Mega SSD Review Part 3 – Super Talent’s Firmware Mystery

Mega SSD Review Part 3 – Super Talent’s Firmware Mystery

It has come to my attention that Super Talent had posted a new firmware version on its website (090508S), and I had missed out that update. Upon updating the drive’s firmware though, I found that the random write speeds have improved dramatically. Without further ado, here’s my findings.

Info tab in HD Tune Pro on the 090508S firmware
Info tab in HD Tune Pro on the 090508S firmware

Looks like Super Talent got it right this time, and the firmware version string is updated. Interestingly, one of the SSD drives that was missing about 500MB had its full capacity restored, and the 48-bit addressing feature is back. Security mode is still not in, however. I’m not sure if Security Mode in HD Tune refers to the password-protected hard drive feature, or the secure erase feature, but I am hoping that secure erase will be back, if it improves the speeds of the drive. Let’s take a look at the sequential read and write performance from this firmware version, compared to the previous ones.

Exceptional sequential reads, as usual
Exceptional sequential reads, as usual
An improvement in average sequential writes
An improvement in average sequential writes

Looks like a big improvement in that area for CPU usage, average reads, maximum reads and average writes, with decreases in performance in a few other areas. Now for the killer test – random read and writes.

Stellar read performance, as usual
Stellar read performance, as usual

Random reads are no sweat for SSDs. But what about random writes?

An improvement in random writes?!
An improvement in random writes?!

It seems that random writes have an improvement of roughly 10 IOPS for 512 bytes, 4KB and 64KB transfers, as compared to the old firmware. A complete breakdown of the write speed comparisons are shown below:

HD Tune - FW090508S Write Operations / second (IOPS)
HD Tune – FW090508S Write Operations / second (IOPS)
HD Tune - FW090508S Write Average Access Time (ms)
HD Tune – FW090508S Write Average Access Time (ms)
HD Tune - FW090508S Write Average Speed (MB/s)
HD Tune – FW090508S Write Average Speed (MB/s)

A complete breakdown of the read speed comparisons are shown below:

HD Tune - FW090508S Read Operations / second (IOPS)
HD Tune – FW090508S Read Operations / second (IOPS)
HD Tune - FW090508S Read Average Access Time (ms)
HD Tune – FW090508S Read Average Access Time (ms)
HD Tune - FW090508S Read Average Speed (MB/s)
HD Tune – FW090508S Read Average Speed (MB/s)

And finally, the IOmeter tests.

IOmeter - FW090508S % CPU Utilization
IOmeter – FW090508S % CPU Utilization
IOmeter - FW090508S Write IOps
IOmeter – FW090508S Write IOps
IOmeter - FW090508S Write MBps
IOmeter – FW090508S Write MBps

Surprising results, as it seems that the performance has been lowered for the ext2 filesystem (in Windows, at least, using the IFS driver). Although CPU utilization is lowered for all filesystems, the write MBps and IOps have decreased. However, performance on ext2 still outpaces the NTFS and raw disk access, while NTFS and raw disk access speeds have increased.

In conclusion, although it is certainly better in benchmarks, it isn’t really better as an “overall experience”, mainly because during write operations on the SSD, Windows XP will often lock up and wait for the outstanding IOs to be finished before I could do anything. What this means is that my Windows XP system will often pause while doing file transfers to and/or from the SSDs, especially when I multitask (e.g. playing music on iTunes while installing a game to one of the SSD drive) – note that my Windows XP installation is on a hard drive and not on the SSD itself. However, it seems limited to only ext2-formatted drives using the IFS driver for Windows – NTFS and raw disk access are snappy.

EDIT: It seems that the firmware is not limited to only Super Talent drives – in fact, if one does a proper search online, the firmware update is supposedly a generic update for all JMicron JMF 602 based SSDs (at risk of voiding your warranty and/or bricking your SSD). If that is the case, I hope that the firmware update works for JMF 601 devices, such as my old Transcend SSD that I bought last year, assuming that I can get the updater to detect the Transcend drive.

EDIT: Since there are no resources on the web, I’ve decided to compile a list of drives that can and cannot be flashed using either the Windows-based update (for USB only) or the DOS update (for SATA only):

Flashable:

  • G.Skill FM-25S2S-64GB (via USB)
  • KingSpec J Series (via SATA)
  • OCZ Core V2 (via USB)
  • OCZ Solid (via USB)
  • Silicon Power (via SATA)
  • Super Talent MasterDrive OX (via SATA)
  • Team Combo (via USB)

Not Flashable:

  • OCZ Apex (2x JMF 602 in RAID configuration with JMF20330 controller)
  • Transcend TS32GSSD25-M (JMF 601 with JMF20330 controller)
Comments are closed.