Raid0 and TRIM Support on Intel 11.5 OROM and Latest v11.5.0.1184 RST Drivers ?

[Fernando]

Question how about older chipset like mine Asus Rampage 3 extreme X58 ? do you know if OROM 11.5 it will work ok on this chipset x58 ? and also latest RST Driver 11.5 ?

Although I don't have an Intel x58 chipset system, I am sure, that the Intel RST(e) drivers of the v11.5 series will support it (the needed HardwareID's are present within the INF files). The Intel RAID ROM module v11.5.0.1414 will be usable too.
The only thing, which has to be investigated, is the behaviour of the x58 chipset SATA Controller regarding the SCSI filter driver named iaStorF.sys. My Intel Z68 chipset system is working without the filter driver in Win8 (the RAID0 is no SCSI disk device), that is why the Trim command may be able to pass through the RAID Controller.

I am awaiting your test results.

Regards
Fernando

[chispy]

Ok Fernando here i go For my testing using ROM 11.5.01414 and RST v11.5.0.1184, on page 1 you can see my speeds using the moded Bios done by ZioGTS for Asus R3E x58 Asus Rampage III Extreme 1601_S02 ZioMod , using ATTO and AS SSD benchmarks on Intel x58 ICHR10 Sata 2 Raid0 128k stripe size. for references i will re-run it again now before i clone this drive , i will be testing using w7 x64bit for 1 week doing heavy benchmarks to see if it is working properly and if maybe Trim works in w7 x64bit or if i get at least a speed boost using windows 7 , Sorry i hate w8 so no testing will be done on w8 as i will not use it anyways. Give me a few hours for back up , cloning and flash Bios Raid OROM , Clean Erase the drives and a fresh new intall of w7x64 bit Raid0 i will post ATTO and AS SSD after im done so that we can compare


" This is the latest available 1601 R3E beta bios with updated IRST, Marvell, Jmicron OROMs and injected SLIC 2.1 tables:

Intel® ICH10R SATA RAID Controller: v11.2.0.1527
JMicron JMB36X Controller: v1.07.28
Marvell 91xx SATA 6G Controller - Bios : v1.0.0.1033*
Marvell 91xx SATA 6G Controller - Firmware: v2.2.0.1125*
Marvell 91xx SATA 6G Controller - Bootloader: v1.0.1.0002b*
SLIC 2.1: ASUS (SSV3/MMtool method) "
Asus Rampage III Extreme 1601_S02 ZioMod
Here-http://www.bios-mods.com/forum/Thread-Asus-Rampage-III-Extreme-1601-S02-ZioMod-OROM-modules

[chispy]

Dual Post

[acebmxer]

If anyone with a Gigabyte motherboard looking for a modified bios can find them here http://forums.tweaktown.com/gigabyte...fied-bios.html.

[chispy]

Intel Raid ROM V.11.5.0.1414
Intel RST Driver 11.2.0.1006
Test # 1
Asus R3E X58 ICHR10
W7x64 bit
Raid0 128k Stripe size used
AS SSD
ATTO

7/18/2012 11.55PM


By chispy at 2012-07-18


By chispy at 2012-07-18


By chispy at 2012-07-18

Findings: Bios Raid ROM 11.5.0.1414 its faster than Bios Raid ROM 11.2 even with RST Driver 11.2.0.1006 as shown in the bench tests AS SSD and ATTO. It wont let me update to Intel Driver RST 11.5 , the system says it's using the latest driver and it's up to date. I will now re-install w7 x64bit installling RST Driver pressing f6 during w7x64 fresh installation and i will report back in a few hours.



