Performance of the Intel RST/RSTe AHCI/RAID Drivers

Hi Fernando,

The Samsung 840 PRO with Rapid mode enabled, OROM 12.7.0.1936 and driver 12.8.2.1000 rocks…check out the results I just recorded…no stuttering either !!

delprong

Samsung 840 PRO Rapid 4.3 - 02.png



Rapid mode means those results are about how fast is your RAM which is obviously much faster than any SSD :wink:

@Fernando
Here are my experience with Magician RAPID mode on P8Z77-V DELUXE AHCI W7-x64 SP1 840 PRO 256GB SSD system.
Figures are quite ‘extraordinary’…

840 PRO AHCI_RAPID_ANVIL.png

840PRO AHCI_RAPID_ATTOBENCH.PNG

840PRO AHCI_RAPID_CDM.PNG

840PRO AHCI_STANDARD_MAGICIAN43.PNG

RAPID2.PNG

Fernando,
did you get this strange report comparaison (sometimes very low IOPS value) get with Magician 4.3 ?

RAPID_History.PNG

I haven`t tested it yet, but sometimes I get intensive stuttering.

is win8.1 uefi install possible icw rapid mode?


Yes, the OS Win8.1 x64 had been installed in "clean" UEFI mode (CSM disabled) and the Samsung RAPID mode runs fine.



Which config has the best response in win8.1 for you: 1x 840 pro in rapid mode or 2x 840 pro in raid-0 ?

As you can see >here<, the benchmark results with a single Samsung 840 PRO running in RAPID mode are far better than those I got with both Samsung 840 PRO SSDs as members of a RAID0 array.
Nevertheless I am not totally satisfied with the Samsung RAPID mode. The user doesn’t get the full benefit from the RAM caching speed done by the Samsung RAPID mode and shown by the synthetic benchmark results. My system was snappier as long as I ran the SSDs as RAID0 array.

Hmm, my results give me mixed feelings. 4K went down but 4K-64Thrd went up. Seq stayed the same. So, not sure which is more important, 4K-64Thrd or just 4K.

11.2.0.1006



12.8.10.1005


So far stability seems just fine to me. Need to try MSAHCI, if I can figure out how that is done. These results make it clear it is time to upgrade my motherboard, haha.

Yes, I tested the RST(e) drivers v12.9.0.1001 with my Z77 AHCI system as well.

Here are the AS_SSD benchmark results I got by using the following AHCI drivers in non-RAPID mode (from the left to the right:
1. Win8.1 stock MS AHCI driver (STORAHCI.SYS), 2. Intel AHCI driver RST(e) v10.8.11.1000 WHQL, 3. Intel AHCI driver RST(e) v12.9.0.1001 WHQL)

[[File:AS-SSD_512GB-Samsung-840-Pro_Win8.1-AHCI_RST(e)-v12.8.11.1000.png|none|auto]][[File:AS-SSD_512GB-Samsung-840-Pro_Win8.1-AHCI_RST(e)-v12.9.0.1001.png|none|auto]]
These results verify, that the performance of all 3 AHCI drivers is nearly identical, when the SSD is running in clean AHCI mode (without additional RAM caching).
Note: This conclusion is only valid for SSDs running in AHCI mode and not for SSDs, which are members of a RAID0 array.

On the other hand I got interesting results with the same drivers after having enabled the Samsung Magician RAPID mode (from the left to the right:
1. Win8.1 stock MS AHCI driver (STORAHCI.SYS), 2. Intel AHCI driver RST(e) v10.8.11.1000 WHQL, 3. Intel AHCI driver RST(e) v12.9.0.1001 WHQL)

[[File:AS-SSD_512GB-Samsung-840-Pro_Win8.1-AHCI_RST(e)-v12.8.11.1000_RAPID-mode.png|none|auto]][[File:AS-SSD_512GB-Samsung-840-Pro_Win8.1-AHCI_RST(e)-v12.9.0.1001_RAPID-mode.png|none|auto]]

As you can see there are big performance differences between the tested 3 AHCI drivers, if they are running in RAPID mode.

