TRIM Support in RAID 0 - Intel RST 11.5 WHQL released

They work with any controller right? No matter if it's SF, Samsung or Marvell?

Now... I seriously think of adding 2nd 830 to my setup next month. Would be pure fun
 
There is nothing in the release notes that mentions TRIM support in RAID. It would be nice if it does, but if this were the first driver to fully support it I would expect Intel to mention SOMETHING on the driver page. Also, this is not a general release, it is listed for just a selection of the latest Intel boards, if you check (as of now) for the general chipset driver it is still 10.8.x.
 
Last edited:
not proven to work yet, only speculation.
 
Last edited:
I've been running the 11.5.0.1414 OROM on my Gigabyte Z68XP-UD3 along with the Intel RST 11.5.0.1184 driver on Windows 7 x64 for a few weeks now. I'm running two Crucial M4 256GB SSDs in a RAID0 array and am quite enjoying it.

While I'd like to believe TRIM is working and TRIM commands are being passed to the SSDs in the array, there is no way to tell for sure short of running benchmarks and watching for performance degradation.

Regardless simply updating your system with these drivers won't magically enable TRIM for SSDs in a RAID array of any type. At the very least you will need to be running an Intel 11.5.x Option ROM...

There are also rampant rumors floating about that Windows 7 simply cannot support TRIM in RAID0, but Windows 8 can with the proper combination of OROM and drivers. This is complete news to me and I tend to think it's hogwash.
 
Last edited:
Blue Falcon said:
There are also rampant rumors floating about that Windows 7 simply cannot support TRIM in RAID0, but Windows 8 can with the proper combination of OROM and drivers. This is complete news to me and I tend to think it's hogwash.
Click to expand...


eh, not hogwash. problem with windows7 is a logical raid volume is treated as a SCSI device, and the SCSI layer in Win7 will not pass TRIM commands. This has been changed in Windows8. If this was a simple issue Intel would've done it a long time ago, the problem is you can build filesystem awareness for raid volumes into the windows driver and OROM but the windows file system layer is still the middleman and thus without Microsoft's involvement they're just paddling pork.

Intel's wording in the release notes were also seemingly intentionally vague - "TRIM support for RAID0 in Windows" -- with no version specified. I guarantee they mean Windows 8.

Everyone doing "testing" on various forums jumping for joy thinking TRIM is working = placebo due to garbage collection.
 
Last edited:
Intel's wording in the release notes were also seemingly intentionally vague - "TRIM support for RAID0 in Windows" -- with no version specified. I guarantee they mean Windows 8.

Everyone doing "testing" on various forums jumping for joy thinking TRIM is working = placebo due to garbage collection.
Click to expand...

It's sad to believe that some people still believe that a miracle will happen. :D
 
Old Hippie said:
It's sad to believe that some people still believe that a miracle will happen. :D
Click to expand...
It would make me sad, if nobody would believe in miracles.
According to the findings of a German user named btester TRIM is working within his RAID0 array while running Win7 x64 in combination with the Intel RAID ROM v11.5.0.1414 and the RST drivers v11.5.0.1207. He verified it by opening parts of his RAID drive with a HexEditor. Look into the screenshots of >this< post.
Here is a translation of his text:
btester said:
I have tested this new Intel RAID ROM for SATA - v11.5.0.1414 + RST 11.5.0.1207 with my Z77X-UD5H under WIN7-64bit with 2 SSD's as RAID0. After having deleted the data the related sectors are empty again (showing just zeros).
As long as I used older RST drivers and ROM versions I got these sectors with the deleted data completely full with YYYYYYY, now - with the actual v11.5 Drivers/ROM combination - I only see 0000000, that means the cells have been cleaned by TRIM.
The same thing happens with a single SSD running in AHCI mode after having done a quick formatting.
According to my knoledge TRIM is working now within an RAID0 array of SSD's.
Click to expand...
 
