AHCI mode Worth Installing to Replace IDE mode?

[wulf109]

That's the question.

[nic2k4]

You'd gain NCQ for increased performance, that's basically what was giving SCSI drives an advantage over IDE. If you have SSD's you need it enabled to use TRIM.

[hello_hello]

I don't know if it applies to newer versions of Windows, but to change a controller from IDE to AHCI mode requires different drivers, and if Windows is expecting an IDE boot drive it won't boot. When it came to XP I think the simple fix was to re-install Windows, but there my be ways around it when using newer versions I'm not aware of.

If the SATA controller has a RAID mode, it's usually recommended to use it in preference to AHCI (at least Intel does). You still get all the AHCI goodness, but you can also ruin RAID if you want to.

For me, the min advantage of AHCI mode is it usually (although it's controller dependent) supports hot swapping. Going back to having to reboot to unplug/connect SATA devices would be a real drag. I'm not sure mere mortals would notice any real performance difference.

[lordsmurf]

ways around it

Read this:
Easy way to convert IDE mode to AHCI in Windows XP, for SATA hard drives ?

All you need is a PCI card (temporarily). eBay, $25.

[hello_hello]

Interesting..... So I assume from that, if you already have a second hard drive connected, and it's aleady running in AHCI mode, switching the OS drive shouldn't be a problem?

Maybe the OP should also be aware, if they're running XP, it needs 3rd party AHCI/RAID drivers to be loaded at the F6(?) prompt when installing as it doesn't have any of it's own.

[glockjs]

to go out of your way to do it...no. if you have the option there? yes absolutely.

[themaster1]

If the SATA controller has a RAID mode, it's usually recommended to use it in preference to AHCI (at least Intel does). You still get all the AHCI goodness, but you can also ruin RAID if you want to.

Do you know if i can use raid drivers without actually doing a real raid (no data copy to other drives) so i can benefit with the ncq feature

I have 3 pc's with old mobos (2003-2006) all have raid drivers , on the one i use daily i use the ide drivers
On these mobos only ide, raid mode available

[wulf109]

I slipstreamed sata\ahci drivers into my XP Pro disk and intalled on a formatted HD. Installed on an Intel I7-920 and the ahci drivers seemed to speed up BR encodes by nearly 20%. A disk that converted in 80 min with IDE drivers.converted in 60 min with AHCI drivers. A repeated the test 3 times to make the results were valid. The same test on an AMD computer produced no change in encode times,strange.

[hello_hello]

Do you know if i can use raid drivers without actually doing a real raid (no data copy to other drives) so i can benefit with the ncq feature

I have 3 pc's with old mobos (2003-2006) all have raid drivers , on the one i use daily i use the ide drivers
On these mobos only ide, raid mode available

Does the chipset support NCQ when running in RAID mode?
My memory on all this has faded somewhat, but I think AHCI mode is basically an Intel thing which is now widely supported, but AHCI mode aside, there's only IDE mode or RAID mode. RAID mode should offer the AHCI features but they have to be supported by the chipset. For instance I'm not sure NCQ was supported at all by earlier SATA devices/chipsets (SATA 1.5Gb/s). Same with hotswapping. To the best of my knowledge, there's no need to actually run drives in a RAID configuration when using RAID mode with any chipset.

Edit I just found this:
http://forums.overclockers.co.uk/showthread.php?t=17504457
Seems NCQ is an optional feature for SATA 1.5Gb and SATA 3Gb, but I'm still not sure it was very widely supported by SATA 1.5GB chipsets. I assume it's pretty standard now.

[hello_hello]

I slipstreamed sata\ahci drivers into my XP Pro disk and intalled on a formatted HD. Installed on an Intel I7-920 and the ahci drivers seemed to speed up BR encodes by nearly 20%. A disk that converted in 80 min with IDE drivers.converted in 60 min with AHCI drivers. A repeated the test 3 times to make the results were valid. The same test on an AMD computer produced no change in encode times,strange.

If you were comparing encoding speed after a fresh installation of Windows, could there be other factors involved aside from IDE vs AHCI? Because even a single PATA hard drivbe should just be cruising along while keeping up with an hour long encoding job.

Plus, I've never seen any NCQ benchmarks which say it offers anything but a minor performance increase. For instance Toms Hardware compared two Seagate 7200.7 hard drives, one with NCQ and one without. The drive without NCQ had an average read transfer speed of 28.3MB/s, while the drive with NCQ averaged 28.4MB/s.
Now admittedly that review is very old, and drives have gotten faster since than, but does NCQ make any more of a difference? I'm not sure....

[nic2k4]

Plus, I've never seen any NCQ benchmarks which say it offers anything but a minor performance increase.

NCQ isn't supposed to make the drive faster, it allows for commands to be executed in an optimized order. The benefit is that a single application won't hog the drive.

[hello_hello]

Well if optimizing the order in which a drive executes commands isn't for the purpose of making it faster, then what's the point of it?

[fritzi93]

to go out of your way to do it...no. if you have the option there? yes absolutely.

Exactly.

When I built the computer listed in my computer details, it had an SSD right from the start using AHCI. It wasn't long before I put in a bigger one (90 GB Vertex), and relegated the 64 GB SSD to my wife's office computer, IDE mode. I benchmarked it before and after and the difference was small. I have it written down somewhere, but I'm talking nothing to get excited about.

As to TRIM, the firmware will do garbage collection okay anyway without TRIM if it's fairly recent manufacture.

As to SATA mechanical hard drives, as opposed to SSDs, I doubt AHCI will make a significant difference.

BTW, I just had a little problem with bad clusters on the Vertex (OCZ) drive. I was alerted to it when trying to image the drive. Acronis popped an error about being unable to read a bad cluster. Retrying just returned the error. Hitting ignore allowed the imaging to complete. I tried running system file checker and that errored too. Running garbage collection did nothing.

So I did some searching, not being sure if I could run chkdsk safely on an SSD. Lots of posts saying don't do it, some saying check file system only and don't try to repair bad clusters.

Finally, I spotted a post on anandtech, and the guy basically said geez, you guys seem to think SSDs are arcane devices that'll blow up if you run chkdsk. If you run "chkdsk c: /f /b" windows will nudge the drive into repairing the bad clusters or marking them as unusable (which it should have done anyway, of course). Being somewhat at a loss otherwise, I thought I'd chance it. Yeah, I could have done secure erase and re-imaged (with a month-old image). I thought I'd likely still be able to do that if chkdsk failed to fix it.

Sure enough, chkdsk found a couple bad clusters, both in the same sector, and IIRC it was Windows fonts and Win SXS. It reported check run successfully on reboot, volume clean. Then system file checker completed successfully as well, reporting that it had made repairs.

That was a week ago, and everything is working properly ever since. I was also able to image the drive with no errors. All other tests come up good. From what I can gather, a power outage is the most likely cause of the problem. SSDs are sensitive to improper shutdown and power outage. Every now and then a squirrel fries itself on the transformer up the road.

[RabidDog]

AFAIK, NCQ will come into its own when there are multiple requests to the same disk, generally multiple programs running or a multi user system.

and if Windows is expecting an IDE boot drive it won't boot.

This caught me out, badly. Wanted to try AHCI and ended up re-installing windows ..