• Please review our updated Terms and Rules here

Best disk interleave?

I've asked question about interleave some time ago.

It depends on computer, Spinrite will tell you wich is best for 5150.
I think, that SCSI drives should have interleave=1 for best performance, but I might be wrong, especially if it's slow 5150.
 
I can't get SpinRite to work, it gives memory error!

With just 256kB it says it is not enough memory.

With the QuadRam board, it says "bad memory" and it exits to DOS

With only another memory expansion it freezes at one point when it tests the memory.

It also happens at the same point on my XT (640kB mainboard).

I won't know why it does that, but I tested the memory in CheckIt and it appears to be fine? :cry:
 
Aww, yes it do freez on my XT too, I runned it on 8086.

There should be similar utility somewhere around.

Also you can format disk with interleave 1, 2, 3, etc and find the fastest one, but it's time consuming.
To check speed use CheckIt, remmember to put memory board, I think it needs at least 512 kB to run.
 
I've now tried both Spinrite and Norton Calibrate, but neither works. CALIBRAT.EXE simply says it cannot detect the optimal interleave.

Any other software to try?
 
There is a program called Optune that you might be able to find an old copy of which I use set the optimum interleave for any drive/controller combination that works on just about anything.

Nice thing about it is that, if the drive already has an operating system and other software on it, it will optimize the interleave by moving the software, optimizing the track, putting the software back and moving on to the next track.
 
dongfeng said:
My 5150 has a SCSI hardcard fitted with a Seagate ST325N 20MB drive. I wish to low-level format this drive, but what would the recommended interleave be?

http://chemeng.p.lodz.pl/zylla/seagate/scsi/st325n.txt

I have currently set the interleave as 3. The controller also detects the drive parameters automatically, which is neat!
Click to expand...

It 80% depends on the speed of the controller card and 20% on the speed of the host system. Spinrite or Norton Utilities' CALIBRAT should be able to determine what the optimal interleave is. Spinrite is a wicked cool program so use that if at all possible.

The wrong interleave on an MFM/RLL disk can slow things down by an order of magnitude so it's worth it to get it right. SCSI, I've never tried -- I'm not sure it's possible.

It is nearly useless to set the interleave on a *different* system unless that system is identical to the one it will finally reside in. For example, my XT usually has a 4:1 interleave for best performance... that same hardware in a 286 nets a 2:1 interleave. So always test/set the interleave on the same system.
 
dongfeng said:
I can't get SpinRite to work, it gives memory error!

With just 256kB it says it is not enough memory.

With the QuadRam board, it says "bad memory" and it exits to DOS

With only another memory expansion it freezes at one point when it tests the memory.

It also happens at the same point on my XT (640kB mainboard).

I won't know why it does that, but I tested the memory in CheckIt and it appears to be fine? :cry:
Click to expand...

Spinrite requires 512K RAM to operate, I think (could be wrong).

Spinrite 5.0 doesn't work with some DOS versions (it hangs during the memory test). I believe it works with MS-DOS 6.22. I contacted Steve Gibson about this and he claims he's simply asking DOS for a free block of RAM and it gives him something starting in ROM (FFFF:0000) but, personally, I think he's got a bug in the program somewhere...
 
Trixter said:
It 80% depends on the speed of the controller card and 20% on the speed of the host system. Spinrite or Norton Utilities' CALIBRAT should be able to determine what the optimal interleave is. Spinrite is a wicked cool program so use that if at all possible.

The wrong interleave on an MFM/RLL disk can slow things down by an order of magnitude so it's worth it to get it right. SCSI, I've never tried -- I'm not sure it's possible.

It is nearly useless to set the interleave on a *different* system unless that system is identical to the one it will finally reside in. For example, my XT usually has a 4:1 interleave for best performance... that same hardware in a 286 nets a 2:1 interleave. So always test/set the interleave on the same system.
Click to expand...

Hi All,

My apologies for bumping the thread. I thought it was better than starting a new one on the same subject. I was thinking about how in the past I've used Norton Utilities (IIRC) to change the interleave on the fixed disk drive on an IBM PC/XT. Again, IIRC, IBM formated it with a 3:1 interleave and Norton said it would speed up with a 2:1 interleave.

That all seems well and good if the goal is to speed up disk throughput. But what if your idea of increasing performance is to increase the drive's longevity? I would think that doing disk-intensive operations like backing up the disk might cause the stepper motor (which moves the heads of the disk to the correct cylinder) to heat up enough that it could shorten the motor's life.

I just read today in my new PC Technical Refernce that the motors in the fixed disk drive are thermally isolated from the rest of it, which allows disk reads without a warm-up period. BTW, the reason the PC Tech Ref manual has info on the fixed disk is because you could hook an expansion unit with one up to the PC. The manual infers that you can't put a fixed disk in the 5150 (which is because the power supply would be overtaxed).