Fernando 1 said:
It would make me sad, if nobody would believe in miracles.
According to the findings of a German user named btester TRIM is working within his RAID0 array while running Win7 x64 in combination with the Intel RAID ROM v11.5.0.1414 and the RST drivers v11.5.0.1207. He verified it by opening parts of his RAID drive with a HexEditor. Look into the screenshots of >this< post.
Here is a translation of his text:
Click to expand...

The musings of an anonymous German user running a hacked beta OROM with a narrow-release driver are generally less than I would accept to put any of my data at risk. When Intel releases a chipset-level release driver with less ambiguous text as to what it does and does not do, then we can look to testing the stability of the driver.
 
mwroobel said:
The musings of an anonymous German user running a hacked beta OROM with a narrow-release driver are generally less than I would accept to put any of my data at risk.
Click to expand...
Nobody is forced to create a RAID array and to use any special RAID ROM or driver version, but users can be happy, if someone is courageous enough to test it.
Intel has already announced in August 2011, that the TRIM support of RAID0 arrays will be integrated into the next build of the RST v11.5 branch, but we haven't yet seen any official announcement by Intel, that it has been done with the official RST drivers v11.5.0.1207.
 
but users can be happy, if someone is courageous enough to test it.
Click to expand...
You're the one that believes in miracles so go for it! :D

I have never had a problem with my RAID0 SSD arrays without TRIM and really don't care about it.

TRIM in RAID will require W8 and I'm not going to deal with that for awhile either.
 
Well if that is Old Hippie's reason then that is no reason at all since it's the filter driver iastorF that converts W7's trim command to SCSI unmap as has been shown previously in the Asus BIOS thread back in May where TRIM worked for single SSD's seen as SCSI devices but not for arrays, using W7 as well as the older W8 and 11.5 versions.

I can not comment for the latest drivers as I have not tried them but what is really sad IMHO is that so many people believe in the misinformation on the net and even sadder still when they continue to believe it and scoff at people who go out of their way to try and prove things. At least try to keep open minded guys ;)
 
Well TRIM support for single SSD's on a controller in Intel RAID mode has been working for some time. But that doesn't mean it extrapolates to striped volumes in terms of how easy it should be for them to implement -- if it was simple then it would've been working a long time ago.

Bottom line nobody's actually been able to prove it works on W7 yet, only claims with vague methodology and screenshots that aren't really proving anything. And the reason people are skeptical is being Intel has said countless times over the past few years that RAID0 trim support was coming "in the next version" and have continued to fail to deliver, or it was gotchas after a lot of buildup like single SSD support only.

But I do look forward to more testing this.
 
Last edited:
Well if that is Old Hippie's reason then that is no reason
Click to expand...
Oh My.

At least try to keep open minded guys
Click to expand...
If you believe it works then go for it.

This whole TRIM thing has been a 'pie in the sky' for everyone except those that know better.

Like I said before, it's not necessary if you know what you're doing with SSDs.

Do what you want but this has been an old wives' tale for a long time and when it's ready Intel will let it be know.
 
odditory, the point is many people believe W7 TRIM will not work with devices seen as SCSI devices as UNMAP is required however this is taken care of by the filter driver. 11.5 drivers present drives to the OS as SCSI devices.

I'm currently not in a position to do any testing at this time but the previous tests I did with older 11.5 drivers showed zero's returned for LBA's that were trimmed and had previously contained incompressible data. Even a 40GiB+ incompressible file was created and checked with no perceivable host writes / media writes to the SSD when trimmed but returned all relevant LBA's as zero's. Just to re-iterate, TRIM did not work for arrays at that time with both W7 and W8, only non-member disks. I have not tested the newer drivers.

Also a driver stack trace was performed to see the TRIM command converted to UNMAP and in the case of using the undocumented registry entry DisableUnmapSupport of iAstorF, which has to be manually created in the windows registry, it was seen the TRIM command was passed through instead of being converted to UNMAP and resulted in returning failure and of course no TRIM where TRIM had been successful before. And after all this people will still say it does not prove anything.

