0

A week ago I upgraded my main hard disk to an SSD (ADATA SU800) hoping to improve the general performance of my PC. I did a clean install of windows (10) and Linux (Pop-OS). I did notice a great improvement on windows, with programs opening up almost instantly. But in Linux, I barely noticed some improvements.

Today, I ran a benchmark in both systems, in windows using crystaldiskmark and in Linux a similar program named almost the same, kdiskmark. Herein I show the results I got.

Windows benchmark Linux benchmark

Why does Linux have the poorest performance? It is related to compatibility or drivers? Is it possible to improve it and get the same results as windows?


Both systems were completely idle (no more programs opened by the user). And I know the results are similar but I want to get the maximum possible performance. I ran the test only one time under both systems but maybe I must have done it thrice or more. And for more information, the Windows partition is NTFS and the Linux one is EXT4.

4
  • You're also testing the filesystems. Mar 23, 2021 at 20:05
  • Did you change to noatime on Linux partition? askubuntu.com/questions/1284302/… And some systems use different schedulers, not sure what default is with Pop. This basically means that data may be written to the disk before the journal. Speed over safety see man tune2fs Not sure what Windows NTFS default is, performance or safety.
    – oldfred
    Mar 23, 2021 at 21:27
  • I don't have the evidence to back this up but my personal experience with windows is that it is slow when accessing small files. It seems to spend a lot seeking. Anecdotally I've seen windows be 10 times slower on a HDD than Linux when accessing millions of small files (same program). An SSD will make up a lot of difference for Windows and provide a smaller boost to Linux. That's just my own personal experience. These benchmarks seem close. Mar 23, 2021 at 21:32
  • In addition, a new or recently defragmented NTFS partition works better, but over time it starts to had bits of files in different locations. With HDD that made a bigger different as another move of drive head, with SSD not as critical. Linux ext4 had little fragmentation by default as it normally writes entire file, but anywhere within partition. This site has many performance tests: phoronix.com/…
    – oldfred
    Mar 24, 2021 at 2:39

1 Answer 1

0

How many times have you run the test under Linux? Was the system completely idle (a web browser running in parallel could actually affect your results in a major way)? Lastly the results look quite similar: I don't see anything to start worrying about - it's not as if Linux performs twice as poor. Also, applications look similar but may use very different underlying read/write routines, OSes work with storage in slightly different ways, there's timing issues, dynamic CPU frequency control which works differently - in the end you cannot compare the results at face value.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .