Compatibility issue between motherboard and HDD?

[raduque]

The short of it is, I have an HDD that is having problems in my WHS box. Timeouts forcing Windows to drop to PIO mode, causing high CPU usage and I/O slowdown.

The drive is a Samsung HD154UI 1.5tb

The long:
My file and backup sever is a home-built WHSv1 box that consists of a Gigabyte GA-965p-S3 motherboard with six SATA ports. 4 are handled by the Intel southbridge chip, and 2 are handled by a third party SATA controller, and I know not who make it.

The HDD exhibits the same behavior regardless of which ports I plug it into. When it's installed into the WHS, and I try to write to it, transfer speeds eventually drop into the 100s of kilobytes per second. Event viewer fills with endless "ATAPI" errors that state the drive is not responding during the timeout period, and at least as many "DISK" errors that say there's a controller error on the drive. In device manager, you can see the drive has fell back to PIO mode.

Now, here's where it gets weird.

I removed the drive from the storage pool, moved it to another SATA port, and booted up the server. Windows tells me, in the WHS Console and Disk Management that the drive is 500gb. BIOS sees the drive correctly. I attempted to use HDDGuru's HDD Capacity Restore, and it sees the drive's full space but refuses to restore it.

I left it alone for a few days, and rebooted the WHS for some reason today, and when I logged into the Console, it tells me the drive has 1397gb. Disk Management asks me to initialize the drive, then create a partition, and format it. Takes about 20 minutes to do all that. I made sure the port it was connected to was on UDMA5, and tried a zero write operation.

Right back to square one. Timeouts, controller errors, fallback to PIO mode. 6 hours I let it run and the progress bar in the 0-write app from HDDGuru moved two blocks.

So I pulled the drive, threw it in my desktop PC, and zero-filled it in a couple of hours, and wrote 1tb to it with no issues. I then threw it into a USB2 enclosure, quick formatted it, and wrote another 1tb to it with no issues. It pegged the speed in Windows 8 at 40MB/s and went no lower than 30MB/s during the transfer.

I have no doubt if I were to put it back into the server, I would have the same issues again.

So now I call upon the gurus here at HDDGuru: Does anybody have any idea what's going on here?

[fzabkar]

The 500GB capacity problem is a Gigabyte BIOS bug.

See ...

GigaByte BIOS bug results in loss of 1TB capacity:
http://malthus.zapto.org/viewtopic.php?f=59&t=150&p=275

The PIO problem appears to be the result of excessive CRC errors. Examine the UDMA Error Count in the drive's SMART report. If the error rate is excessive, you could try limiting the drive's SATA link rate to 1.5Gbps. I believe Samsung's ESTOOL can do this.

Another thing to try would be to disable NCQ, if supported.

[raduque]

199 UltraDMA CRC Error Count 100 100 00000005-1C45 000

Well, I guess that's that then. It doesn't seem to have any issues working fine at UDMA5 speeds in my gaming desktop, so I guess I'll keep it there and store my Steam library on it.

[fzabkar]

That's a lot of errors.

0x51C45 = 334917

http://www.google.com/search?q=0x51C45+in+decimal

[raduque]

So what is it that causes this many CRC errors? Is it the platters, the drive PCB, something else?

[fzabkar]

I believe it's the SATA interface that is the culprit. If so, then reducing the link rate from 3Gbps to 1.5Gbps may solve your problems. There are also registry edits that could possibly mitigate the problem.

See http://www.tomshardware.com/forum/29283 ... -back-mode

[raduque]

I will give that a try and report back, as I'm currently writing data to it again on the desktop, and the CRC error count isn't changing, so it MUST be an issue with that motherboard.

Odd though, as the Samsung HD204UI is working 100%.

Edit: Oh and I've done the uninstall ide/reboot thing a few times.

[fzabkar]

FWIW, there have been compatibilty problems between some motherboards (the AMD SB850 chipset and the Intel P67/H67 chipset) and Samsung-brand hard drives.

http://knowledge.seagate.com/articles/e ... Q/223631en

Yours appears to be a different issue, though.

Edit: Some Samsung models can be jumpered for 1.5Gbps. I don't know whether yours is one of these.

http://web.archive.org/web/201309151008 ... Q/221435en

[raduque]

I have changed the SATA mode, but I'm about 30% through with a read surface scan in ESTool. If it finishes tonight, I'll update tonight, but if not, I won't update till Sunday evening.

[raduque]

Update: After the surface scan completed 100% and the drive passed all the tests, I re-installed it in the server with a brand-new cable. It appears to be working with no issues. I wrote over a terabyte to the drive inside the server with no speed issues, and no event manager errors.

I then formatted it and added it to the WHS drive pool, and it's been working correctly since.