Apart from the nearly unbelievable benchmark results I got running my SSD in RAPID mode, these are the main reasons why I actually prefer the Intel RST(e) drivers v12.8.11.1000 WHQL for my AHCI system:

  1. They are the newest (although "newest" not always mean "best").
  2. Intel has spent a lot of time and efforts to fix the bugs of the v12.8 series. These were the steps until now: v12.8.0.1008 > v12.8.0.1016 > v12.8.1.1000 > v12.8.2.1000 > v12.8.4.1000 > v12.8.6.1000 > v12.8.7.1000 > v12.8.10.1005 > 12.8.11.1000. This development history verifies, that the v12.8 series branch obviously is a very good and interesting one.

You showed the AHCI and RAPID performance difference, but I wonder what is the situation in pure RAID 0…?

Please have a look into the post #2 of this thread (>LINK<).
Although I haven’t yet tested the newest Intel RAID drivers v12.8.11.1000 and v12.9.0.1001 with a RAID0 system, I am pretty sure, that the latest "conservative" iaStor.sys RST driver v11.2.0.1006 will still be the "best" driver for RAID0 systems regarding the performance.

Please have a look into the post #2 of this thread (>LINK<).
Although I haven’t yet tested the newest Intel RAID drivers v12.8.11.1000 and v12.9.0.1001 with a RAID0 system, I am pretty sure, that the latest "conservative" iaStor.sys RST driver v11.2.0.1006 will still be the "best" driver for RAID0 systems regarding the performance.




I meant for a comparison of v12.8.11.1000 and v12.9.0.1001. I agree the v11.2.0.1006 is maybe fastest, but on my SF drives in RAID I do get some bizarre lags (micro freeze-stuttering) once in a while in combination withe newest OROMs, so I don’t use them since no lag with newest…

Maybe I will do some additional benchmark tests once I have recreated a RAID0 system.

Fernando,

I don"t understand why with a very similar ASUS board and AHCI SSD 840 PRO 256 GB configuration i get a very different and less performance result as you have reported.
I get 4333 index and you report 6811.
(on my side only the iRST driver is installed, iRSTe package has been installed for setting SATA link option "disabled" and then i have uninstalled the package).

V128111000_AHCI_RAPIDmode_test5.PNG



Do you see the any reasons why this huge difference ?

But you got far better Read scores than me.
Note: Our systems are not similar:

  1. My tests have been done with a brandnew 512 GB Samsung 840 PRO.
  2. Furthermore I have added 2x4 GB SDRAM, so my system now has at least 16 GB of RAM.
  3. All my tests were done on a fresh installed Win8.1 x64 with integrated Spring Update 2014 (RTM version).

Where and why did you disabled this option?

Our hardware is different and the individual settings as well.

Fernando,
I assumed your signature was up to date.
A 512GB SAMSUNG PRO vs 256 GB cannot explain a 60% better result …
My system has 16 GB RAM of PC3-17100 DDR3 SDRAM UDIMM, so it is the same size you have.

You recommended in the past in some threats to disable the ‘sata link’ option in the iRSTe software application in ‘Performance menu’, to get better performance.
This option is also well commented in iRSTe pakkage when select “?” icon with the mouse. Disabling this option should improvre the performance…

Yes, you have W8.1 x64 but this does not explain the 60 % more …

This is what ANVIL test gives:

840 PRO AHCI_RAPID_ANVIL.png



Do you get a similar result on your side ?

Fernando,

Do you have done an ANVIL, RAPID mode test with the 512 GB SSD configuration ?
If AS SSD is 60 % better, ANVIL test sould be also about 60 % better, i.e. about (15000 + 60%)= 24000 ??
Did you get a such ‘extraordinary’ result ?

EDIT by Fernando: Double post deleted

Not yet.

Do you really believe, that my 512 GB sized Samsung SSD is 60% faster than your 256 GB SSD?
Please remember: These are synthetic benchmark tests, which do not have much to do with the real work of a "standard" user with his PC.

Fernando are you still using RAPID with your Samsung ?

I gave it up a few weeks after it came out.


FYI: New 840 / 840 Pro Firmware is out

New Firmware for Samsung 840 / 840 Pro (Mar’ 2014)

Yes, I am still testing this Samsung Magician feature.
Big advantage: It is very easy to disable and enable it.

Thanks for the info!
Have you already flashed it?