Second Part

Test # 2

Intel Raid ROM V.11.5.0.1414
Intel RST Driver 11.5.0.1184
Test # 2
Asus R3E X58 ICHR10
2x Samsung 830 128GB SSDs Raid0 , Secure Erased Both SSDs
W7x64 SP1 bit Clean Fresh Install with all updates
Raid0 128k Stripe size used
AS SSD
ATTO

7/19/2012 3:32AM

AS SSD


By chispy at 2012-07-19

ATTO


By chispy at 2012-07-19

My Findings and Conclusion: More Performance out of the combination of using both , Bios ROM 11.5 and RST Driver 11.5 , so far this is the best combination of driver RST and Bios Raid ROM for a good bump in performance , Writes did improved the most. Much more ballanced performance across. Very stable so far. I will hammered it with a lot of work and benchmarks for one week to see how the performance has degraded , or if by miracle some kind of TRIM or garbage collection its past trhu the Raid0 but as simple drives and not as a Raid0 array , time and testing will tell us next week wednesday 25th of July i will post the same benchmarks AS SSD and ATTO to see how far degradation has taken this array in one week thru hell.


--------------------------------------------------------------------------

Test # 3 and conclusion.

7/25/2012 10:34AM

One week has passed since i have started torture testing my Raid0 array , testing mode used as follow: 3 consecutive runs of each benchmark 5 times a day for a whole week AS SSD , ATTO , Cristal Disk , Anvil Storage Utilities , HD Tune , HDD Tach , Lots of writes and Lots of Reads + Large transfer files of BR Movies and huge pictures from diferent drives to and from the Raid0 array non stop for hours on end each day for 7 days , drive was used 24/7. Here its the real time performance of the same Raid0 array of 2x 128 Samsung SSDs on Intel ICHR10 x58 Using Intel Raid ROM V.11.5.0.1414 , Intel RST Driver 11.5.0.1184.

AS SSD


By chispy at 2012-07-25

ATTO


By chispy at 2012-07-25


Final Conclusion and my own opinion / understanding: YES some kind of Trim and or garbage collection its working under w7x64bit restoring performance back to optimal speeds , It is to my opinion and understanding that under w7 x64 there some new files within the driver that Fernando discover earlier namely C:\windows\system32\DRIVERS\iaStorF.sys , this file its some kind of new filter for the function of the new 11.5 RST Driver , this filter must be seen the SCSI device in the Raid0 array as single drives and passing TRIM and or some kind of Garbage Collection as it is the only logical conclusion after the rigorous torture testing done for 7 days to this Raid0 array , i observe very bad degradation on Reads and Writes after the 5th pass of all the bencmark test , the system Reads went down to 294 max reads and 232 Max writes on AS SSD and ATTO , but once i let iddle for a couple of hours the Performance went back up and was completely restored to optimal speeds. I will try to contact Intel for an explanation and if they answer my questions i will post them here , feel free to do your own testing , im done with mine and i did prove to myself that yes definitely something its improving performance back after iddle time of about 2 hours of iddle system time. Feel free to discuss my findings and if you have a better theory and understanding with real data as proff + extensive testing please let us discuss here , Long live W7 x64bit :thumbup:



By chispy at 2012-07-25

[Fernando]

Findings: Bios Raid ROM 11.5.0.1414 its faster than Bios Raid ROM 11.2 even with RST Driver 11.2.0.1006 as shown in the bench tests AS SSD and ATTO.

This is an interesting result.

It wont let me update to Intel Driver RST 11.5 , the system says it's using the latest driver and it's up to date.

The MS Operating Systems prefer WHQL certified drivers. That is why you get the misleading message, if you try to replace a WHQL driver by a driver without WHQL certifcate.
This is the way to force the driver installation of an unsigned driver: Right click onto the Controller > "Update driver software" > "Browse my computer ...." > "Let me pick ...." > "Have Disk" > Point to the related INF file of the desired driver > "OK".

[chispy]

Fernando i updated my last post with the new benchamarks and much beter numbers using Intel Bios ROM 11.5.0.1414 in combination with RST Driver 11.5.0.1184 take a look at the jump in performance on those writes , so far so good.

[chinobino]

Interesting results Chispy.

I found I got the best results so far from the Intel RST driver v11.2.0.1006 with OROM v11.2.0.1527 (already in the latest BIOS for my motherboard).

I am extremely close to the maximum performance I can possibly get from 4 x 80GB Intel X-25m in RAID0 (using the Intel ICH).

