Hard drives with 4kb sector-size are now available

[rloew]

Is it possible just to choose appropriate partition starting locations to get pre-Vista systems to work well with 4 KB sector drives, as long as they emulate 512-byte sectors (as I'd kind of expect them to do for a long time, at least as an option, if disk manufacturers don't wish to kiss the legacy aftermarket good-bye)?

Actually DOS 7 and Windows 9x can support 4KB Sector Drices WITHOUT 512 Byre Emulation with the proper modifications. I just secceeded in Booting Windows 98SE from a simulated 4KB Logical Sector Hard Drive.

At least if you were creating a partition for a Win2K or later NTFS system, where (IIRC) the file clusters (if a multiple of 4 KB themselves) begin at 4 KB multiples from the partition's start (yes, that offset may be configurable in the Volume Boot Record, but just in case some brain-damaged software is ignoring that it seems best to leave it 'standard').

But in FAT partitions (in all Windows environments, or just in Win9x?) my vague recollection is that clusters (even if a multiple of 4 KB themselves) are not so nicely aligned with respect to the partition's start. Perhaps RLoew's format utility changes the Volume Boot Record and first-cluster offset to fix this - but if one knew the 'standard' offset (if there is one, you could simply create a FAT partition with 4 KB clusters and see what the alignment was of some file in it) you could, again, just adjust the partition's start location to make the file clusters fall on disk 4 KB boundaries.

I have noted your concern and updated my Formatter to minimize the need to change the Standard Offset. It may still be necessary to move the Partition to use the Standard Offset, but now it will only need to move one Cylinder.

[billtodd]

Thanks for replying.

Is it possible just to choose appropriate partition starting locations to get pre-Vista systems to work well with 4 KB sector drives, as long as they emulate 512-byte sectors (as I'd kind of expect them to do for a long time, at least as an option, if disk manufacturers don't wish to kiss the legacy aftermarket good-bye)?

Actually DOS 7 and Windows 9x can support 4KB Sector Drices WITHOUT 512 Byre Emulation with the proper modifications.

I was attempting to ascertain whether merely positioning partition start locations strategically on 4 KB sector drives that DO support 512-byte-sector emulation would be sufficient to allow several legacy operating systems of interest installed within those partitions without any modification (or special formatting) to suffer no significant performance degradation (e.g., to avoid almost all read/modify/write sequences caused by the larger sector size) - at least as long as those systems used file clusters that were multiples of 4 KB. As long as those systems do maintain consistent offsets to the first file cluster (and then manage free space such that clusters are tightly packed) I couldn't see any reason why this wouldn't work, but it's always nice to have such suppositions confirmed.

[rloew]

Thanks for replying.

Is it possible just to choose appropriate partition starting locations to get pre-Vista systems to work well with 4 KB sector drives, as long as they emulate 512-byte sectors (as I'd kind of expect them to do for a long time, at least as an option, if disk manufacturers don't wish to kiss the legacy aftermarket good-bye)?

Actually DOS 7 and Windows 9x can support 4KB Sector Drices WITHOUT 512 Byre Emulation with the proper modifications.

I was attempting to ascertain whether merely positioning partition start locations strategically on 4 KB sector drives that DO support 512-byte-sector emulation would be sufficient to allow several legacy operating systems of interest installed within those partitions without any modification (or special formatting) to suffer no significant performance degradation (e.g., to avoid almost all read/modify/write sequences caused by the larger sector size) - at least as long as those systems used file clusters that were multiples of 4 KB. As long as those systems do maintain consistent offsets to the first file cluster (and then manage free space such that clusters are tightly packed) I couldn't see any reason why this wouldn't work, but it's always nice to have such suppositions confirmed.

Yes. But Partition Size is also a factor.

The trick is knowing how to do it and not breaking something else in the process.

My tools have been updated to handle this issue for Windows 9x.

[billtodd]

I was attempting to ascertain whether merely positioning partition start locations strategically on 4 KB sector drives that DO support 512-byte-sector emulation would be sufficient to allow several legacy operating systems of interest installed within those partitions without any modification (or special formatting) to suffer no significant performance degradation (e.g., to avoid almost all read/modify/write sequences caused by the larger sector size) - at least as long as those systems used file clusters that were multiples of 4 KB. As long as those systems do maintain consistent offsets to the first file cluster (and then manage free space such that clusters are tightly packed) I couldn't see any reason why this wouldn't work, but it's always nice to have such suppositions confirmed.

Yes. But Partition Size is also a factor.

If the partition start location is set such that the file system clusters (as long as they're multiples of 4 KB in size) will be 4KB-aligned on the disk, then unless there's some other non-file-structured system data that gets frequently written I would think that where the partition ended wouldn't matter (i.e., if it ended at some odd boundary the file system simply would leave the final few sectors that were insufficient to create another cluster unused - just as it has always done).

The trick is knowing how to do it and not breaking something else in the process.

