a 6300. You will notice the improvement immediately. Therefore if 3 is normal, 4 should do for the 6300. This is probably even truer for the plus. The reason is the accelerated clock speed on the motherboard, which is not affecting the speed of the onboard drive clock, which is tied to rpm.
enchant@oliveb.UUCP (Dan Crocker) (08/18/87)
In article <45900066@uxe.cso.uiuc.edu>, epb19@uxe.cso.uiuc.edu writes: > > When the interleave factor on a drive is default 5, use 7 on a 6300. > You will notice the improvement immediately. Therefore if 3 is normal, > 4 should do for the 6300. This is probably even truer for the plus. > The reason is the accelerated clock speed on the motherboard, which > is not affecting the speed of the onboard drive clock, which is tied > to rpm. What????? -- Here comes the supernatural anesthetist If he wants you to snuff it All he has to do is puff it He's such a fine dancer
timothym@tekigm2.TEK.COM (Timothy D Margeson) (08/19/87)
In article <45900066@uxe.cso.uiuc.edu> epb19@uxe.cso.uiuc.edu writes: >When the interleave factor on a drive is default 5, use 7 on a 6300. >You will notice the improvement immediately. Therefore if 3 is normal, >4 should do for the 6300. This is probably even truer for the plus. >The reason is the accelerated clock speed on the motherboard, which >is not affecting the speed of the onboard drive clock, which is tied >to rpm. I would laugh, but this person thinks they know something about computers! You folks out there writing this tripe, be serious. If you don't know what you are talking about, please refrain from posting misinformation that may be taken seriously by some innocent bystander. For anybody who might have been taken in by the above posting, it is tripe, and not an once of it makes any sense, especially the part about the 6300+. -- Tim Margeson (206)253-5240 PO Box 3500 d/s C1-937 @@ 'Who said that?' Vancouver, WA. 98668 {amd..hplabs}cae780!tektronix!tekigm2!timothym (this changes daily)