New Intel RST drivers v9.5.6.1001 available

[DZand]

At Station-Drivers there are new Intel AHCI and RAID drivers v9.5.6.1001 available. They are dated 12/30/2009 and belong to the new Intel(R) Rapid Storage Technology. You can get the new RST Console and driver pack here: http://www.station-drivers.com/page/intel%20raid.htm http://www.station-drivers.com/page/intel%20raid.htm

Currently only the 64bit driver version is available.

The new RST Console and driver v9.5.6.1001 does support all Intel ICH7-10 and 5-series Southbridges running in AHCI and RAID Mode.

I installed the new 64bit driver v9.5.6.1001 onto my Intel ICH10R system running 2x160GB Postville SSD's as RAID0. My first impression is very good.

It would be interseting to know, if these drivers do support the Trim command.

Regards

Fernando

[VPent]

Hi Bobins and UglyPercy,

I checked DCP latency in my system and the max value, in only 2 or 3 spikes, was ~4000 us, or 4 ms, under the following conditions:

- running a antivirus scan, and

- streaming a video

- a period of aprox.50 min.

- disregarding the 2 or 3 spikes the normal latency was ~100us

So, in my case the latency is not an issue, and i think the problem is a little more complex, since it depends of what is the load (applications running, if the antivirus is load intensive, etc) on the computer, and of course what is the computer's hardware resources.

For video and audio streaming, there is, at least, another parameter to be considered: the latency between the server from which we are dowloading the stream and the computer, which typically is of 50 - 300ms or 50,000us - 3000,000us, and that can cause audio/ video stutterings .

Also i think the "diagnostic" given by DPC Latency Checker is too simplified, misleading, and alarmist, in my case: "Some device drivers on this machine behave bad and will probably cause drop-outs in real-time audio and/or video streams. To isolate the misbehaving driver use ..."

From what i understood this "diagnostic" is not considering that: the spikes only happened 2 or 3 times in a period of almost one hour, the buffer of several seconds that almost all "real time" applications have, and that even good harddrives have an access time of 9,000- 18,000 us, so comparable to the spikes.

And of course, many things besides "drivers behaving badly" may cause an increased latency on a computer.

As an objective result, i have never observed in my system any stuttering effects in audio/video streams.

Sorry for invading this thread that is supposed to be about Trim support; maybe the latency subject deserves an exclusive thread...

My two cents...

regards

system: 2x 80GB G2 Intel SSD in raid conf.

Intel ICH10R sata raid controller/ driver version 9.5.6.1001

W7 64 bit

[idata]

Very good post, valdir. I should clarify briefly (my system specs are referenced in earlier postings): I upgraded from 9.5.4 to 9.5.6, and made various other changes to my system. It was a few days before I noticed that, playing local AVI files in VLC Media Player, videos had begun freezing randomly, say for 1/4 second every 5-10 seconds. It took hours for me to find references to DPC latency and "dpclat", which pronounced my system defective within about 10 seconds every time I ran it. I downgraded to 9.5.4 and am back to normal. Unfortunately, I have two Corsair SSD's in a striped RAID plus four Seagate 1.5TB's in a RAID-10, so I can't say whether this is an SSD or mechanical hard drive issue, or both.

[idata]

I saw a performance decrease on my 160gb m25 g2. AS SSD benchmark dropped 9 points in write and 1 point in read compared to the generic MS AHCI driver.

[DZand]

mjemirzian schrieb:

I saw a performance decrease on my 160gb m25 g2. AS SSD benchmark dropped 9 points in write and 1 point in read compared to the generic MS AHCI driver.

This means nothing, because the differences are so small, that you will not feel them while working. Furthermore you will never get exactly the same benchmark result even with the same driver.

Much more important is the stability of the system. What is your experience with the RST driver v9.5.6.1001 compared with the MS AHCI driver?

[idata]

Small increments may not be noticable, but they add up. If you ignore 10 'small' speed tweaks out of 10, your system will be noticably slower even if each individual step is a small one.

9 points is a significant difference in AS SSD, about 5 MB/s. The sequential write test dropped from 104.29 MB/s to 101.60 MB/s,and the '4k' write test dropped from 43.70 MB/s to 38.70 MB/s, and write access time increase by .012 ms. This is far more than a 'random variation' in testing. You WILL notice a 5 MB/s write difference if you are copying large files.

>What is your experience with the RST driver v9.5.6.1001 compared with the MS AHCI driver?

I did not notice a difference in the stability of the system. I didn't use the 9.5.6.1001 driver long enough to see whether TRIM was enabled or not, but if it does not have TRIM support that's just another reason to not use it.

