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

[Dufus]

If anyone sees a mistake or has an advice how to make it better, please let me know it.

Well unless it's maybe using 1C07! Give me a while and I'll see if I can write something to read the real ID and try to take some of the unknowns out.

Hellhammer, if you follow Fernando's trim method first to see if it's working IMO that would be better than hammering your SSD's unnecessarily If it doesn't work with the simple test, especially if you've rebooted after deletion, then it's not likely to work anyway. Also trim itself just tells the firmware to unmap the LBA's, it is up to GC to do the erasing at it's pleasure. IOW your test is really a combination of trim and GC and might vary with different SSD firmwares depending on how they handle the trim.

I have tested both a 5-series (P55) and 6-series (P67) and confirmed trim working with drivers 11.0 to 11.5. Personally I prefer the RST drivers, 11.2 is not a bad choice IMHO.

 

[Ourasi]

Hmmm. Does the quick format execute trim? How long did you wait after formatting to test?

Fernando's latest OROM does indeed zero out my C300 array, cleans it right up imidiately and restores performance...

@ Hellhammer:
It is a pity, that you obviously did not get TRIM support within your RAID0 after having flashed a BIOS containing the Intel RAID ROM v.11.6.0.1702, which as been customized by me today.

@ all:
This is the related part of the unmodified OROM (uploaded by Dufus):

And this pic shows the changes I have done today (red hex code pairs):

If anyone sees a mistake or has an advice how to make it better, please let me know it.

TRIM works as it should do, but only on my Intel 520 array. There it zeros out the deleted file imidiately. But on my C300 array, it does nothing unless I quickformat. C300 array are on SATAII Intelports, and are a split array made in Intel Raid Bios, into 2 equal arrays for easy testing stuff. This may have something to do with it, but not likely since quickformat does everything right. Is it possible SATAII have a different ID, or is the C300 FW just not working right with TRIM-raid0, I do not know..

Anyway, here is the proof that Fernando's latest orom TRIM works on a P67 motherboard on the Intel SATAIII SSD 520 array:

Before delete/TRIM:


After delete/TRIM:


Edit: The C300 array does delete/TRIM the txt file, but only after an reboot. So, there we go, all my SSD arrays now TRIM-positive on my P67:

C300 array before delete/TRIM:


C300 array after delete/TRIM/reboot:

 

[Hellhammer]

Hellhammer, if you follow Fernando's trim method first to see if it's working IMO that would be better than hammering your SSD's unnecessarily If it doesn't work with the simple test, especially if you've rebooted after deletion, then it's not likely to work anyway. Also trim itself just tells the firmware to unmap the LBA's, it is up to GC to do the erasing at it's pleasure. IOW your test is really a combination of trim and GC and might vary with different SSD firmwares depending on how they handle the trim.

I have tested both a 5-series (P55) and 6-series (P67) and confirmed trim working with drivers 11.0 to 11.5. Personally I prefer the RST drivers, 11.2 is not a bad choice IMHO.

I tried the hex editor thing as well and the result was the same. I know how the drives I'm using handle TRIM since I've tested them, performance restored after running a TRIM pass.

 

[Dufus]

Here, try this.

SataID

Edit: If you get a chance to run that hellhammer maybe you can post the result, anyway, I'm off to get some sleep.

 

[Dufus]

Is it possible SATAII have a different ID

No, in RAID/AHCI it's all the same controller.

 

[Ourasi]

No, in RAID/AHCI it's all the same controller.

I was just speculating and typing things I knew it could not be, as quickformat did TRIM, but a simple filedelete did nothing. Anyway, C300 array have done the delete/TRIM after an reboot, so it's just what they do. So both my arrays is now working at TRIM/raid0 on my Asus Sabertooth P67..

