RAID0 trim and SRT seems possible on 6-series and 5-Series

Discussion in 'Memory and Storage' started by Dufus, Nov 8, 2012.

  1. Foonus

    Foonus Junior Member

    Joined:
    Sep 2, 2012
    Messages:
    17
    Likes Received:
    0
    Ya, I smell the stench of corporate politics and profiteering... Imagine how badly it will stink if they manage to force AMD to go tits up.

    If what you found is verified, its indeed amazing how fast they can respond to no doubt the article on Anandtech and cripple the thing, as opposed to enabling it for everyone after users have been harping about it for over a year. Relate the time it took to do this in comparison to the time it took them to release a RST without a memory leak after their buggy one!

    These guys are worse than forum trolls, not only are they negative but they literally have the power to do what a troll can't.. undo the progress the community has made; and they are Greedy enough to do it.

    Whoever decided to do this, let alone rush it , deserves a GSP kick in the groin and a lump of coal for Christmas, though the kick wouldn't do much as they don't even have the testicular fortitude to tell you to your face that they are crippling the communities progress in the patch notes.

    For Intel the excuse of "validation costs" doesn't apply to crippling hardware, only updating it.
     
    #276 Foonus, Dec 6, 2012
    Last edited: Dec 6, 2012
  2. Dufus

    Dufus Senior member

    Joined:
    Sep 20, 2010
    Messages:
    469
    Likes Received:
    0
    That might work davidm71. My guess is in earlier boards, pre PCH, the alternate ID was never passed on so that the OROM firmware could use it. IOW having the correct ID will not help in this case.

    If you are really curious to find out the real ID it might be possible to activate it by setting the proper register on the SATA controller, if not locked by the BIOS.


    NotSmart, unfortunately SRT needs more than just OROM support, which is why my first test failed, it also need BIOS support and adding that is not so simple and will vary from BIOS to BIOS.


    Latituder, I wrote my own software for doing this and the times help not only to get an idea of trim performance but also in a case where RSTe drivers are used. With RSTe and iastorF a failed trim command gets changed to successful when reported back. Failures tend to return very quickly so by using the time a very quick response of just a few microseconds can be an indication that it actually failed.

    At the moment the test software is not suitable for distribution as it has little in the way of error checking and is fixed to test just a few LBA's outside of my file system. IOW it could easily mess things up on a different system where variables are not known.

    I would like to produce something that could be used by everyone and provide some certainty of trim test but the deeper I dig into it the more problems I seem to find, even RST driver bugs and SSD firmware bugs. :( I have not given up this idea yet but have little time to spend on it right now.


    Fernando, I ran across this
    posting as CPL0 I was asked by Yavi back in May on hardforum how to test for trim and while using my own methods said that Winhex could be used to check LBA's. I certainly was not aware of someone having already posted this method so I apologize if it seems I was stealing your ideas without giving credit.
     
  3. Fernando 1

    Fernando 1 Senior member

    Joined:
    Jul 29, 2012
    Messages:
    325
    Likes Received:
    0
    This story is nothing to worry about. The "Easy Trim Test Method" by using a hex editor, which I have published >here< on 09/01/2012 at Station-drivers Forum, was not my discovery at all. It has been in July this year, when I ran across >this< post written by the XtremeSystems Forum user Mark HFR. All I had done was to test the method in July 2012 with an Intel Z68 RAID0 and later with an Intel Z77 RAID0 system with the result, that TRIM was just working on 7-Series chipset RAID0 systems. Since I couldn't find any detailed instruction about how to do the hex editor test method, I wrote later a guide in German and English language.
    So your publication at [H]ard|Forum dated 05/26/2012 (>Link<), which I have seen the first time now, has been written much earlier. So I rather have to give credits to you regarding the TRIM test method than you to me vice versa.
     
