Trim status checking

How, what, where and why - when using the software.

Trim status checking

Postby i3v » 2016.06.11. 18:35

Hi,

I'm wondering how Hard Disk Sentinel checks the "TRIM feature status".
If I'm correct, it checks two things:

  • Whether OS recognizes SSD correctly (and, thus, sends TRIM commands)
  • Whether SSD itself declares the support of TRIM and DRAT (via "word169->bit0", "word69->bit14" and "word69->bit5")
    ( https://koitsu.files.wordpress.com/2013/02/trim_behaviour.png )

So, given two TRUE logical values here, hdsentinel would report, that TRIM is functioning OK, right?
So, it's a "theoretical conclusion", not an "experimental", is it?
Thus, hdsentinel would report "false positive" if there's something wrong with how controller (e.g. RAID) passes TRIM commands, for example, in the case, described here : http://www.win-raid.com/t1366f25-Nightmare-to-find-a-SSD-with-functional-TRIM-on-LSI-SAS.html .

I received some help on the linux-scsi list where it was suggested that for trim to work with LSI controllers, deterministic read after trim (read zero) is needed


I've noticed these two topics :
  1. http://www.hdsentinel.com/forum/viewtopic.php?f=32&t=7609&p=10522&hilit=TRIM#p10522
  2. http://www.hdsentinel.com/forum/viewtopic.php?f=6&t=9798&p=13004&hilit=TRIM#p13004
, also mentioning TRIM. I do understand them, but they do not answer my question.
i3v
 
Posts: 1
Joined: 2016.06.11. 18:05

Re: Trim status checking

Postby hdsentinel » 2016.06.12. 12:30

Hi,

> If I'm correct, it checks two things:

No. These would show status of the OS (if it uses TRIM command or not), the device itself.
But normally these are both true: Windows (until someone manually disables) always use TRIM, SSDs (at least newest models) all support TRIM.
So checking these would not help to determine if TRIM is *really* working, really reaches the SSD?

This only depends on the actual disk controller (where the SSD connected) and its driver. Plus possible other elements "between" the SSD and Windows, for example USB adapter / enclosure (if used).

> So, given two TRUE logical values here, hdsentinel would report, that TRIM is functioning OK, right?

NO.
Hard Disk Sentinel (in addition) checks if TRIM is actually working by determining if the DEVICE_TRIM_DESCRIPTOR is supported by the driver of the actual disk controller and its driver.
See: https://msdn.microsoft.com/en-us/library/windows/hardware/ff552562%28v=vs.85%29.aspx

> So, it's a "theoretical conclusion", not an "experimental", is it?
> Thus, hdsentinel would report "false positive" if there's something wrong with how controller
> (e.g. RAID) passes TRIM commands,

Absolutely NOT: Hard Disk Sentinel would not report false positive.
Hard Disk Sentinel would show correctly if the current hard disk controller and driver does not pass TRIM command as its driver does not implement the functionality which would "pass" that comman to the SSD itself.

Generally
www.hdsentinel.com/faq.php#trim
should give information about this situation and suggest driver change / update which may help in this situation. The Driver Zone page ( http://www.hdsentinel.com/driverzone.php ) designed to help to find the best working drivers for various controllers.
This is common for many controllers (not only SAS controllers but also simpler RAID/non-RAID (AHCI) controllers too).
User avatar
hdsentinel
Site Admin
 
Posts: 2408
Joined: 2008.07.27. 17:00
Location: Hungary


Return to Questions

Who is online

Users browsing this forum: No registered users