10-28-2011 05:03 PM - edited 10-28-2011 05:12 PM
Ironclaw, I've just finished to run successfully (no checksum errors) 25 passes of your 4GB test on a SD28 firmware (drive was originally SD23, then updated to SD24, then SD28, then zero-fill, then re-flash again to SD28).
The test was performed in cygwin's mintty terminal, under WinXP SP3, with 2 NTFS partitions on 2 hard drives connected directly on SATA II. The test drive was the Seagate Momentus XT 500 GB; the input test file was on a WD Scorpio Black 500 GB (WD5000BEKT) (by the way... the Scorpio is an excellent drive)
I can confirm that SD28 in cygwin under Win XPSP3 has NO file corruption issues. This is valid ONLY IF the update to SD28 has been done correctly: see post 18 here:
11-03-2011 03:19 AM
OK, FILE CORRUPTION with the SD25 firmware in WINDOWS for the Momentus XT 500 GB drive.
The file corruption appeared only after copying about 25 x 4GB files under WINDOWS XP SP3 32bit.
I did a first test of 15 x 4GB and everything was OK. Then in the second batch of 15, the 10th file is damaged. There is a block of zeros in this file of about 512 kB where data has not been written !
This is unacceptable and I'm sure that a careful investigation will show corruption on any SD25 firmware. One needs to copy from a SATA, high speed connected drive a random-filled 4 GB file (as Ironclaw generated) to the MXT at least 30 times (maybe more). That is more than 100 GB of data and it takes some time.
My batch file for testing looks like this:
sha1sum.exe random_4000mb > d:\mxt_out\check_in.txt
copy random_4000mb d:\mxt_out\random_4000mb.01
sha1sum.exe d:\mxt_out\random_4000mb.01 >> d:\mxt_out\check_out.txt
copy random_4000mb d:\mxt_out\random_4000mb.02
sha1sum.exe d:\mxt_out\random_4000mb.02 >> d:\mxt_out\check_out.txt
copy random_4000mb d:\mxt_out\random_4000mb.03
sha1sum.exe d:\mxt_out\random_4000mb.03 >> d:\mxt_out\check_out.txt
copy random_4000mb d:\mxt_out\random_4000mb.04
sha1sum.exe d:\mxt_out\random_4000mb.04 >> d:\mxt_out\check_out.txt
continuing up to 30 or more. The input (good) file is called random_4000mb and alongside the sha1sum.exe file reside on the non-MXT drive. D is the partition on the MXT to which I copied the data.
People, if you use SD25, take your time and test that drive thoroughly... let the tests run for 2-5 hours.
I got blue screens with SD25 twice just yesterday.
One more thing: SD25 is in NO WAY faster than SD28 for me. It is exactly the same speed. They both boot in 1m10s for me and behave very similarly in windows/movies/games.
It seems to me that the only reliable firmware and also fast enough is SD28 at this point.
12-17-2011 06:15 AM
I believe that the lags/stuttering issues (maybe also connected to the End-to-End errors in SMART, not entirely sure about this) are directly caused by the SD25 firmware. It seems that the problems occur ONLY on drives which, at a given point had the SD25 firmware (even if, afterward one updated further to SD28, for example).
SD25 messes up drives really badly. We've already shown that it corrupts files under all operating systems. But I strongly believe that it also messes with the SSD and/or with some very important cache-controlling firmware components.
I can tell you that I have never seen the lags/stutterings/freezes on the following 2 drives:
1. originally SD23, updated to SD24, then to SD28 directly. Perfect drive. No corruption
The second drive:
2. originally SD22, never updated to anything else. It works very well, no problems. No file corruption.
It seems that the file corruption was introduced in SD24 (I'm not entirely sure about this). Can anyone confirm ? However, SD25 clearly shows serious problems with corruption.
I've seen the lags/freezes only on drives which, at a given moment in time, had SD25 on it (especially for a longer amount of time). Those drives sometimes showed the End-to-End errors in SMART.