 

[Fernando 1]

@ Sunburn74:

Here is the download link to the modified BIOS F10b for your GA-P67X-UD3-B3:
http://win-lite.de/benutzer/fernando//Diverses/p67xud3_mod_by_Fernando.7z

At first step I have opened the BIOS with the tool CBROM32_198. Then I replaced the original Intel RAID ROM v10.1.0.1008 by the modified v11.6.0.1702. Since the sensitive MINIT file was beneath the RAID ROM module, I have recovered the original BIOS structure at least.
Nevertheless I am not sure, that you will be able to flash this modified BIOS without getting a checksum error message.
Note: GIGABYTE is the only mainboard manufacturer, which doesn't follow the CBROM rules.

Good luck!

 

[Fernando 1]

Well unless it's maybe using 1C07!

That is what I am thinking too after having read the TRIM test results from Hellhammer, who didn't succeed with my newest OROM file from today (with unlocked 1C04) and Ourasi, who succeeded with my yesterday's OROM (with unlocked 1C07 instead).
Question:
Which DeviceID did you put into the RAID ROM hex code for your P67 TRIM test?

Fernando's latest OROM does indeed zero out my C300 array, cleans it right up imidiately and restores performance...

That is really good news, not only for yourself, but for us all! :awe:
Thank you very much for having tested the Intel RAID ROM module, which has been modified by me yesterday.

 

[Ourasi]

That is what I am thinking too after having read the TRIM test results from Hellhammer, who didn't succeed with my newest OROM file from today (with unlocked 1C04) and Ourasi, who succeeded with my yesterday's OROM (with unlocked 1C07 instead).

FYI: I actually tested with the orom posted 25.11.2012:1142 earlier this evening in post #102, so this is what I'm using now to modd for other users with P67.. Just so you know exactly what orom I used..

That is really good news, not only for yourself, but for us all! :awe:
Thank you very much for having tested the Intel RAID ROM module, which has been modified by me yesterday.

Yes, very good news indeed, and I'll try to remember to post who made it from now on
The Intel SSD 520 array did TRIM imidiately, with no delays.. The fact that the C300 array waits until reboot or logoff/on to TRIM files marked for deletion, is pretty smart when one/system is deleting during benching and highperformance tasks, as you never get a dip in performance. This was a linus torvalds biggest consern about TRIM onthefly a while back if I remember correctly..

 

[managerman1985]

1. Intel 7-Series and C600-Series (X79) SATA RAID Controllers get full "TRIM in RAID0" support by the original Intel RAID ROM v11.6.x.xxxx (no modding required).

Fernando,

Thanks for all your work regarding this issue. I need your guidance for my situation. I have an Asus Rampage IV extreme x79 motherboard. The latest bios from Asus is 3101 and has raid option rom 3.5.xxxx. Do you have a modified bios with 11.6 option rom? I am assuming I would need the 11.6 rom to enable trim on this board.

Any info would be great. Thanks

-M

 

[Sylver123]

Hi guys.

I am wondering if this would work on my ASUS Rampage III Extreme which has a x58 chipset.

Is it possible you experienced modders that know what you're doing could please mod the
Intel OROM 11.6.0.1702 for me with the ids I can find for my motherboard's onboard
Intel RAID Controller to see if RAID0 trim can be enable on this board too?

I would be very grateful if somebody could do this for me if you can. I went into the
device manager and here is all the information I could find for my Intel RAID Controller.

Hardware ids:

PCI\VEN_8086&DEV_2822&SUBSYS_82D41043&REV_00
PCI\VEN_8086&DEV_2822&SUBSYS_82D41043
PCI\VEN_8086&DEV_2822&CC_010400
PCI\VEN_8086&DEV_2822&CC_0104

Compatible ids:

PCI\VEN_8086&DEV_2822&REV_00
PCI\VEN_8086&DEV_2822
PCI\VEN_8086&CC_010400
PCI\VEN_8086&CC_0104
PCI\VEN_8086
PCI\CC_010400
PCI\CC_0104

 

[MaxICH]

I tried the hex editor thing as well and the result was the same. I know how the drives I'm using handle TRIM since I've tested them, performance restored after running a TRIM pass.