In the case of zero's being returned after TRIM it should in some cases be fairly straight forward to see if TRIM was successful. Note however different firmware may behave very differently and of course after TRIM, GC still needs to do it's job.

IMHO how much TRIM will bring to arrays if it's implemented may vary from improvement to no noticeable change or even in some cases failure (BSOD/lockups). Hopefully time will tell.
 
So you're using the proper OROM and still not getting TRIM through RAID? This makes no sense to me -- the chipset shouldn't figure in at all.
 
Fernando 1 said:
Yes, and I am not the only Z68 user, who has tested it with the result: No Trim within the RAID0 (look >here<).!
Click to expand...

Like I said before, it's not necessary if you know what you're doing with SSDs.

Do what you want but this has been an old wives' tale for a long time and when it's ready Intel will let it be know.
Click to expand...

Thanks Fernando 1.
 
will someone please post documentary reference/working proof as to how ahci trim hints to unmap hints are 'taken care' of by the intel filter driver? with a windows OS? and not the results of a hex editor with a bunch of zeros (programmed nand). i want to see some ones/0xff action/non jonestown trim is working in win7 proof coolaid.
 
Old Hippie said:
...If you believe it works then go for it.

This whole TRIM thing has been a 'pie in the sky' for everyone except those that know better.

Like I said before, it's not necessary if you know what you're doing with SSDs.

Do what you want but this has been an old wives' tale for a long time and when it's ready Intel will let it be know.
Click to expand...
thank the deities and common sense for this most sensible of replies, and i would also ask the bare a55ed raid trim espousing emperors to put some pants on...
 
If you want to run RAID0 even without TRIM, go for it. My Vertex 2 has TRIM enabled through AHCI and it still degrades to shit-like performance a month after being secure erased.
 
bigdogchris said:
If you want to run RAID0 even without TRIM, go for it. My Vertex 2 has TRIM enabled through AHCI and it still degrades to shit-like performance a month after being secure erased.
Click to expand...

Found your problem: You bought an OCZ drive. Get something without shit for firmware (Plextor, Samsung) and profit.
 
Well calling it shit may be a little harsh but I definitely would've been in OCZ's face for a replacement that doesn't go porn and blow it with degrading multiple writes over time. But yeah that's a known issue with that particular drive rather than a failing of TRIM. All the O/S can do is issue the command, if the drive doesn't cooperate well talk to OCZ.
 
Last edited:
Funny thing is I've been in the industry for 25+ years and yet I would still never proclaim myself to be 'one that knows better' or 'an expert'. To me that just reeks of an elitist and inflexible attitude, and that stank sure is hard to shake off....


And now we have Fernando saying that Z68 chipsets can't support it because a few Z68 chipset users performed his 'test' without success while ONE Z77 chipset user got positive results.

aaaand we have Hippie quoting a supposed 'insider' at Asus (who is fairly well known to pull stuff right out of his ass I might add) saying he heard from his own personal 'inside source' that TRIM in RAID won't be here until 2013 at the earliest...

I hereby come to the conclusion that everyone is pretty much full of shit on the topic.
 
Blue Falcon said:
And now we have Fernando saying that Z68 chipsets can't support it because a few Z68 chipset users performed his 'test' without success while ONE Z77 chipset user got positive results.
Click to expand...
If you want to flame against me, you should read my posts more carefully: I wrote about TWO Z77 chipset users, who got positive results. Meanwhile there are some more Z77 users, who confirmed these results (look >here<).

Before I am leaving this thread, I just want to mention, that I didn't start it. I thought, that there would be someone here, who really is interested in this topic.
It is much easier to flame against the testers than to do the experiments yourself.

Bye!
Fernando
 
Fernando 1 said:
Before I am leaving this thread
Click to expand...
FWIW I found your posts interesting and agree that if people are going to respond with asinine comments then perhaps hardforum isn't the place to post such findings.
 
You must log in or register to reply here.
Back
Top