OpenWrt Forum Archive

Topic: stable vs trunk performance

The content of this topic has been archived on 29 Apr 2018. There are no obvious gaps in this topic, but there may still be some posts missing at the end.

Hi,

Is there such a thing like a development/debug build for OpenWRT?
Could there be a reason for a performance difference between a stable image downloaded from this website and a manually built image from trunk?
(I'm thinking maybe some optimization flags)

This is totally speculative, but a couple of months ago I was testing the openvpn performance on an WR1043ND and I noticed a slight difference (10-15%) between my trunk build (slower) and stock Barrier Breaker 14.07. (faster)
I was using the exact same settings, keys and clients, and tested on wire/LAN.
Also, both builds used the same openvpn and openssl versions.

I know that there are other factors that may come into play, I'm just asking to make sure this isn't one of them.

Thanks.

Well I supposed that it is inevitable.
When you keep adding new functions and processes on a resource constrained embedded device something got to give without optimizing existing stuff for every new stuff added.

I understand that. I guess I didn't ask the question right.
I'm interested strictly in build options/settings, if there are any differences between trunk and stable, which could affect the performance in any way.

Thanks.

Not answering your question... but OpenWrt is usually built with -Os and if you build it yourself you can use -O2. It will probably produce a larger and faster image.

Tested Os and O2 has not much difference in performance but big difference in size on mips32r2

(Last edited by alphasparc on 19 Jun 2015, 14:30)

Also, if you build your own firmware, you can select/deselect any packages/ptions to include/exclude, respectively, into/from firmware. You can even trim the kernel as you wish.

The discussion might have continued from here.