Intel Raid ROM 11.2.0.1527
Intel RST Driver 11.2.0.1006
Test # 1
ASRock Z77 Fatal1ty Pro (BIOS 1.30)
Windows 8 RP x64 bit
Raid0 32k Stripe size used
AS SSD

Windows Write cache buffer flushing disabled
windows_8_rp_intel_rst_11.2.0.1006_z77_orom_11.2.0.1527_write_cache_buffer_flushing_disabled.png

Intel Raid ROM 11.2.0.1527
Intel RST Driver 11.2.0.1006
Test # 2
ASRock Z77 Fatal1ty Pro (BIOS 1.30)
Windows 8 RP x64 bit
Raid0 32k Stripe size used
AS SSD

Windows Write cache buffer flushing enabled
windows_8_rp_intel_rst_11.2.0.1006_z77_orom_11.2.0.1527_write_cache_buffer_flushing_enabled.png

[andressergio]

this are my results on the M5G Z77 with 2x Intel 520 series SataIII 120GB

BEFORE

BIOS 0903 + Intel® RAID ROM 11.0.0.1339 + Intel® Rapid Storage Technology 11.1.0.1006 RAID (vanila BIOS not modded)



AFTER

BIOS 1101 + Intel® RAID ROM 11.5.0.1414 + Intel® Rapid Storage Technology 11.5.0.1184 Beta (modded BY stasio (BIG THANKS !!!)




overall improved will keep testing !!!

[Fernando]

Now We have That next release driver the Intel talked about out in the wild wich its version v11.5.0.1184 RST Drivers , my question is does it enable Trim over Raid0 yes or no ? Does it work in both OS Windos 8 and Windows 7 ? If anyone has a definite answer with data and proff please share your findings

Maybe you will get an answer to your question by my recent contribution at station-drivers (look >here<).

@ andressergio:
That looks really good!

@ all:
Although the users with an Intel RAID0 system already may know it, I want to mention, that the following settings are necessary to get the best Intel RAID0 write performance:

1. enable both write cache options within the Device Manager ("Drives" > "Intel RAID0" device >"Properties" > tab "Policies")
2. enable "Write Back Caching" from within the Intel RSTe Console ("Adminstration" > "Advanced")

[Zaxx]

@ chisby...afaik Windows 7 is not capable of passing the trim command (I forget the actual reason) to raided drives...it's gonna be a Windows 8 only thing I'm afraid. I have no problem with MS tinkering with the user interface (i.e. Metro and no start button) BUT I can't for the life of me figure out why they didn't make it possible to switch back to the original desktop style for those who prefer it (~90% imo) instead of the touch screen geared Metro..."I don't want no stinking tablet!"

[andressergio]

Maybe you will get an answer to your question by my recent contribution at station-drivers (look >here<).

@ andressergio:
That looks really good!

@ all:
Although the users with an Intel RAID0 system already may know it, I want to mention, that the following settings are necessary to get the best Intel RAID0 write performance:

1. enable both write cache options within the Device Manager ("Drives" > "Intel RAID0" device >"Properties" > tab "Policies")
2. enable "Write Back Caching" from within the Intel RSTe Console ("Adminstration" > "Advanced")

thanks Fernando

looks good with new ROM and drivers...i think i had better reading with no 11.5 drivers but i have to try it cauz if you see the 4k section improved a lot !

cheers !!!
Sergio

[andressergio]

@ chisby...afaik Windows 7 is not capable of passing the trim command (I forget the actual reason) to raided drives...it's gonna be a Windows 8 only thing I'm afraid. I have no problem with MS tinkering with the user interface (i.e. Metro and no start button) BUT I can't for the life of me figure out why they didn't make it possible to switch back to the original desktop style for those who prefer it (~90% imo) instead of the touch screen geared Metro..."I don't want no stinking tablet!"

are using W8x64 Zaxx i got tired of the damn lockups that lot of ppl have...

