Hi, as most of you propably already know: Microsoft released a hotfix for all 4K (aka adv. format) drives http://support.microsoft.com/?kbid=982018 That of course includes my WD15EARS drive. Unfortunately the drive reports a physical sector size of 512 Bytes (checked via ‘fsutil fsinfo ntfsinfo’) instead of 4096 and thus the MS patch doesn’t have any effect for my drive. Anyone experiencing the same behavior? Any workarounds known? I really would like to fix the performance problems I have with smaller files. OS: Win 7 x64 Drive: WDC WD15EARS-00MVWB0 Firmware: 51.0AB51
TechGuy wrote:
Unfortunately the drive reports a physical sector size of 512 Bytes (checked via ‘fsutil fsinfo ntfsinfo’) instead of 4096 and thus the MS patch doesn’t have any effect for my drive.
Um… if the patch is for AF drives that report themselves as 4096, and your AF drive is reporting itself as 512 already, then isn’t Microsoft telling you you don’t need the patch?
“This hotfix is only applicable to Advanced Format disks which report themselves as having a 4 KB physical sector size”
None of the other “Issues” listed as being addressed seem to be related to any issue you’re encountering. The only potential one is Issue #2… but the issue only occurs when at the end of a file with a buffered write, in where the buffered write is padded out to the “logical” size instead of the “physical” size. On a drive that reports 512 logical and 4096 physical, there’s a discrepancy. If your drive is reporting both logical and physical as 512, then there’s no discrepancy for the patch to be dealing with.
From what I can see, WD has already dealt with this in their “Advanced Formatting”, the way the controller handles the data back and forth… it seems to only be an issue with other manufacturer’s drives that report a logical 512 and a physical 4096.
From what I can tell, the drive’s on-board controller is doing any necessary padding (presumably because WD designed it that way, because Windows wasn’t padding properly) and not relying on Windows to do it.
RoofingGuy wrote:
Um… if the patch is for AF drives that report themselves as 4096, and your AF drive is reporting itself as 512 already, then isn’t Microsoft telling you you don’t need the patch?
“This hotfix is only applicable to Advanced Format disks which report themselves as having a 4 KB physical sector size”
- *> You forgot to quote the rest of the sentence “and which emulate a logical addressing interface of 512 bytes”
Imho that’s exactly the case for the drive since it internally operates with 4k sectors but has an addressing interface of 512 bytes. All 4k drive rules apply to the mentioned drive including the partitions boundary rules, i.e. if the drive doesn’t report the correct physical sector size value it uses then the partitioning/disk tools (and the OS itself) won’t be able to detect which boundaries should be used. That’s btw the reason why all Linux users do have problems using this drive.
Regarding the padding: well it would actually be very interesting to know if the onboard controller handles the paddings properly as described here http://support.microsoft.com/kb/981208 or if it fills up the 4k sectors.
> Anyways, thanks for your input! Hopefully WD will make a statement regarding this issue …
Well, I didn’t “forget” to quote it… I just chose not to, since it was unnecessary. All “Advanced Format” drives emulate 512-byte logical sectors… that’s what makes them “Advanced Format” as opposed to being 512-native or 4096-native.
The issue Microsoft is discussing, is not whether the logical sector is reported as 512 bytes… because that’s a known fact. The issue Microsoft is discussing is a performance loss if the final byte of a file won’t be properly padded if the drive is reporting a physical size of 4096. But since your drive is not reporting physical 4096, the condition Microsoft is describing does not seem to apply.
It would seem to me to make more sense if Microsoft stopped worrying about physical and logical and any potential differences, and just merely padded the final chunk out to a full Allocation Unit. Since a cluster must be a whole number of physical sectors anyways (well, not just a whole number, but a power of two), it’d seem to remove any problems with whether a physical sector gets “filled” or not at the end of the file.
As described, padding is one thing, proper alignment another. The patch MS provides is not just about the padding, it’s also about the alignment and a way for the OS, drivers and tools to get the physical sector size of a drive via the IOCTL_STORAGE_QUERY_PROPERTY command. And again - as described - this information is needed by all low level disk tools including the OS itself otherwise they will have more or less problems just like Linux/MacOS creating partitions on those drives.
So the conditions do at least partially apply, no matter if the drive’s logicboard performs well or poorly doing the padding itself.
Btw, WD lists the hotfix to improve reliability and performance ‘including Western Digital’s Advanced Format drives’:
And that’s simply not true if the OS is not even able to detect the drive as a advanced format drive.