If you have a Z68 borad, try to use the orom with only the changes I posted in post #26:

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".

That worked for me and in theory should work for other Z68 boards.

 

[Engineer]

Here, try this.

SataID

Edit: If you get a chance to run that hellhammer maybe you can post the result, anyway, I'm off to get some sleep.

I just tried this on my Biostar P67 board and it came back with "Intel SATA controller not found".

 

[MaxICH]

I just tried this on my Biostar P67 board and it came back with "Intel SATA controller not found".

Same thing on my Z68

 

[ney2x]

I hope Hellhammer tried to use Windows 8 Defrag to optimize his SSD. That's what I did yesterday on my friend system which has 2 x 120Gb Samsung 840 SSD. Samsung SSD's are not Sandforce Controller, right? From what I learned from other forums, only Sandforce Controlllers has "aggresive" TRIM, which means TRIM kicks-in as soon as you delete/format a drive.

I noticed 1 flag difference from mine and from Fernando (I highlight them with blue).

Fernando's modification

My modified raid

ooopppss sorry, I didn't read Fernando's post # 93 about the image I posted.

 

[Brahmzy]

Great news Ourasi! TRIM IS WORKING on my P8P67 Deluxe Vertex 3 240 RAID0 array. I have verified 5 or 6 times using the hex editor. No reboots, no recreating the array. Just instaneous. I tried 3 times before flashing your BIOS and a definite no-go - hex showed stuff was still there. I waited 2-3 minutes and it didn't happen. I flashed your BIOS, re-setup all of my settings, tested again and voila - the hex editor showed an immediate TRIM job - like 5-6 times, each time successfully and immediately.

Awesome work! I owe ya one!

 

[Fernando 1]

FYI: I actually tested with the orom posted 25.11.2012:1142 earlier this evening in post #102, so this is what I'm using now to modd for other users with P67.. Just so you know exactly what orom I used...

Thanks for this information.

@ all, who have already downloaded any of my modded Intel RAID ROM files:
Until now I have offered at least 2 differently modified Intel RAID ROM v11.6.0.1702 files with the following names and details:

• "8086-2822_v11601702_mod_for_6-Series_TRIM.bin" dated 11/23/2012 at 17:08
  Details: Within this ROM module I have set the FLAGS for the DeviceID's 1C06 and 1C07 to "7" (TRIM support enabled).
• "8086-2822_v11601702_mod_for_P67+Z68.bin" dated 11/25/2012 at 11:42
  Details: Within this module TRIM support I have enabled TRIM support for the DeviceID's 1C04 and 1C06.

So both files hopefully will support TRIM for the Intel SATA RAID Controllers with the DeviceID 1C06 (should be Z68), whereas the other supported DeviceID is different (1C07 resp. 1C04).
Until now we do not know exactly, which 6-Series chipsets are using the DeviceIDs 1C07 and 1C04.

EDIT:

@ managerman1985:
Users with an X79 chipset RAID0 have to wait until the manufacturer of their mainboard offers a BIOS with included Intel RAID ROM v11.x.x.xxxx. The "normal" Intel RAID ROM modules v3.x.x.xxxx for X79 chipsets do not support TRIM in RAID0 and a switch to the v11.x.x.xxxx can not be done by modding the ROM file.

@ Silver123:
Since we do not yet even know the correct internal DeviceID of the Intel X58 chipset SATA RAID Controllers, we are not able to unlock the TRIM in RAID0 feature for X58 RAID0 systems. Furthermore I doubt, that this will be possible by simply modifying some hex codes of the RAID ROM file.

 

[Dufus]

I just tried this on my Biostar P67 board and it came back with "Intel SATA controller not found".

Thank you Engineer and MaxICH, appreciate the feedback. Although I was not expecting it to work on older systems it's interesting you got that message, thanks.