[chispy]

Guys i keep hammering the Raid0 array daily torture test , with all benchmarks available , also doing a lot of writes and reads all day , transfering huge images , pictures , BR Movies and what not , you guys will be surprised at my findings next wednesday i will have some proff , until then i will keep the torture going non stop , to check for Raid0 degradation in performance if there is any

[Computurd]

Chispy, there is a simple test that you can do to see if TRIM is working that will not take as long, it is just a simple run of PCMV hdd suite when the drives are freshly sanitary erased. Then, write the entire capacity of the drive to 100% fill, then delete. Run PCMV HDD suite again, and compare the results

[Fernando]

@ all:

Just for your information:
Yesterday night Intel has published new official RST(e) drivers v11.5.0.1207 WHQL. You can get them from >here<.

[radier]

Thnx for the info !

[chispy]

*** Updated My Post Number 30 with my findings and conclusion after extensive testing and data collection for 7 days with Regards to Trim and or Garbage Collection on W7 x64bit , a big yes it is working.

@ Computurd , i will try that later today and will post back the results.

im still amaze of those results and I believe Intel needs to disclose information regarding this matter , i will try to contact them and if i get an answer i will post it here , lets keep discusing this subject i am intrigue.

[Zaxx]

@ all:

Just for your information:
Yesterday night Intel has published new official RST(e) drivers v11.5.0.1207 WHQL. You can get them from >here<.

Big thx for the info. Well, I have one half of the equation...now to find a 11.5 orom bios for an Asus P8Z68-V Pro Gen3. I'm getting curious now.

[Fernando]

[

*** Updated My Post Number 30 with my findings and conclusion after extensive testing and data collection for 7 days with Regards to Trim and or Garbage Collection on W7 x64bit , a big yes it is working.

That is really very good news! Thank you for your efforts to test it!

..now to find a 11.5 orom bios for an Asus P8Z68-V Pro Gen3. I'm getting curious now.

Why don't you do it yourself? It is very easy, if you follow my guide (>LINK<). The required Intel RAID ROM v11.5.0.1414 can be downloaded from >here<.

[m.oreilly]

win7 does not issue an unmap command. can't do it. only win8 can. win7 is missing this option in it's api. chisp, are you running win8?

[andressergio]

@ all:

Just for your information:
Yesterday night Intel has published new official RST(e) drivers v11.5.0.1207 WHQL. You can get them from >here<.

wow cool thanks Fernando !!!

still no clue if TRIM works for RAID0 on W7 right ? i read a zillion threads but only clue i get is to look in CristalDiskInfo and says YES but how do you execute it or works alone ? i remember that for my intel there's the SSD toolbox but i dont know if they will release a newer one

cheers
Sergio

[chispy]

Hola MR. Oreilly im running W7x64bit SP1 all Updates , I'm so intrigue by my findings and still trying to find a logical explanation for this matter. But as per my long 7 days long test it Does shows Performance its back up to optimal speeds after the System iddle time for ~ 2 hours , " from my post # 30 , 7 Day Test """" i observe very bad degradation on Reads and Writes after the 5th pass of all the bencmark test , the system Reads went down to 294 max reads and 232 Max writes on AS SSD and ATTO , but once i let iddle for a ~couple of hours the Performance went back up and was completely restored to optimal speeds"""

[andressergio]

Hola MR. Oreilly im running W7x64bit SP1 all Updates , I'm so intrigue by my findings and still trying to find a logical explanation for this matter. But as per my long 7 days long test it Does shows Performance its back up to optimal speeds after the System iddle time for ~ 2 hours , " from my post # 30 , 7 Day Test """" i observe very bad degradation on Reads and Writes after the 5th pass of all the bencmark test , the system Reads went down to 294 max reads and 232 Max writes on AS SSD and ATTO , but once i let iddle for a ~couple of hours the Performance went back up and was completely restored to optimal speeds"""

i just observed that ASSD benchmark varies a lot on every shoot i give, im not sure if trim is working...using RAID ROM 11.5 latest + latest drivers Fernando publlished

btw on this combo my 4k write went a up a lot ! i got a score of 1180 now sometimes more sometimes less

[Zaxx]

win7 does not issue an unmap command. can't do it. only win8 can. win7 is missing this option in it's api. chisp, are you running win8?

That what I thought...posted on previous page. Thus far, being able to trim raided drives array isn't NEAR enuff to force me to use Metro...unless they make the old UI an option I'm not budging from 7. My array has been doing just fine thus far.