  4. damageboy

    damageboy Junior Member

    Joined:
    Mar 21, 2007
    Messages:
    17
    Likes Received:
    0
    OK, so far no good.
    I've updated the OROM... Here are complementary screenshots:
    Before (System Report):
    http://imgur.com/uWqJm,FLTPH,LRTB2,fkQOb,9BvGt,bPC07#0
    Before Deleting with the old OROM:
    http://imgur.com/uWqJm,FLTPH,LRTB2,fkQOb,9BvGt,bPC07#2
    After Deleting with the old OROM:
    http://imgur.com/uWqJm,FLTPH,LRTB2,fkQOb,9BvGt,bPC07#3
    After BIOS Update, Everything booted cleanly!, (System Report):
    http://imgur.com/uWqJm,FLTPH,LRTB2,fkQOb,9BvGt,bPC07#1
    Before Second Delete:
    http://imgur.com/uWqJm,FLTPH,LRTB2,fkQOb,9BvGt,bPC07#4
    After Second Delete (Waited 10 minutes):
    http://imgur.com/uWqJm,FLTPH,LRTB2,fkQOb,9BvGt,bPC07#5

    As it is apparent from the last screenshot, the TRIM didn't happen here.
    The raid was created with the previous OROM (Which will be my next attempt, to re-create with the new OROM).

    Any help/ideas will be appreciated.
     
    #279 damageboy, Dec 6, 2012
    Last edited: Dec 6, 2012
  5. Dufus

    Dufus Senior member

    Joined:
    Sep 20, 2010
    Messages:
    469
    Likes Received:
    0
    Just to clarify Fernando, testing trim by checking LBA's seemed a logical thing to do and this appears evident by the way similar testing has been done long before my post on HF. So no, I do not deserve any credit for this and this is not what my post is about. It's about someone accusing me of stealing, which I find highly objectionable.

    IMHO the credit is still yours by the way you have done so much work towards providing answers and help to the community, much more than I have ever done. You have my utmost respect for that. :thumbsup:

    damageboy, did you try rebooting and checking again? Sometimes it appears the trim results are negative but after a reboot are good, or sometimes just placing the disk offline and online with diskpart, or running chkdsk without automatically fixing. I don't know if it's because of caching that the old data is returned or simply that the trim command has not yet been processed by the disk. The Intel driver seems to block disk flush commands for RAID arrays so that doesn't help to diagnose the problem.
     
  6. damageboy

    damageboy Junior Member

    Joined:
    Mar 21, 2007
    Messages:
    17
    Likes Received:
    0
    I've re-created the RAID array.
    This time the RST wizard recomended a stripe-size of 16KB (previously it was 128KB).
    So I accepted as it seemed more relevant for a boot drive with many small files.

    The test didn't go any better apparently:
    http://imgur.com/L1NWH

    The same sector remained as it was before, after an additional 10 minutes of use...

    After I rebooted the sector was overwritten by some NTFS data-structure, but still did not contain 00's or FF's, I assume that the overwriting implies it just happened to use the same sector...
     
  7. damageboy

    damageboy Junior Member

    Joined:
    Mar 21, 2007
    Messages:
    17
    Likes Received:
    0
    I thought about rebooting, and have indeed done it, but the same sector got overwritten by some NTFS data structure (Starts with INDX which I remember to be an NTFS stream).
    I'll try again...
     
  8. damageboy

    damageboy Junior Member

    Joined:
    Mar 21, 2007
    Messages:
    17
    Likes Received:
    0
    I did:
    mountvol x: /d
    diskpart:
    select volume 3
    offline volume
    online volume

    mountvol x: ...

    Didn't seem to matter...

    The question I have is: is there a more "accurate" trim tool that can return a value from the controller explaining what happened?
     
  9. davidm71

    davidm71 Member

    Joined:
    Nov 13, 2008
    Messages:
    50
    Likes Received:
    0
    Unfortunately my knowledge is limited and don't know enough to hack on that level. Willing to beta test any ideas you guys might have on my X58 setup though..
     
  10. Engineer

    Engineer Elite Member

    Joined:
    Oct 9, 1999
    Messages:
    37,893
    Likes Received:
    2
    Has anyone else tried this yet? I jumped to conclusions on this yesterday when I installed it on my AHCI system and didn't see an OROM in the properties page and concluded that Intel had turned it off. Now it seems that they may have really done that. If so, I hope AnandTech does a followup on the article reflecting that. Give them bad press over it.
     
