Reduce code size of node

[cuttlfish]

Hi,

Is there any way to reduce the code size of nodejs. On a cross compile platform it comes to be about 9 MB.

Also is there a way to reduce initial heap allocated by V8

Thanks
David

[mikeal]

pinging @trevnorris since he likely knows the answer.

[trevnorris]

By code size do you mean the built binary? As for v8, there are ways to constrain how much maximum memory it can use, but don't know of a way to limit memory consumption at startup.

[cuttlfish]

Hi Trevor,

Yes I mean the built binary. I tried changing the v8 gcc flag to -Os, that
did succeed in making it smaller, but starts up slower I guess due to less
optimization

David

On Fri, Oct 23, 2015 at 11:41 AM, Trevor Norris notifications@github.com
wrote:

By code size do you mean the built binary? As for v8, there are ways to
constrain how much maximum memory it can use, but don't know of a way to
limit memory consumption at startup.

—
Reply to this email directly or view it on GitHub
#19 (comment).

[trevnorris]

I'm not really sure. you could compile w/o openssl if you don't need crypto.

[srl295]

@cuttlfish what operating system and platform?

[cuttlfish]

This is linux running on a ARM platform that is similar to raspberry pi

On Wed, Dec 9, 2015 at 5:06 PM, Steven R. Loomis notifications@github.com
wrote:

@cuttlfish https://github.com/cuttlfish what operating system and
platform?

—
Reply to this email directly or view it on GitHub
#19 (comment).

[amulpatel]

I am also looking to reduce node memory size around 20 MB in openwrt. currently it is consuming around 50 MB.

is there any way to reduce memory consumption ? even i am ready to change code to reduce memory i.e v8, uv or other deps.

Thanks,
Amul Patel

[bnoordhuis]

There is an --optimize_for_size flag in newer versions of node and you can limit the size of the heap with --max_old_space_size=... (and --max_semi_space_size and --max_executable_size) in all versions.

You can also tweak common.gypi to build at -Os instead of -O3. The smaller binary, besides using less memory, also tends to run faster on systems with small L1 code caches.