PC running slower RST 11.5 vs. 11.1? (SRT enhanced mode)

[idata]

I recently upgraded the RST driver on my HP Envy 4T from 11.1.0.1006 to 11.5.0.1184 to take advantage of the new TRIM features.

Boot seems faster, but application launches and overall performance were sluggish. When I ran PCMark Vantage, my system was more than 50% slower with 11.5 (24000 --> 11000).

Any idea what is going on? Will this be fixed in a new release?

[idata]

Did you run PCMark only once or multiple times?

[idata]

I ran PCMark HDD Suite only about 5 or 6 times. After the first run (4K), the score jumped to 11K and stayed flat. Previously, I saw results in the 23-25K range.

[idata]

Where did you come up with IRST 11.5.0.1184, which cannot be downloaded from Intel? That's likely a beta driver that was rumored to support TRIM in RAID 0, although whether it actually does is debatable. The latest IRST beta driver posted on that French download page is 11.5.0.1191, so you could try that one.

The latest WHQL certified IRST driver is 11.5.0.1207, which was actually removed from the download list of any Intel boards with chipsets that support RAID, and is found only as an AHCI driver now. Intel now provides IRST 11.2.0.1006 for their own 6 and 7 series chipset boards (except X79, which uses RSTe).

I would not expect a beta driver to perform as well as an officially released driver. It seems the IRST 11.5 drivers have been problematic, since Intel does not now provide any 11.5 driver for their RAID chipsets and boards. The first officially released 11.5 driver, 11.5.0.1207, had a memory leak problem, caused BSODs in some applications, caused the SATA interface to be seen by Windows 7 as SCSI, and performed a bit slower in benchmarks. I experienced all of these things on a Z77 chipset board using IRST 11.5.0.1207. I can't recall an IRST driver being removed like that since I started using Intel Matrix RAID.

Some people claim TRIM does not exist over SCSI, and that Windows 7 is incapable of sending the correct command to TRIM RAID 0 volumes, which is UNMAP, not TRIM. Official, reliable information on this topic is hard to find, if it exists at all. 

[idata]

Thanks, for the tips.

Yes, this is a beta driver from station-drivers.com, not an official driver from Intel. I will check out 1191.

I played around with 1184 some more and discovered that I can get performance back up to the exact original levels (24K) if I check the "Turn of Windows write-cache buffer flushing on the device" option, but I don't want to risk the data loss in the description.

Ideas? Or, should I just wait for an official WHQL driver?