[DZand]

mjemirzian schrieb:

Small increments may not be noticable, but they add up. If you ignore 10 'small' speed tweaks out of 10, your system will be noticably slower even if each individual step is a small one.

9 points is a significant difference in AS SSD, about 5 MB/s. The sequential write test dropped from 104.29 MB/s to 101.60 MB/s,and the '4k' write test dropped from 43.70 MB/s to 38.70 MB/s, and write access time increase by .012 ms. This is far more than a 'random variation' in testing. You WILL notice a 5 MB/s write difference if you are copying large files.

You should consider, that benchmark results are virtual ones. Furthermore there is no single driver, which is the "best" for all different hardware/OS/BIOS configurations and for every individial pc usage preference.

Own experience:

Recently I have tested the following Intel MSM and RST drivers on my my Intel ICH10R system running Win7 x64 with 2x160 GB Postville SSD's as RAID0:

1. Intel MSM 8.6.2.1012 (= Win7 in-box RAID driver dated 07/14/2009)
2. Intel MSM 8.9.0.1023 (= latest official MSM driver dated 06/04/2009)
3. Intel RST 9.5.0.1037 (= first official RST driver dated 10/02/2009)
4. Intel RST 9.5.6.1001 (= latest RST driver dated 12/17/2009)

Here are the results:

The test results show, that the READ Scores (left column) of my SSD RAID system were the best with the MSM driver v8.9.0.1023, whereas the RST driver v9.5.6.1001 induced the highes WRITE Score (right column). Although I didn't realize any noticeable performance difference myself while working with them, I got the feeling, that the RST driver v9.5.6.1001 is the best for my system (no dropouts, very reactive, absolutely stable).

Another question to you:

Did you check, that you got your old test results back after having reinstalled the formerly "better" driver?

[idata]

I've read a few of the threads here about the new RST drivers. They've been available since Jan 14, 2010 but still nobody knows if they support trim mode or not in a single drive configuration with AHCI and Win7.

1. Does the new RST 9.5.6.1001 driver support AHCI? I believe the last RST driver did not support AHCI or is that incorrect?

2 Is anybody with a single Intel SSD running these drivers? If so, does performance drop over time and does the SSD Optimizer refresh the performance. If the performance stays the same over a week of use then one could assume trim commands are passed to the drive? Is this that difficult to test?

3. Why can't Intel be clear about this? Is it that hard for a representative to state if auto trim is supported with Win7, RST 9.5.6.1001 and AHCI??

[GCatt]

TRIM support in AHCI mode is one thing, but I think most people are waiting for Intel to support trim in RAID mode.

[idata]

Yeah, but can anybody even show that the latest RST driver supports trim with even one SSD drive? Page one Post # 29 of this thread, Fernando claims this could be true based on a forum thread. I tried google translate and the impression I get from that thread is nobody has tested this yet.

Somebody should consider writing a program which simply tests if the trim command has gotton to the hard drive or not. TRIM PASSED. TRIM FAILED is the only result the program needs to output. This would be a very helpful tool for many people - actually everyone with an SSD drive. If intel was smart they would add this test into the SSD TOOLBOX!! Anyone up for this challenge? (unfortunately I don't know how to program)

[idata]

With so many community members waiting for information on TRIM support in the new RST drivers, I think total silence from Intel is due to competitive market concerns. Maybe Intel doesn't want competitors to know how far along (or how far behind) they are in RAID drivers supporting TRIM (with a single drive or an array). I'd also like to know if RST will ever pass through SMART information from individual drives in an array. Some customers with RAID wrote about putting drives back in IDE mode occasionally just to read SMART values. I hope Intel realizes there are many of us using RST drivers. Maintenance and early warning for performance/reliability is very important to us!

Raidman, TRIM status indicator (success/fail/never run) is a great idea too.

[idata]

The issues I have with DPC latency were on Western Digital Raptor (mechanical) drives in a Raid 5 array. I've not yet switched to SSD so the "TRIM" support is not my main concern. The 9.5.4 drivers are flat & green in dpclat but the 9.5.6 drivers spike at 6000us ever 4-5 seconds.

I switched to the 9.x.x drivers because th 8.8 version doesn't oficially support Windows 7 (64bit) and the 8.9 version drops drives from the raid randomly and eventually trashed my boot partition requiring a complete rebuild :-(

[idata]

Well I have to admit that i'm very pleased with the new RST drivers, my system definitely feels a bit more responsive and has even reduced my bootup time to roughly 10 secs! (after splash screen etc...)

And I dont care whether they support TRIM or not as I have the SSD toolbox running every week anyway so no biggie.....

walshy ")