Hello
I search several times in the forum and I haven't seen anything like this, so i decide to post.

I have a 2.5" HD samsung HM321HI (320Gb) SATA, it was in a Iomega USB case, one day didn't allow me to write anything, all kind of errors when I try to write, but I could read all the data just fine.

I low level format it with VIVA and, now it's working again but the write speed is terrible around 1.5 M/s, while reading I get around 90 M/s.

I repartition and format, then write some big test files (1GB each), reboot and read them, same story.

Write speed 1,5Mb/sec, Reading speed around 85 Mb/sec.

any ideas what could be wrong?, I run MAKEBAD utility and didn't detect any slow sector (obviously because it only reads!)

thank you
Indkt.

What is the source for your 1GB test files? Is it possible that your source drive (?) is the write bottleneck? What is the write speed if you copy a file from one location on your problem drive to another location on the same drive?

_________________
A backup a day keeps DR away.

Hello!
The source drive is a secondary WD10EARS (not system) brand new.

I have tested with a good drive via eSATA and I have read and write around 90M/s (WD scorpio black 320Gb)
when I use this drive connected in the same way, write speed is between 500k/s and 1.5M/s, but read is around 80M/s.
also tested the same drive in another machine via USB-SATA adapter, and I have 33M/s read (as expected for USB 2.0 , and again 1.5M/s write)

i don't know if this is defective surface, and the drive is doing several passes to make sure it write it properly, or if this is something else, this is not the first time i see something like this in samsung drives but never knew why this happens.

smart data seems to be ok (for what I understand) no current pending sectors, and only 8 reallocated sector count.

Another test I did yesterday, was to divide the disk in 2 partitions of the same size, then I ran the test in the secondary partition to see what happens if I write to the upper half of the disk, and in this area the write speed is far better, around 40M/s

what do you think?
thank you!
Indkt.

I have a few comments, from my (non-DR) experiences:

a) Any non-zero reallocated sector count value, was found in the Google data centre study to have a strong correlation with a disk that will suffer premature failure. Of course a manufacturer would not accept a disk for warranty replacement if a low count is the only observable problem. In your case, you did have other symptoms at first (e.g. all writing failed) and so it seems likely that the Samsung disk is faulty.

What is "VIVA" that you say you used, to try to fix the original "all writes fail" problem?

b) I suggest that you do a full read benchmark graph on the "slow writing" Samsung disk e.g. using HD Tach, and attach the graph to your reply. I know you apparently see fast reading speeds, but that is not detailed enough - the graph of read speed is more useful. You could use MHDD (with the correct BIOS disk controller setting) to perform a similar test, if preferred. Remember that disks must read, before they do a write...

c) You may not have the tools to actually confirm that there is a problem with this disk, if the only symptom you now have, is slow writing - that may not be considered an error by diags. However you could run the Samsung testing utility from their website, to see if you are lucky that it confirms a failure. I do not know if that will work on disks in a USB enclosure (I don't have any Samsung disks in my lab at the moment, to test).

d) Since the write speed varies depending on where you are writing (based on the results of your recent test), then again, this points strongly towards a disk fault.

e) I don't know if there is a serial port on the 2.5" Samsung disks, like on their 3.5" disks. If it was my disk, then I would investigate this, and then check to see if the disk was also logging any messages on that port.

Finally: Now that we have established this is very likely to be a genuine disk problem, what other result are you expecting from your questions exactly?

FYI, there have been other "slow disk" threads here, where normal tools were unable to prove any failure, even though the disk behaviour was clearly abnormal.