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

[MaxICH]

In my case all I did was change "07 1C 00 00 01" to "07 1C 00 00 07" and "06 1C 00 00 01" to "06 1C 00 00 07".

Maybe for my Z68 board only the last one was needed, but I changed both just in case.

 

[Evildeffy]

Hi Dufus,

First off you just made a whole lot of my friends with a P67/Z68 chip very happy.

Second, considering that X58/6-series/7-series are all the ICH10R family, you think you can alter that Option ROM to include the X58 series for RAID0 Trim?

You would make me beyond happy with this.

Thanks for all your effort in any case!

 

[Makaveli]

wow great find guys.

I think now I will finally look at updating the OROM on my board.

 

[Engineer]

Just posting to give :thumbsup: and a huge Thank You to Dufus (and Fernando and everyone else who has helped along the way in the quest for TRIM for RAID 0).

You guys are awesome and happy Thanksgiving!

 

[davidm71]

Was wondering if you guys know the alternate id for an X58 motherboard like the evga classified? Going to investigate right now in device manager and see if I can find it and try that mod,

Thanks

 

[davidm71]

Hi again..

I tried to mod my 11.6.0.1702 bios file with these 8086 ids: 3a20, 3a25, and 3a05 and set the flag to 07. I'm running 11.6.2.1002 on an Evga Classified 759 X58 motherboard with two HyperX3Ks in raid 0 slave drive. I ran the Trim test routine using the hex editor and the OS did not Trim the drive. Running Win 7 64 also...

Anyhow would this only work Windows 8? Do I need to reset the raid array? Reinstall the drivers?

Thanks.

 

[MaxICH]

Hi again..

I tried to mod my 11.6.0.1702 bios file with these 8086 ids: 3a20, 3a25, and 3a05 and set the flag to 07. I'm running 11.6.2.1002 on an Evga Classified 759 X58 motherboard with two HyperX3Ks in raid 0 slave drive. I ran the Trim test routine using the hex editor and the OS did not Trim the drive. Running Win 7 64 also...

Anyhow would this only work Windows 8? Do I need to reset the raid array? Reinstall the drivers?

Thanks.

Where did you find those id's? Are you sure they are the right ones?

 

[davidm71]

Where did you find those id's? Are you sure they are the right ones?

I googled my chipset and assumed its an ich10r and googled those terms until I found a list of hardware ids. I looked up my raid controller in device manager 8086 2822 and found subsys of 3A20 which was close to the other ids found with a google search so I added them all in there. I even tried running Aida64 and hardware info but didn't know or couldn't find the alternate ids..

Thanks

 

[Fernando 1]

@ davidm71:

You will find the possible internal DeviceIDs of all Intel 5-Series SATA Controllers on page 14 of >this< official paper published by Intel.
AFAIK the correct DeviceID for your X58 RAID Controller is 3B25. That would be in hex code "25 3B".

 

[davidm71]

@ davidm71:

You will find the possible DeviceIDs of all Intel 5-Series SATA Controllers on page 14 of >this< official paper published by Intel.
AFAIK the correct DeviceID for your X58 RAID Controller is 3B25. That would be in hex code "25 3B".

Thanks but I looked at that data sheet and none of the chipsets mention the x58 platform. Are you sure that's correct for x58? I thought mine was 3A25?

Thanks

 

[Fernando 1]

Thanks but I looked at that data sheet and none of the chipsets mention the x58 platform. Are you sure that's correct for x58?

I am not sure about that, but I thought, that the X58 Express chipset belongs to the Intel 5-Series.
All official Intel 5-series SATA AHCI Controllers DeviceIDs begin with "3B".

 

[davidm71]

Ill try it out but the thing is in device manager for the Raid Controller properties the hardware is 8086 2822 with subsys 3A25. Maybe that has nothing to do with it..

Thanks

 

[tweakboy]

It's been a while since I've looked at trim but some good news, RAID0 trim seems possible with pre-7 series chipsets.

Tested on ASUS P67 EVO


Running 11.6 OROM and 11.2 driver with no modification on W7SP1x64 with 2x V4 in RAID0, sectors 16 thru 31 were filled with 8KiB of test data and some crude software written to send a trim command to unmap them. The command was rejected by the driver, no trim performed.
http://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/1.png


After some rough modification (needs sorting out when I have the time) for personal educational purposes , using the same V4 setup the LBA's of a picture file was recorded using Winhex then the file deleted.

Before deletion
http://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/2.png

After deletionhttp://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/3.png

Success, the file was trimmed as seen by all ones being returned for the trimmed LBA's. Note that returning all ones is specific to V4 SSD firmware, other SSD's may return all zero's instead.


