exFAT File-System Performance On Linux 5.9

[phoronix]

Phoronix: exFAT File-System Performance On Linux 5.9

Now that the Samsung-contributed open-source exFAT file-system kernel driver has matured quite nicely since being merged earlier this year as a replacement to the short-lived staging exFAT driver based on an older code-base, here is a look at how exFAT is performing on the Linux 5.9 kernel compared to EXT4 and F2FS as well as the existing exFAT FUSE file-system implementation.

http://www.phoronix.com/vr.php?view=29697

[ajboyle]

Why not have vfat in the comparisons?

[onlyLinuxLuvUBack]

I think people would be interested in a part II of this article where you compare these numbers against a samsung high endurance sdcard in a usb3 reader.

[muncrief]

Wow, that's an incredible improvement over the FUSE implementation. I had no idea it was so large. Thank you Michael.

[polarathene]

I think people would be interested in a part II of this article where you compare these numbers against a samsung high endurance sdcard in a usb3 reader.

Against external SSD / HDD disks would also be interesting. A Samsung T5 or T7 perhaps, the T5 at least is known to be pretty decent for USB external SSD, though I think EXT4 performing better would be a given and F2FS probably would still do well.

[Croco44]

Does anyone know if exFAT supports TRIM and whether any of the linux exFAT drivers support it? Also worth pointing out... without TRIM, the order in which these benchmarks were run may matter... even if the drive was re-formatted in between tests. Re-formatting a flash drive doesn't guarantee (I think) that the flash controller marks the blocks as free. For some flash drives a Secure Erase command can "reset" the drive to a state where all blocks are marked as free, but this doesn't work with all drives.

[bug77]

Strange it manages to lose two fights to the Fuse driver. Maybe those tests are acting up?

[fkoehler]

Strange it manages to lose two fights to the Fuse driver. Maybe those tests are acting up?

The other FSs probably try to steal the benchmark with illegal reads and writes. There is no way, kernel driver loses to FUSE, unless FUSE is cheating and the benchmark is rigged, in which case we will not accept it.

Also Michael treats kernel Exfat very very badly and very very unfair. All I know, kernel exfat was ahead, it was winning huge on Application startup time and then all those late tests came in, right out of nowhere. Benchmark should have stopped right then !!!!

Make exfat great again.

[cyclistefou]

Wait, dozens of seconds for opening xterm or gnome-terminal? On any FS? What kind of test conditions are those?

[Schugy]

I think people would be interested in a part II of this article where you compare these numbers against a samsung high endurance sdcard in a usb3 reader.

After use of UFS cards in Acasis UFS USB 3.0 readers I prefer them over UHS-I/UHS-II SD cards. Downside is the JMicron chip that doesn't support UAS. SDexpress is going to be even better if the SDA is no concern.