  11. ney2x

    ney2x Junior Member

    Joined:
    Nov 21, 2012
    Messages:
    19
    Likes Received:
    0
    As far as I know, there is no "RAID Option ROM version: 11.x.x.xxx, under System Report in IRST GUI, if you are on AHCI mode.

    I know that Fernando has ASUS P8Z68-V (same as mine) and I am waiting for his confirmation/verification if TRIM (RAID-0) is working (not working on my test) if using the latest IRST 11.7.0.1013 together with modified-RAID OROM 11.6.2.1702.
     
  12. Fernando 1

    Fernando 1 Senior member

    Joined:
    Jul 29, 2012
    Messages:
    325
    Likes Received:
    0
    Maybe you are able to verify what happened with the previously deleted JPG file by doing a search for a significant text or hex code fragment of the file.
    If the hex editor shouldn't find it anymore within the related RAID0 partition, TRIM may be active within your RAID0.

    Maybe I will do the test, but it will cost me a lot of time, because my current system is an Intel Z77 AHCI one.
     
    #287 Fernando 1, Dec 6, 2012
    Last edited: Dec 6, 2012
  13. retiredfields

    retiredfields Junior Member

    Joined:
    Dec 6, 2012
    Messages:
    3
    Likes Received:
    0
    First, my many thanks to those who contributed to this thread.

    I have an Asus P8P67 Deluxe, with a Samsung 830 raid(0) boot array and an Intel X-25 raid(0) array for virtual machines.

    I am now running bios 3602 with the Intel 11.6.0.1702 rom and the current IRST 11 driver on Windows 8.

    My testing shows that the Intel array immediately zeroes out a file, when it is permanently deleted, indicating that TRIM is working. However, the Samsung array does not. Even after waiting 30 minutes, the Samsung array still does not appear to have processed the TRIM command.

    Both arrays were pre-existing and not recreated with the new orom.

    Any suggestions or comments? Any Samsung 830 owners out there?

    Ken
     
  14. Fernando 1

    Fernando 1 Senior member

    Joined:
    Jul 29, 2012
    Messages:
    325
    Likes Received:
    0
    Hello Ken,
    welcome at Anandtech Forum!
    Has the Intel RAID ROM file been modified for TRIM support or does your mainboard BIOS contain the original ROM?

    Does this happen automaticly or are you using the "Optimize" button of Intel's SSD Toolbox?
     
  15. retiredfields

    retiredfields Junior Member

    Joined:
    Dec 6, 2012
    Messages:
    3
    Likes Received:
    0
    Hello Fernando,

    My thanks to you directly for all your work and instruction guides.

    Yes, the Intel ROM has been modified. Sata ID 1C04 on controller 2822, changing entry 1 to 7 in the rom. TRIM is definitely working for the Intel array. The Samsung array is not responding as I expected.
    Yes, this happens automatically with the Intel raid(0) array.
     
  16. kulica

    kulica Junior Member

    Joined:
    Nov 20, 2012
    Messages:
    5
    Likes Received:
    0
    Hi there. You were right, there was a problem with my rig :S Today I also updated my crucial firmware, now it works as it should. By the way TRIM does work on intel RST 11.7 driver. There is a proof here.

    Original image:
    [​IMG]

    Offset:
    [​IMG]

    Driver 11.6.2:
    [​IMG]

    New driver info:
    [​IMG]

    New offset (actually it's the same):
    [​IMG]

    And finally the TRIM test on 11.7:
    [​IMG]
     
    #291 kulica, Dec 6, 2012
    Last edited: Dec 6, 2012
  17. Fernando 1

    Fernando 1 Senior member

    Joined:
    Jul 29, 2012
    Messages:
    325
    Likes Received:
    0
    So here we are with the test results running Windows 8 Pro x64:
    1. TRIM is working fine within my Z68 chipset RAID0 after having flashed a BIOS with the modded Intel RAID ROM v11.6.0.1702.
    2. After having updated the Intel RST(e) drivers from v11.6.2.1002 to v11.7.0.1013 I got exactly the same result: TRIM is still working fine.
    The TRIM activity has been tested by using HxD according the "Easy TRIM Test" layed down at Station-Drivers (>LINK<).

