False Premises, Misconceptions

[ParanoidLittleMan]

I saw during last couple decades, and even in last months over and over again some simply wrong assumptions, claims related with Atari ST, TOS mass storage, hard disks or now rather Flash card based media. So, will write here some facts, some ideas why wrong claims became spread, explanations ...

 

"Max size of first partition (C) is 32 MB, or by some even only 16 MB."

TOS has same size limit for all partitions C-P . And if someone don't know it yet:  512 MB by TOS 1.04-4.02 . 256 MB by TOS 1.00 and 1.02 .

Why this misinfo. ? Because some (probably most of early hard disk drivers) had such limits. And the reason for limit was not TOS, but way how those (autoboot) drivers loaded self - loading is not possible as regular file load, since that can happen from hard disk only when hard disk driver is operational, so load of usually some 10-40 KB code must be done with code in MBR, where space is limited - only some 390 bytes. And that needs to load possibly fragmented file HDDRIVER.SYS from root of C . Well, that's just very hard. So, they needed to limit size of partition C to save some bytes in MBR, so that code fit there.

Why to care about fragmentation at all ? I guess because possible updates, which go on then not empty C . So, that can happen.

But ... there is better solution, what costs only some + 10-20 KB -  add  so much to first version of HDRIVER.SYS, and ask users in manual that copy/install it as first on C - then will be not fragmented for sure. And possibly longer  updated versions will fit in that extended space, so will be not fragmented.

Here to say, that I used completely different place for that SYS, where no fragmentation can happen, and code is very short. Update too - in time.

Surely, in early years of ST 32 MB partition did not sound bad on drives of 80-160 MB capacity. Today is different, especially when there is SW what needs to be on C .

 

"It is good to have less partitions of larger sizes instead more of smaller sizes"

Yeah, some think that more partitions use more RAM, cause slower work. But truth is exactly opposite. Plus, disk space usage is more efficient with smaller partitions.

So, first count of partitions does not affect speed. May need very little more RAM (like +250 bytes more), but that depends on driver SW. What needs more RAM is actually larger partition size - if only 1 is such, driver must set larger so called BCB buffers. And that's 10-s of KB-es . Still not so bad.  But, larger buffers mean slower work. Especially with TOS-es so called large logical sectors.

Probably most interesting is efficiency of space usage. I will not go here in deep details of FAT16, cluster sizes ... Who is interested can look online about - several  of sites, pages deal with it.  Shortly: FAT16 means max 2 POW 16 allocation units in 1 partition - 65534 (2 less actually) . And 1 file, 1 DIR will allocate min 1 unit, or if is longer by only 1 byte 2 units, and so on ... In practice it means that if file is 133 bytes, it will take 16 KB on TOS partition of 260 MB. Because allocation unit (cluster) size is 16 KB then.  Wait ... cluster size on partition above 256 MB should be 8 KB, not 16 - because 512MB/65536 is 8 KB. True, but unfortunately TOS partitions have actually 15-bit FAT and not 16. In other words, max cluster count is 32766.  In case of TOS 1.00-1.02 it is 16382 only - even less efficient (and reason for smaller max part. size). 

In real usage all above means that used space on disk can be much more than sum of file sizes on it. In case of lot of short files much more - like 30% .

With smaller partitions that will be less. So, advice to people with smaller capacity medias: more partitions of smaller size, and put those with lot of short files on smaller partitions.

Everyone can check is it correct what I wrote here:  in Desktop select hard disk icon (like C-P) , then File, Show Info .  There is "Bytes used:" line - and that's misleading, wrong - that's actually sum of sizes of all files. But DIR-s use 'bytes' on drive too - create new folder and will see that value of 'Bytes used' remained same. And value can be odd - unlike real used space . 

There is easy way to see really used space - write down what is shown as "Bytes available" when partitions (or floppy) is empty. Then check that value after added files, DIRectories (with possible content) in partition or floppy . Subtract it from initial (when it is empty) value, and will see that it is more or much more than "Bytes used" value.  That diff. is the slack.  Smaller when partition is smaller. This is not just some 'be economic at all cost'  ? talk - not everyone has so large Flash cards or hard  disks, and even speed is little better with less slack.

 

 

 

[+DarkLord]

Agreed. I've been running my BBS on a 4 gig SCSI hard drive, via a Mega ST4 with TOS v1.04 since the early 90's.

 

I've got partitions C through K, each slightly under 512 megs.

 

It's worked just fine all these years....

 

[zzip]

I got the bad partition size information from the person who helped me set up my ST hard drives.  I bought my hard drive second hand, so no manual for the disk interface or the SCSI<->MFM controller.    So much of it was new to me at the time, ACSI, SCSI, MFM, RLE, partitioning, hard disk drivers that it all felt overwhelming, so it was very easy to take the word of someone who seemed to know what he was doing :)

[ParanoidLittleMan]

People like simple explanations, yes or no answers (or black or white), but things are in most cases gray, something between, and there are many factors which need to be considered if want really to use it best possible. Plus, things around change, so what was best choice for most decades ago is now far from it.

I saw some really shallow 'opinions' many times, and there are sites which just keep them. Plus, SW sellers (as sellers generally) don't care for facts, they just want to present it as best possible - so if that SW can max 32 MB partition C, that's because 'more is not possible with Atari ST'.