Does Your SSD's File System Affect Performance?

AS SSD: Random And Sequential Throughput

Random 4 KB read and write operations with an unsaturated queue (a queue depth of one) represent a fairly typical load scenario.

Writes are faster, as the drive can quickly find a free sector to write to, while the read operations are specific to a given sector. Writes to the Samsung 830 are significantly slower using FAT32, though the Zalman F1 drive seems to suggest this isn't an issue inherent to the file system itself.

A very large queue depth of 64 pending commands allows the drives to optimize read operations, leading to much improved throughput.

On exFAT and NTFS, this has a very positive impact, while the drives working with FAT32 suffer enormous performance hits. However, a queue depth of 64 is really only theoretical, since desktop systems hardly ever experience queues depths in excess of four.

Sequential throughput helps illustrate FAT32's lagging performance in a more realistic situation, and its throughput is severely limited, especially on the Zalman SSD. Since storage drives tend to be used for holding larger files (like the aforementioned videos), this metric is ostensibly the most relevant to us.

Create a new thread in the UK Article comments forum about this subject
This thread is closed for comments
Comment from the forums
    Your comment
  • jamesedgeuk2000
    Could you not have benchmarked HPS too by doing the tests with a Mac setup to duel boot OSX and Windows 7? (thus letting the 7 install see the HPS file system)
  • djamorpheus
    What about linux filesystems?
  • audiovoodoo
    No mention of cluster size impact on performance?
  • audiovoodoo
    What are you talking about?
  • Anonymous
    Question is does the FAT32 partition was aligned? Data on FAT32 starts straight after two FAT tables (their size depends on amount of clusters). So even if volume for FAT32 partition is aligned after creating such partition the FAT32 clusters may not be aligned to SSD physical sectors.