I've updated it so hopefully it should work even on the older boards. davidm71 maybe you would try this on your X58

Hellhammer, I understand you have your own testing method and that is fine. I'm not trying to knock it, just providing some info on trim and GC as well as appreciating that you already know what to expect when trim is working with your own methods.

 

[Fernando 1]

I've updated it so hopefully it should work even on the older boards.

Contrary to the previous version this updated tool shows the "normal" and the alternate DeviceID of my Z77 SATA AHCI Controller:

Thank you very much for offering this very useful small application, which may help not only the users, but the supporters of the Forums too.

 

[Engineer]

Thank you Engineer and MaxICH, appreciate the feedback. Although I was not expecting it to work on older systems it's interesting you got that message, thanks.

I've updated it so hopefully it should work even on the older boards. davidm71 maybe you would try this on your X58

Hellhammer, I understand you have your own testing method and that is fine. I'm not trying to knock it, just providing some info on trim and GC as well as appreciating that you already know what to expect when trim is working with your own methods.

My Biostar P67 now reads 1C02. (Both ID's).

 

[Ourasi]

Contrary to the previous version this updated tool shows the "normal" and the alternate DeviceID of my Z77 SATA AHCI Controller:

Thank you very much for offering this very useful small application, which may help not only the users, but the supporters of the Forums too.

This explains why your 25.11.2012:11.42 orom was the only one that worked on my P67 Fernando:

 

[Fernando 1]

My Biostar P67 now reads 1C02. (Both ID's).

This verifies, that you are currently running your HDD/SSD in AHCI mode.

This explains why your 25.11.2012:11.42 orom was the only one that worked on my P67 Fernando:

Thanks for this important feedback, which verifies, that we obviously are on the right track.
Nevertheless there is still 1 question open:
Why didn't Hellhammer with his P67 system get TRIM working?
AFAIK he has used exactly the same modified ROM as Brahmzy and you.

 

[Hellhammer]

Here, try this.

SataID

Edit: If you get a chance to run that hellhammer maybe you can post the result, anyway, I'm off to get some sleep.

If I've understood correctly, only device IDs 1C04 and 1C06 are supported? Fernando, any chance you could add support for 1C02 as well if it's not supported yet?

I hope Hellhammer tried to use Windows 8 Defrag to optimize his SSD. That's what I did yesterday on my friend system which has 2 x 120Gb Samsung 840 SSD. Samsung SSD's are not Sandforce Controller, right? From what I learned from other forums, only Sandforce Controlllers has "aggresive" TRIM, which means TRIM kicks-in as soon as you delete/format a drive.

Like I said earlier, formatting the drive issues a TRIM command to all user-accessible LBAs and it's instantaneous. It works with a single SSD so it should work with RAID 0 as well, assuming that TRIM works. Anand did this with his Z77 based setup and it worked similarly to how a single SSD would behave.

 

[Fernando 1]

If I've understood correctly, only device IDs 1C04 and 1C06 are supported? Fernando, any chance you could add support for 1C02 as well if it's not supported yet?

That is very mysterious. Do you have a RAID0 array at all?
1C02 is the DeviceID of the "Intel(R) Desktop/Workstation/Server Express Chipset SATA AHCI Controller", which should support TRIM automaticly under Win7/Win8.
So there is no need to modify the Intel RAID ROM for your system.

 

[MaxICH]

Thank you Engineer and MaxICH, appreciate the feedback. Although I was not expecting it to work on older systems it's interesting you got that message, thanks.

I've updated it so hopefully it should work even on the older boards. davidm71 maybe you would try this on your X58

Hellhammer, I understand you have your own testing method and that is fine. I'm not trying to knock it, just providing some info on trim and GC as well as appreciating that you already know what to expect when trim is working with your own methods.

The new version reports:

Controller id: 2822
Alternate id: 1c06