You are not logged in.

#1 2022-06-13 16:35:24

dawnofman
Member
Registered: 2019-07-26
Posts: 140

[solved] does alignment still makes sense on 4Kn SSDs ?

I always formatted my 512e drives as following:

sgdisk --new='1:4096:+whateverG' ...

This ensured a good chunk of un-allocated space (8 MB on 512e drives) since the days the SSD vendors were not always disclosing the page size of some drives; eg: Intel.

Yesterday I formatted many new drives that were 4Kn (and some that come 512e from the factory but I switched to 4Kn; eg: SK Hynx P31 Gold 1 TB NVMe) to later realize that I was still doing the 1:4096:+ as I ever did (16 MB) and started guessing ... is this really necessary on these drives ?

I think offset=0 for 4Kn drives is not an issue anymore ... am I right ?

Last edited by dawnofman (2022-06-17 20:27:36)

Offline

#2 2022-06-14 11:44:50

nl6720
The Evil Wiki Admin
Registered: 2016-07-02
Posts: 714

Re: [solved] does alignment still makes sense on 4Kn SSDs ?

There are various myths surrounding alignment on SSDs, so I won't claim to know what's right and what's wrong.

The general practice by partitioning tools is to align to MiB, see https://wiki.archlinux.org/title/Advanc … _alignment.

For your sgdisk command, use 0 as the starting sector to align the start and the -I option to align the end:

sgdisk -I --new='1:0:+whateverG' ...

Offline

#3 2022-06-15 00:01:11

dawnofman
Member
Registered: 2019-07-26
Posts: 140

Re: [solved] does alignment still makes sense on 4Kn SSDs ?

nl6720 wrote:

There are various myths surrounding alignment on SSDs, so I won't claim to know what's right and what's wrong.

There are various myths indeed; and there are various horror stories also ... so I get my bases covered.

nl6720 wrote:

The general practice by partitioning tools is to align to MiB, see https://wiki.archlinux.org/title/Advanc … _alignment.

Yes, I did know of that wiki entry, but quite frankly, I did visit it long-time ago because I've been using my current drives ever since. Now that I have new drives and that I am deploying BTRFS some granted practices (mine of course) come to question. This is why I was using larger than 1 MB offsets -just to be quite quite sure. There's a lot of info that surfaced over the years for those Intel SSD 520/530 series (and earlier ones) -they didn't tell you back in the day the page size for any of those. I even deployed X32Es which were the first Intel 32 GB SSDs and being SLC endured years of flawless operation eventually removing them for lack of speed: SATA 1.5-Gbps ... but 10-years-or-more later I still use them for test drives smile ... those SLC 32 GB drives survived lots of newer MLC consumer drives.

nl6720 wrote:

For your sgdisk command, use 0 as the starting sector to align the start and the -I option to align the end:

sgdisk -I --new='1:0:+whateverG' ...

I always size my partitions using MiB and/or GiB, never decimal sizes, so I guess using the -I switch (which I just learned it exists) won't make things differently (providing I start my partition offset aligned of course; be it at 0 or 4096 or whatever multiples the drive page size).

Offline

Board footer

Powered by FluxBB