Silicon Image chipset and corrupted image

I'm getting corrupted images using the Silicon Image chipsets 3512 and 3124. Both are PCI Sata controllers, 3512 having one internal and one external port, and the 3124 having two internal and two external ports(eSata). The drive being imaged is connected to the internal port and the destination drive is connected to the eSata port on the same card. If I image to a USB drive from the Sata controller everything is cool. I'm glad I caught the corruption before I needed to do a restore.

 

Are you doing this within Windows or with the TI rescue CD?

As a first test, I would calculate the MD5 checksum for a large image file already created on the SATA HD connected to the board and then copy it to the eSATA drive and recalculate the MD5 checksum. It must agree. I would then copy the file from the eSATA drive back to the internal drive and recheck it again to ensure it agrees. If it doesn't then you have a problem with the device.

A free MD5 checksum calculator is available here:
http://www.irnis.net/soft/xcsc/

My guess is that the above test will not show a problem but at least you will know the device can handle large files. It could be TI is really hammering the card since it is using both channels at a high-rate and the checksum test is really using only one at a time.

 

There are so many instances where this program just doesn't work as advertised.

Now with the SI 3124 card I've set up a raid1, with an Acronis Secure Zone partition.

Created an image to the secure zone and upon validation I discover it too is corrupted. This after I was advised by the program that the backup operation completed successfully.

And the log file was absolutely worthless... error message stated that the validation operation completed with errors, see log for more details. Then I read the log file only to see "operation completed with errors"... But the error code 0x640005 should help me troubleshoot.. not.

 

Hello ssorg,

Thank you for choosing Acronis Disk Backup Software.

Please notice that most likely the reason are faulty data transfer routes. Please test the transferring as seekforever suggested. Please also notice that corruption may be a result of a faulty memory module. To check that, please download memtest86+ at http://www.memtest.org/ for free, and run it for several hours.

Thank you.
--
Marat Setdikov

 

I'll give memtest a shot, but I doubt memory is the problem here, this is the most stable box I've ever put together.

Tried operating with both controllers installed, and backing up to 3512's eSata, but I had the same result of a corrupted image.

I'm about to give up on the idea of an eSata backup drive on this box, not going to waste anymore time on it. Raid1 and a good image no matter how slow is good enough I think.

Here are a few more interesting quirks I've run into with this system:

-ATI boot disk doesn't see controller SIL3124 (should according to Acronis)
-BartPE won't see USB hard drive, but will see same hard drive in same enclosure with a 1394 firewire connection

The eSata enclosure was an early suspect, but I'm able to backup to it successfully on another Vista machine's eSata port where the other channels are running an nVidia stripe. Interesting to note also that the Acronis boot disk will absolutely destroy an nVidia stripe, hence the venture into BartPE. Thanks Bart !

That's all for now - I'll eat crow for sure if memtest comes back with errors.