I don't understand HOW anything could be broken as long as one used standard (pre-Vista) partitioning utilities to create the partitions: the entire point of this exercise would be to create bog-standard partitions using standard 'cylinder' boundaries carefully chosen to make virtually all disk access activity by legacy file systems 4KB-aligned (as long as those file systems used clusters that were multiples of 4 KB in size).

Even if one made a mistake in such choice the worst that should happen in the presence of 512-byte sector emulation is that small-write performance would suffer - at least as long as one refrained from using Vista/Win7 Disk Management to muck around with the partition structure without first telling it to use old-style partition alignment (a Registry modification which Microsoft appears to support).

Am I still missing something?

[rloew]

I was attempting to ascertain whether merely positioning partition start locations strategically on 4 KB sector drives that DO support 512-byte-sector emulation would be sufficient to allow several legacy operating systems of interest installed within those partitions without any modification (or special formatting) to suffer no significant performance degradation (e.g., to avoid almost all read/modify/write sequences caused by the larger sector size) - at least as long as those systems used file clusters that were multiples of 4 KB. As long as those systems do maintain consistent offsets to the first file cluster (and then manage free space such that clusters are tightly packed) I couldn't see any reason why this wouldn't work, but it's always nice to have such suppositions confirmed.

Yes. But Partition Size is also a factor.

If the partition start location is set such that the file system clusters (as long as they're multiples of 4 KB in size) will be 4KB-aligned on the disk, then unless there's some other non-file-structured system data that gets frequently written I would think that where the partition ended wouldn't matter (i.e., if it ended at some odd boundary the file system simply would leave the final few sectors that were insufficient to create another cluster unused - just as it has always done).

The trick is knowing how to do it and not breaking something else in the process.

I don't understand HOW anything could be broken as long as one used standard (pre-Vista) partitioning utilities to create the partitions: the entire point of this exercise would be to create bog-standard partitions using standard 'cylinder' boundaries carefully chosen to make virtually all disk access activity by legacy file systems 4KB-aligned (as long as those file systems used clusters that were multiples of 4 KB in size).

Even if one made a mistake in such choice the worst that should happen in the presence of 512-byte sector emulation is that small-write performance would suffer - at least as long as one refrained from using Vista/Win7 Disk Management to muck around with the partition structure without first telling it to use old-style partition alignment (a Registry modification which Microsoft appears to support).

Am I still missing something?

The Data area is after the FAT tables, so the size of the FAT Tables can affect the alignment. This is why the Partition size is important.

Aligning some standard Partitions require a legal but non-standard modification to the Format. If your Computer uses 240 Head mapping, it may be impossible to avoid this issue regardless of Partitioning, if you use standard settings.

[billtodd]

Am I still missing something?

The Data area is after the FAT tables, so the size of the FAT Tables can affect the alignment. This is why the Partition size is important.

So for FAT partitions there is NOT any fixed offset to the first cluster, and one would either need to use a standard partition size (for which that offset had been determined previously) or create the partition, check the offset, and then recreate it at a 'cylinder' boundary which would cause that offset to create aligned multiple-of-4KB clusters. (A complementary approach could tweak the partition size to affect the offset, I suppose - though since each 'cylinder' comprises around 2,000 4 KB sectors the required size change could be prohibitive unless the partition could be forced to end at other than a 'cylinder' boundary, which would introduce an undesirable peculiarity.) Throw in the 240-head case which you mention below and the benefits of your special format utility become more evident.

Is the NTFS offset fixed (or at least always created such that it's a multiple of 4KB)? I seem to recall discussions w.r.t. the alignment of database clusters that suggest that it is - and I probably could (and should) just try to dig up the spec to verify this.

Aligning some standard Partitions require a legal but non-standard modification to the Format. If your Computer uses 240 Head mapping, it may be impossible to avoid this issue regardless of Partitioning, if you use standard settings.

Since every 'cylinder' will begin at a 4KB boundary. This would be fine for any (non-FAT) primary partitions on the drive if their file systems aligned the first cluster on a 4 KB boundary w.r.t. the partition start (though you'd need to leave an unused 'cylinder' prior to the first partition), but logical partitions could never be optimally aligned.

Or you could perform all partitioning operations on a different system which used standard (255 x 63) 'cylinders' before populating the partitions on the target system - as long as you were prepared NEVER to use the target system to alter the partition layout (well, some utilities might be smart enough to handle it correctly, but Partition Magic certainly wasn't back when it was still otherwise reasonably usable). Fortunately, I don't foresee using a 4KB-sector drive in my ancient Thinkpad any time soon (do they still use that 240-head pseudo-geometry today?).

Thanks again for having taken the time to help clarify things. I may be overly paranoid about using any atypical partition layouts (internal or external), but between seeing how Vista/Win7 and older Windows systems can screw each other up and my past experience with Partition Magic's refusal even to LOOK at a disk partitioned on a system with differing pseudo-geometry without first 'fixing' it and thereby making it unusable I prefer to be conservative where possible in this area.

