0

Summary

USB drives connected to the rev2 running Raspbmc read/write twice as quickly as the rev1 running Raspbian. Why is there a difference in speed?

Testing

I tested the same external USB hard drive connected to both machines. I created a 50 MB test file from random data.

$ dd if=/dev/urandom of=/media/HDD/randomness bs=1M count=50

On the Raspbmc (two replicates):

$ time cp randomness randomness2
real    0m1.922s
real    0m1.984s

On the Raspbian (two replicates):

$ time cp randomness randomness2
real    0m4.649s
real    0m4.709s

I compared the mount options. On the Raspbmc

$ mount | grep media
/dev/sda1 on /media/HDD type ext4 (rw,nosuid,nodev,uhelper=udisks)

On the Raspbian (mounted with -o noatime)

$ mount | grep media
/dev/sdb1 on /media/HDD type ext4 (rw,noatime,data=ordered)

I tested mounting the Raspbian HDD with noatime,nosuid,nodev,data=writeback, but I still get read/write times that are about double that of the Raspbmc

real    0m5.127s
real    0m4.966s

Why is the rev1 Raspbian Pi hard drive not as quick as the rev2 Raspbmc?

  • @Wilf I actually create a static random file, then reuse the same file for the benchmarks. – Sparhawk Jul 26 '14 at 22:41
0

This appears to have not much to do with the distribution in use. I tried swapping the rev1 and rev2 boards around. When running the Raspbian system on the rev2 board, I can get similar speeds to the Raspbmc.

I was quite surprised, because I see no references to changes in USB throughput in the rev 2 release notes. However, it seems that there is a difference in RAM in my case.

Adding free -m and GPU allocation from /boot/config.txt, my rev1 has 231+16=247 MB, and the rev2 has 374+128=502 MB. Reading further, it seems that rev1s and early rev2s have 256 MB of RAM, and later-model rev2s have 512 MB (although I'm not sure why the slight discrepancy in my case).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.