    Conclusion: Intel's new RST(e) drivers v11.7.0.1013 obviously do not touch the TRIM in RAID0 support of 6-series chipset mainboards with a BIOS, which contains an especially modded version of the Intel RAID ROM v11.6.0.1702.
     
    #292 Fernando 1, Dec 6, 2012
    Last edited: Dec 18, 2012
  18. Engineer

    Engineer Elite Member

    Joined:
    Oct 9, 1999
    Messages:
    37,893
    Likes Received:
    2
    :thumbsup:
     
  19. Ourasi

    Ourasi Junior Member

    Joined:
    Jun 15, 2009
    Messages:
    19
    Likes Received:
    0
    Same results here, TRIM is working fine on an Intel SSD 520 Raid0 array on my P67 chipset... Intel SSDtool seems a bit unstable with this new driver, hangs and then works again randomly, not in much use anymore so no biggie..
     
  20. MaxICH

    MaxICH Member

    Joined:
    Nov 19, 2012
    Messages:
    30
    Likes Received:
    0
    I just tried it on my P8Z68-V Pro with the latest IRST 11.7.0.1013 and it works just as it worked with the previous 11.6.0.1030.
     
  21. ney2x

    ney2x Junior Member

    Joined:
    Nov 21, 2012
    Messages:
    19
    Likes Received:
    0
    Thanks for those clarification. Well, I guess my 2 rig has the problem, maybe the system files are messed up because they're "production systems". Thanks again.
     
  22. damageboy

    damageboy Junior Member

    Joined:
    Mar 21, 2007
    Messages:
    17
    Likes Received:
    0
    Hi guys, specifically Dufus and Fernando...

    Apparently it was indeed working all along.
    The thing I missed was that since I was using Windows 8, the TRIM isn't triggered as immediately as one would hope but seems to be rather delayed.

    What I did to trigger it was run the defrag tool.
    The defrag tool specifically reports that it is trimming the drive...
    And indeed after the trim operation I could see that my Vertex 4 RAID-0 is filled with 0xFF where previously it had a JPG file...

    I'll try to make a video out of the process of how I do the test, for the rest of the community...

    Thanks a lot!
     
    #297 damageboy, Dec 7, 2012
    Last edited: Dec 7, 2012
  23. damageboy

    damageboy Junior Member

    Joined:
    Mar 21, 2007
    Messages:
    17
    Likes Received:
    0
  24. Fernando 1

    Fernando 1 Senior member

    Joined:
    Jul 29, 2012
    Messages:
    325
    Likes Received:
    0
    Thanks for the video, which verifies, that TRIM is active within your P55 RAID0 array after having inserted the modded Intel RAID ROM v11.6.0.1702 into the BIOS. :thumbsup:
    That is very good news not only for other users with a quite similar Intel P55 SSD RAID0 system, but for me as well, because now I know, that the especially modified Intel RAID ROM file, which I am offering >here< for P55 systems, definitively enables the TRIM in RAID0 feature for these older chipsets.
     
  25. Mike73

    Mike73 Junior Member

    Joined:
    Dec 7, 2012
    Messages:
    5
    Likes Received:
    0
    Hello, this is my first message.

    This is my configuration:

    1)- Windows 7 x64;
    2)- Mainboard Asus P8P67 Deluxe - Bios 3509 - Orom 11.6.0.1702;
    3)- Driver Intel Rapid Storage 11.6.2.1002 WHQL;
    4)- 2 SSD Vertex 3 MI - Firmware 2.02 - RAID 0;

    I tried using the software HxD and it seems that TRIM does not work (I do not see 00 or FF).

    Tips?

    Thank you.