[LEDE-DEV] What optimizations does low-end MIPS userspace need?

Alexandru Ardelean ardeleanalex at gmail.com
Sat Apr 15 23:55:39 PDT 2017


On Sun, Apr 16, 2017 at 1:47 AM, Jay Carlson <nop at nop.com> wrote:
>
>> On Apr 15, 2017, at 1:57 AM, Alexandru Ardelean <ardeleanalex at gmail.com> wrote:
>>
>> On Sat, 15 Apr 2017 at 02:01, Jay Carlson <nop at nop.com> wrote:
>>
>>> I’ve been fooling around with various userspace toolchain hacks for mips, trying to make things faster globally. But then I realized: I don’t know what “faster” is. Or what “better” is, in general.
>>>
>>> I want my Python/Firmata apps to launch faster on mt7688, but there is a clear metric for winning there; I don’t need your help. :-)
>>
>> Strictly speaking for Python (if you haven't already) you might want to check some recent changes.
>>
>> Python now ships with bytecode files instead of source.
>> Seems that app startup is improved by even a factor of 9x on some low end CPUs.
>> Something like 500 msecs to 70 msecs to start a simple Hello World script.
>
> Very cool. That’s
>
> "python: use default host build prefix, remove cross-compile workarounds for host” [1]
>
> and
>
> "python,python3: drop remove .pyc & .pyo files” [2]
>
> right?
>

Ah, last reply got through only to you, and not the list.
I've sent it from my phone, and for some reason, the Plain Text mode
wasn't enabled on it.
Will have to check that ; my browser has it enabled.

The commits are in this PR:
https://github.com/openwrt/packages/pull/4130/commits
Some are cleanups, and a few actually deal with the byte-code changes.

This commit should be the more important one : "python: split source
packages away from compiled packages "
But, it has some pre-work commits.


> --
> Jay Carlson
> nop at nop.com
>
> [0]: Am I supposed to be citing the openwrt package repository? If so, how?

not sure ; i don't know of a rule

> [1]: https://github.com/openwrt/packages/commit/b70b978cc61bdd6fcdc7f8f7bae8ee3ef16baf4a
> [2}: https://github.com/openwrt/packages/commit/05f8d6edf06c93dadab34f14a5fad48449dd7eda
>
>



More information about the Lede-dev mailing list