Edited June 8, 2010 by billtodd

[rloew]

I have solved the last Software barrier to using 4KB Logical Sector Drives. I can now boot from them as well as run DOS and Windows.

The only issue left is the BIOS, over which I have no control. If necessary I can create a new BOOTMAN Package to bridge the gap.

[rloew]

While Western Digital appears to be heading in the direction of 4KB Logical Sectors to break the 2TiB barrier, it appears that Seagate has decided to break the 32-Bit Sector barrier for their 3TB Hard Drive. This means there will be two different issues for Windows 9x. It looks like I may have to break out my other set of Patches before I get to put my new ones to use.

[Wolfgang16]

Hello,

internal data of the new Seagate 3TB USB drive has just been published here:

Link

These data include a dump of the first 1MB of the drive and can be downloaded

Download

Well this is interesting... The drive under Windows does show up as a native USB device with 4K sector size. However, it also has a 512-byte MBR in the first 1/4 of sector 0. According to the MBR, the first partition is NTFS and starts at sector 7 (?!) and ends at 732564000 = 2794GB when interpreted as 4K sectors.

It's a pretty simple kludge really, the MBR is supposed to be the first sector on a 512-byte disk but they've just shoehorned the same 512 bytes into a 4K sector and left the remaining 1536 empty, counting on the OS not to care that it's a 4K disk when it really should be a 512-byte disk.

So it's not the NTFS format that's been kludged, it's the partition table. For all intents and purposes it's a 4K drive, but uses a 512-byte MBR when it should be using GPT. Windows seems to just read the first 512-bytes to get the partition table and interpret it in 4K mode (which it really shouldn't). I can understand why older partitioning tools would be confused by it...

May be this can also help for Win98.

[rloew]

Hello,

internal data of the new Seagate 3TB USB drive has just been published here:

Link

These data include a dump of the first 1MB of the drive and can be downloaded

Download

Well this is interesting... The drive under Windows does show up as a native USB device with 4K sector size. However, it also has a 512-byte MBR in the first 1/4 of sector 0. According to the MBR, the first partition is NTFS and starts at sector 7 (?!) and ends at 732564000 = 2794GB when interpreted as 4K sectors.

It's a pretty simple kludge really, the MBR is supposed to be the first sector on a 512-byte disk but they've just shoehorned the same 512 bytes into a 4K sector and left the remaining 1536 empty, counting on the OS not to care that it's a 4K disk when it really should be a 512-byte disk.

So it's not the NTFS format that's been kludged, it's the partition table. For all intents and purposes it's a 4K drive, but uses a 512-byte MBR when it should be using GPT. Windows seems to just read the first 512-bytes to get the partition table and interpret it in 4K mode (which it really shouldn't). I can understand why older partitioning tools would be confused by it...

May be this can also help for Win98.

I haven't seen the Drive for sale locally and I'm not sure if Windows 9x USB Drivers can handle it, but the format is compatable with my Large Sector Patches.

It appears from the Identification Data on the Drive inside it, that it uses 512 Byte Logical Sectors and requires more than 32-Bits to address.

I do have Patches for supporting more than 32-Bit Addressing but with less capability.

Edited August 10, 2010 by rloew

[rloew]

I have completed testing on a new set of Patches that will support this Hard Drive when it becomes available as an Internal Drive.

Using an Emulation Overlay, I was able to Partition, access and Boot From Sectors that appeared to be above 2TiB.

[rloew]

I received a Seagate 3TiB Go Flex External Drive with USB 2.0 this morning.

It uses 4KB Sectors at the USB Interface, while using 512 Byte Sectors at the Internal SATA Interface.

None of the BIOSes I have tested support it. So Booting from it and/or running DOS without a Driver requires BIOS modifications.

Windows 98SE, with ME USB Drivers and the necessary Patches, does support it, so the drive is useable as an USB External.

I have yet to disassemble it to test the bare SATA Drive.

[rloew]

I received a Seagate 3TiB Go Flex External Drive with USB 2.0 this morning.

It uses 4KB Sectors at the USB Interface, while using 512 Byte Sectors at the Internal SATA Interface.

I disassembled it to test the bare SATA Drive. It wasn't easy.

None of the BIOSes recognized it as 3TB. They all said it was 801GB.

They also wrap around at 2TiB back to Zero so a DDO is required to use it.

The SI 3512 and the JMicron SATA Cards correctly recognize the drive as 3TB and properly handle requests above 2TiB.

I have a Patched IO.SYS that support Partitions above 2TiB.

[rloew]

I have now tested both the Seagate Go-Flex 3TB USB Drive and the Internal Western Digital 3TB Drive.

I have developed a complete Package (TBPLUS) to support these Hard Drives, and larger. It is now available on my website.

This should help keep Windows 98 alive for another 15-20 Years.

[osRe]

Has anyone tested the performance of 4K drives with aligned partitions on Win9x? Is there performance degradation?

Even if the data area is 4K aligned and uses 4K clusters, wouldn't there be degraded performance on FAT accesses (which aren't cluster-bound)?

Edited August 22, 2011 by shae