Welcome, Guest. Please login or register.

Author Topic: PFS3 slow or fast?  (Read 4282 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline Karlos

  • Sockologist
  • Global Moderator
  • Hero Member
  • *****
  • Join Date: Nov 2002
  • Posts: 16879
  • Country: gb
  • Thanked: 5 times
    • Show all replies
Re: PFS3 slow or fast?
« on: May 15, 2011, 11:30:02 AM »
It's not generally the raw disk transfer rate that is being discussed when people compare filesystems (as it always tends to be limited by the hardware interface), but the speed of filesystem operations in general. In other words, how quickly does it scan directories, open files, move, delete and so on.
int p; // A
 

Offline Karlos

  • Sockologist
  • Global Moderator
  • Hero Member
  • *****
  • Join Date: Nov 2002
  • Posts: 16879
  • Country: gb
  • Thanked: 5 times
    • Show all replies
Re: PFS3 slow or fast?
« Reply #1 on: May 15, 2011, 12:07:27 PM »
Quote from: AmiDude;637934
OK, thanks for the explanation, I understand. Do you know which program I can
use for measuring the speed of the filesystem?


There are some old tools, the name of which presently escapes me, that were able to specifically benchmark various facets of the filesystem. However, SysSpeed does have some tests, if I recall clearly.

If you really wanted to do a like for like comparison, the only real way to do it would be to format a partition as one file system, copy a load (many thousands) of files and directories on it and time how long it takes to copy them there in the first place, list them all (and any other routine tests you can think of) before finally timing how long it takes to delete them all. Then do exactly the same tests after reformatting the same partition with the new filesystem and compare results.

This won't give you an absolute performance ratio since you still have the constant time it takes to read all the data from the original source drive in the first step and the overhead of physically drawing and scrolling your console but if you are careful in running your tests (using exactly the same methods in each case) these contributing terms should be identical in each case. You could probably minimise their total contribution by redirecting the output of each test to NIL.
int p; // A