Funny. I just watched a video on partition alignment today, and downloaded Minitool Partition Wizard. I was unable to find 9.1 though. I downloaded 10.2 of the Windows version, and also a versions 7 and 8 bootable CD. I assume those will work on Win98 machines, as long as the machine supports booting from CD. There was also flash bootable images, but I didn't download those.

maybe you can try the following.
partition a harddisk with msdos, pcdos, freedos, w95, w98, wme, w2k, wxp, wvista, w7, ... linux, bsd, ... os2, ...
and after each partition creation use your alignment program to check for misalignment and report here your findings 😀

Note that I've not tested any of what I downloaded yet, so not sure if they support partition alignment.

As covered in this? https://www.youtube.com/watch?v=SdNG7YImRQQ&t=62s

Misalignment was the norm because the first partition would start on the second 'cylinder' which was represented as 63 sectors. What's the rationale to explain why this would affect performance on drives that use native 512-byte sectors?

bakemono wrote on 2025-01-01, 22:06:

Misalignment was the norm because the first partition would start on the second 'cylinder' which was represented as 63 sectors. What's the rationale to explain why this would affect performance on drives that use native 512-byte sectors?

IBM has a paper on that which I am reading right now. As far as I know, it is an issue on drives with large sectors.

https://www.ibm.com/support/pages/misalignmen … n-be-twice-cost

weedeewee wrote on 2025-01-01, 21:43:

maybe you can try the following.
partition a harddisk with msdos, pcdos, freedos, w95, w98, wme, w2k, wxp, wvista, w7, ... linux, bsd, ... os2, ...
and after each partition creation use your alignment program to check for misalignment and report here your findings 😀

All my drives have a single partition.

From what I have found, misalignment had a correlation with the size and the age of the drive, and not the OS which formatted it.

My Conner CFS270A 270Mb had no misalignment problems. Neither did the 2.1Gb Seagate SCSI Hdd. SATA drives formatted in Windows XP, 150 and 500Gb, had no misalignment. SATA HDD, 75Gb, formatted in Windows XP, had misalignment. 40Gb Seagate HDD limited to 32Gb via jumper, formatted and used in a Pentium 2, Windows 98SE, had no misalignment. A Seagate 40Gb and a Maxtor 60Gb, used in another Windows 98 PC (Athlon XP 2000+) did have misalignment. So at this point in time I don't know why and how it happens.

What was important is - I was using those disks without even knowing they were misaligned. And fixing one of them shaved 20+ seconds off my boot time on one PC.

Grem Five wrote on 2025-01-01, 21:57:

As covered in this? https://www.youtube.com/watch?v=SdNG7YImRQQ&t=62s

Phil mostly concentrates on flash media deterioration.

bakemono wrote on 2025-01-01, 22:06:

Misalignment was the norm because the first partition would start on the second 'cylinder' which was represented as 63 sectors. What's the rationale to explain why this would affect performance on drives that use native 512-byte sectors?

Maybe to be on the safe side.

Many disks have "hardware" 4k sectors but are capable to use "logical" 512 byte sectors.

The performance decreases (and wear increases!) when partition and/or clusters are misaligned with "hardware" sector size. Or so they say. In most of my flash cards, the sector and cluster size are way smaller than the ERASE BLOCK SIZE. So if your device has 128k memory page size, I guess the correct place to align partitions should be at 128k boundaries. Also, first cluster should be aligned to that boundary and the cluster size should be a 128k multiple... or you'll end erasing the whole page every time it has to erase a single cluster.

Modern OSs and software hide (or does not explain well) what are doing at every moment. Some questions...
- Alignment programs DO take care about first cluster alignment or only about start of partition alignment?
- Does an OS "know" how long is the erase block size on an flash device and/or takes it in account to minimize erase/write cycles?
- Or can the memory controller on the flash device "group" writes?

Just an update; unlike Windows XP, realigning the HDD partitions on the Windows 98SE computer did not result I any noticeable performance increase.

Zup wrote on 2025-01-02, 08:51:

Modern OSs and software hide (or does not explain well) what are doing at every moment. Some questions... - Alignment programs D […]
Show full quote

Modern OSs and software hide (or does not explain well) what are doing at every moment. Some questions...
- Alignment programs DO take care about first cluster alignment or only about start of partition alignment?
- Does an OS "know" how long is the erase block size on an flash device and/or takes it in account to minimize erase/write cycles?
- Or can the memory controller on the flash device "group" writes?

That's actually a very good question. Will need to dig into the documentation to see if those features are well documented.

Minitools actually moved my primary partition 1mb ahead to align it, while Paragon did what I expected it to do.

In contrast, Paragon broke my Windows 98 installation (the MBR was fine, but after that I only got a pulsing cursor - no IO.SYS, no LOGO.SYS, no activity on the HDD).
Booting from Win98SE installation media and trying to run WIN would BSOD the system. I had no issues formatting the aligned partition and reinstalling Windows98SE. This time I remembered to backup my Morrowind saves.

Minitools did not break any of my Windows installations.

Zup wrote on 2025-01-02, 08:51:

Maybe to be on the safe side. […]
Show full quote

bakemono wrote on 2025-01-01, 22:06:

Misalignment was the norm because the first partition would start on the second 'cylinder' which was represented as 63 sectors. What's the rationale to explain why this would affect performance on drives that use native 512-byte sectors?

Maybe to be on the safe side.

Many disks have "hardware" 4k sectors but are capable to use "logical" 512 byte sectors.

The performance decreases (and wear increases!) when partition and/or clusters are misaligned with "hardware" sector size. Or so they say. In most of my flash cards, the sector and cluster size are way smaller than the ERASE BLOCK SIZE. So if your device has 128k memory page size, I guess the correct place to align partitions should be at 128k boundaries. Also, first cluster should be aligned to that boundary and the cluster size should be a 128k multiple... or you'll end erasing the whole page every time it has to erase a single cluster.

That's not the only thing your should consider.

Legacy operating systems use CHS address scheme and need to be aligned at least to a starting cylinder/head value (sector number 1). Most legacy partition tools align them even to a starting cylinder with head number 0. An exception is the first partition which will be aligned to the 64th sector which will be unaligned on any disk that uses sectors bigger than 512k like almost every SSD. Unaligned access does not mean simple write a sector but will cause a read and write operation everytime if not cached by the disk.
Modern operating systems use LBA address scheme. They do not care about being aligned to CHS sectors.
Operating systems since Windows Vista, 7, 8, 8.1, 10, 11 create megabyte-aligned partitions. They are compatible with 4k alignment of course.

Best is when your partition and format (!) scheme is when the layout has first data sector aligned to your physical disk behaviour (512 bytes, 4k, or ERASE BLOCK SIZE).
For FAT16 you have to calculate bootsector + FAT + copy of FAT + rootdirectory.

For myself, I do not know any of the operating systems tools that considers this.