Same setup with 2x 60GB Agility 3 in RAID0 and writing 8x 1GiB files of incompressible data.http://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/4.png

And after deletion.
http://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/5.png
If trim were not working we would still see the big dip in read performance.


Switching to W8DPx64 and 11.5 drivers using the 8KiB of test data written to sectors 16 thru 31.

Vertex4 RAID0 before
http://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/6.png

Vertex4 RAID0 after
http://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/7.png

Agility3 RAID0 beforehttp://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/8.png

Agility3 RAID0 after
http://i1142.photobucket.com/albums/n610/CPL0_at_Photobucket/Trim_P67/9.png
Note that Agility3 SSD firmware returns zero's for unmapped LBA's.

There does seem to be a little bugginess with the trim, even with Z77, so maybe Intel are working on that before making RAID0 trim more widely available, I hope so.

It would have been interesting to have looked at trying SRT on P67 and 5 series RAID0 trim but unfortunately I have little to no time for this for now.

Nice, but its gonna copy at around 600mbps,,, not worth the trouble. Wait 2014 and buy a potential 1000mbps single SSD. thx gl

Let's get rid of the screenshots
-ViRGE

 

[Evildeffy]

Nice, but its gonna copy at around 600mbps,,, not worth the trouble. Wait 2014 and buy a potential 1000mbps single SSD. thx gl

And this has exactly WHAT to do with the topic at hand?
We're not benching SSD speed here, we're praying for TRIM on chipsets prior to Z/Q/H-77.

 

[Fernando 1]

@ tweakboy:
Why do clutter up this interesting thread by fully quoting a post containing at least 9 screenshots, which by the way have nothing to do with your contribution.

 

[organshifter]

If I did understand you correcty, it may be possible to enable the "TRIM in RAID0" feature for all Intel RAID0 systems with an 6- or 7-series chipset by just modifying the related "FLAGS" hex code from "01" resp. "06" to "07".
Here is a screenshot with the related hex codes of the v11.6.0.1702 ROM file after having changed the "FLAGS" (= red hex code pairs):

Is this correct?

Kind Regards and thanks for this amazing thread
Fernando

I changed the flags to reflect exactly what you did above, and all went well. I then incorporated the OROM into the most recent bios for my board, and the flash went perfectly. However, after running the TRIM test here, it doesn't appear to be working. Initially, I tested with Intel RST v11.6.2.1002, and it didn't work. So, I removed that version and installed RST v11.6.0.1030. Same results.

Could this be a problem with Win 8 Pro, as there have been reports of TRIM/RAID 0 not working with it? Or maybe I shouldn't have modified as many flags? Any Ideas?

Motherboard: ASUS P8Z68 DELUXE/GEN3
SSD Drives: 2x 128GB Transcend SSD320 RAID0

In my case all I did was change "07 1C 00 00 01" to "07 1C 00 00 07" and "06 1C 00 00 01" to "06 1C 00 00 07".

Maybe for my Z68 board only the last one was needed, but I changed both just in case.

I'm starting to wonder if I should have changed only those two flags as well. Did you re-create your array?

 

[davidm71]

When those 1000mbs ssds come out you bet your bottom dollar someone will raid them up as raid is more than just speed but also redundancy and you will want your trim working!

 

[Fernando 1]

Could this be a problem with Win 8 Pro

Since the TRIM in RAID0 feature works for my Z77 system with Win7 and Win8, I think, that the OS will not be the problem.

 

[Soundmanred]

@ tweakboy:
Why do clutter up this interesting thread by fully quoting a post containing at least 9 screenshots, which by the way have nothing to do with your contribution.

I recommend using the "block user" feature, it will save you from the mind blowing stupidity.

 

[davidm71]

Any X58 people out there willing to give this mod a try? I tried without success. Pretty sure the x58 raid controller id is 3a25. Couldn't get trim working...

Thanks

 

[MaxICH]

Any X58 people out there willing to give this mod a try? I tried without success. Pretty sure the x58 raid controller id is 3a25. Couldn't get trim working...

Thanks

Did you try with 3b25 as suggested? The id you see in Device Manager is not the same you need to enter in the LUT. I have a Z68 and device manager says this:



while the right id I used is 1C06 (061C in the orom) which also matches what Intel says in this document at page 11.

 

[MaxICH]

I finally got time to take screenshots of the tests.

System info

Asus P8Z68-V Pro, BIOS 3402 with Intel orom 11.6.0.1702.

The orom was edited using the LUT info provided by Dufus as follows:

"07 1C 00 00 01" to "07 1C 00 00 07" and "06 1C 00 00 01" to "06 1C 00 00 07"​

then saved back in the BIOS and re-flashed. The SSD drives are 2 OCZ Vertez 4 256GB. The raid 0 was created before editing the orom.