Anyway, I was thinking that if I ever do get another early PC with a hard drive (note my user name), I plan to set the interleave to the factory value, for the purpose of preserving the computer for as long as possible.
 
SCSI drives are a special case. Most PC BIOSes for SCSI do not implement low-level format. You need a special program to perform SCSI LL formats.

I would not expect any commodity interleave optimizer to work with a SCSI drive. Programs such as SpinRite are more suited to IDE, MFM/RLL and ESDI drives.

In addition, CHS geometries for SCSI drives are purely the invention of the BIOS--the SCSI protocol itself makes no provision for CHS geometry--the first sector on the disk is 0 and they're numbered consecutively to the last sector. Any program that attempts to determine interleave by making assumptions about the geometry is going to be wrong.
 
Last edited:
Floppies_only said:
Hi All,

My apologies for bumping the thread. I thought it was better than starting a new one on the same subject. I was thinking about how in the past I've used Norton Utilities (IIRC) to change the interleave on the fixed disk drive on an IBM PC/XT. Again, IIRC, IBM formated it with a 3:1 interleave and Norton said it would speed up with a 2:1 interleave.

That all seems well and good if the goal is to speed up disk throughput. But what if your idea of increasing performance is to increase the drive's longevity? I would think that doing disk-intensive operations like backing up the disk might cause the stepper motor (which moves the heads of the disk to the correct cylinder) to heat up enough that it could shorten the motor's life.

I just read today in my new PC Technical Refernce that the motors in the fixed disk drive are thermally isolated from the rest of it, which allows disk reads without a warm-up period. BTW, the reason the PC Tech Ref manual has info on the fixed disk is because you could hook an expansion unit with one up to the PC. The manual infers that you can't put a fixed disk in the 5150 (which is because the power supply would be overtaxed).

Anyway, I was thinking that if I ever do get another early PC with a hard drive (note my user name), I plan to set the interleave to the factory value, for the purpose of preserving the computer for as long as possible.
Click to expand...

The interleave setting won't effect drive longevity at all. It has no effect on the stepper motor either. The interleave is how data is written to a track, if the interleave is too small, the system won't be able to read the data in one pass and will have to wait for it to come around again. If the interleave is too large, the system will be waiting unnecessarily long for the data to be positioned under the head to be read.
 
strollin said:
If the interleave is too large, the system will be waiting unnecessarily long for the data to be positioned under the head to be read.
Click to expand...

I thought that that would result in a lower stepper motor current over time, resulting in it heating up less. Think about it: if the disk has to wait longer for the right tracks to be under the head, the head will be the part doing the waiting, allowing the stepper motor to cool down more.

Sean
 
Floppies_only said:
I thought that that would result in a lower stepper motor current over time, resulting in it heating up less. Think about it: if the disk has to wait longer for the right tracks to be under the head, the head will be the part doing the waiting, allowing the stepper motor to cool down more.
Click to expand...

Sure, so drives could be formatted at a 1:1 interleave on an XT, meaning that it takes 17 revs to read one track? I submit that unless you're using an old ST506 (unbuffered seek) drive, that any time lost in bad interleave calculation has a negligible effect on positioner operating temperature. Later drives (20 MB and up) usually use a "ballistic" algorithm when seeking, making it hard to accurately determine the positioner current.

Back to the original question--how will the OP change the interleave on his SCSI drive?
 
Chuck(G) said:
Sure, so drives could be formatted at a 1:1 interleave on an XT, meaning that it takes 17 revs to read one track? I submit that unless you're using an old ST506 (unbuffered seek) drive, that any time lost in bad interleave calculation has a negligible effect on positioner operating temperature. Later drives (20 MB and up) usually use a "ballistic" algorithm when seeking, making it hard to accurately determine the positioner current.

Back to the original question--how will the OP change the interleave on his SCSI drive?
Click to expand...

I thought the part about low level formatting a SCSI drive was covered elsewhere in this thread. I was just talking about the MFM drives that came with the original IBM PC/XT, because that is what I've had experience with.

I never meant that a 1:1 interleave was desired. My whole idea was that IBM engineers might have had a good reason for setting the interleave to 2:1 instead of 3:1. I just can't imaging them not being smart enough to set the interleave at the mix that would allow the fastest rate of data transfer. Two to One seemed fast enough for me, anyway.

Sean

P.S.: Has anybody been following auctions for Windows version one that were on ebay recently? There were two 1.01 and on 1.04. The first one to close went for over seven hundred dollars, the last one to close went for between two and three hundred, to a bidder who bid over six hundred for the first one.

Windows seems to hold it's value.
 
You must log in or register to reply here.
Back
Top