Testing

Copied a jpeg file in the root of C: then used Winhex to check the drive



Deleted the file and emptied the recycle bin, then after few seconds reopened the drive in winhex pointing to the same location:



Just as Dufus said the Vertex 4 drives use FF instead of 00 when "trimming".

 

[Dufus]

Sorry for the late reply, I'm quite preoccupied with other things non computer related.

Fernando the alternate ID's are really the true ID of the controller, 2822 is used to make things easier by enabling a generic driver to be used. There are quite a few different ID's, for instance 7 series 1e04, 1e06, 1e07, 1e0e? and 6 series 1c04, 1c05, 1c06, 1c07?. I don't know what 1c07 is for, maybe used on servers . The 6 you see with some of the 7 series (starting with 1E) are probably for the 7 series boards that do not have SRT but have trim in RAID0. To incorporate many boards would mean expanding the LUT, something for Intel to do.

Evildeffy, that's a good question. I have no experience with X58 so unfortunately can not answer it. X58 comes before PCH so is not really the same as a normal 5 series, a bit like X79 is not really the same as normal 7 series.

davidm71, if the X58 is using a standard ICH10R then AFAIK the ID should be 3a05. ID's can be found in the Intel chipset specification update doc's down-loadable from Intel, if however you can check it with software then all the better. I've ran on both W8 (DP) and W7 with drivers from 11.0 to 11.5, although I prefer the RST driver over RSTe, so there should not be a problem utilizing trim on those systems. A windows trim command can also be sent successfully using vista with the 11 series drivers installed, it's just that MS did not implement or update the vista OS to send trim commands on delete/format.

MaxICH, nice work and thank you for showing trim working on your Z68 but please remember this testing was only done as "a proof of concept" and not as a means to provide trim support.

Some things to consider when running trim tests. Some drives have broken trim firmware. Trim is not always guaranteed to happen right away or sometimes even at all. I have noticed in some instances where a trim command was successfully sent but no changes were seen at the LBA level (includes 7 series) until after reboot or running a quick disk check (without autofix) or sometimes simply logging off for a couple of minutes. I don't know if this happens because of the OS/driver side or SSD firmware side, a mystery yet to be solved. If you have been running without trim for a while and finally get a working system remember to run some trim optimizing to clean up the older LBA's that had been deleted before. AFAIK W8 can do this via defrag which doesn't defrag SSD's but re-trims, however W7 will require 3rd party software.

Thanks to all who posted, appreciated. I'll try to update when I can but seems I have little "play" time these days.

 

[davidm71]

Dufus,

I'm pretty sure now that its 3a25 and not 3b25. That intel 5-series specification PDF doesn't mention x58 which has a different raid controller. Not only that the subsys id in device manager for the raid controller is 3A25. Second someone's lspci output in linux of their x58 system I found on a google search says 3a25 for their raid controller. Try googling x58 against 3a25 and you get hits where with 3b25 nothing. So I'm convinced I'm using the right device id. Anyhow thanks for your input and work. Maybe it's a delayed operation..

Thanks

 

[Fernando 1]

I'm pretty sure now that its 3a25 and not 3b25.

You obviously are right. I thought, that boards with an X58 chipset have 5-Series SATA Controller chips, but obviously they haven't.
By the way: >Here< is a list with the DeviceIDs of nearly all SATA Controllers.
My conclusion until now after having seen the test results layed down within this thread:

1. Users with an Intel 7-Series SATA RAID Controller automaticly get the TRIM in RAID0 feature by using an actual Intel RST driver(s)/RAID ROM combination v11.x.x.xxxx or higher.
2. Users with an Intel C600 (e.g. X79) Series SATA RAID Controller will get TRIM into their SSD RAID0, if they are able to use an actual Intel RST driver(s)/RAID ROM combination v11.x.x.xxxx (precondition: the mainboard manufacturer offers a BIOS containing such Intel RAID ROM version).
3. Intel 6-Series SATA RAID Controllers obviously are able to support TRIM in RAID0, but this feature is still locked by the hex code of the currently available Intel RAID ROM modules. It seems possible, that Intel will unlock the TRIM in RAID0 feature for 6-Series SATA RAID Controllers in the near future. Users, who do not want to wait for an Intel RAID ROM version with enhanced TRIM in RAID options, may get this feature already now by a simple hex code modification of the actual Intel RAID ROM v11.6.x.xxxx according to the findings of Dufus (see above).
4. The future will answer the question, if users with an Intel 3-Series, 4-Series, 5-Series SATA RAID Controller will get TRIM into their SSD RAID0.