em++: warning: -pthread + ALLOW_MEMORY_GROWTH may run non-wasm code slowly

[John Dallman]

I'm getting this warning from the Emscripten 6.0.4 version of em++ 

em++: warning: -pthread + ALLOW_MEMORY_GROWTH may run non-wasm code slowly, see https://github.com/WebAssembly/design/issues/1271 [-Wpthreads-mem-growth]

Since issue 1271 was closed in May 2025, should the warning still be produced? 

Thanks in advance.

John

[Alon Zakai]

Ah, we need to improve that message. I'll open a PR.

The current situation AFAIK is this:

1. The design repo bug was closed because there is nothing more to be done on the design side. The intended solution is growable SharedArrayBuffer,

https://github.com/tc39/proposal-resizablearraybuffer

2. That landed in major browsers around 2024, but Firefox only implemented it fully in 2026,

https://github.com/emscripten-core/emscripten/blob/bc7be4e0c44eb0159cbdb4d37fea5526a337e3d8/src/runtime_common.js#L133-L134

3. If you are ok with targeting only browsers that support that feature - most modern ones, but only very recent Firefox - you can pass

-sGROWABLE_ARRAYBUFFERS=2

Mode 2 there is "force this feature on, do not build in any fallback support for browsers without full compatibility".

--
You received this message because you are subscribed to the Google Groups "emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to emscripten-disc...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/emscripten-discuss/CAH1xqgk0UWZ7vV15nMj3d-J5jHyEy_p%3DJKLHG-HC7GOrOFVtDg%40mail.gmail.com.

[Alon Zakai]

PR: https://github.com/emscripten-core/emscripten/pull/27697

[John Dallman]

Alon wrote:

> 1. The design repo bug was closed because there is nothing more to be done on the design side. 

> The intended solution is growable SharedArrayBuffer,

I don't think I understand the implications of a growable SharedArrayBuffer. My knowledge of JavaScript is quite limited. 

I'm porting a large mathematical modelling library to WebAssembly, along with its test harness. I am not providing a JavaScript binding for it: the intention is that the consumers of the library will call it from C/C++ code compiled for WebAssembly.  

Does this have implications for the effects of growable SharedArrayBuffer?

Thanks,

John 

To view this discussion visit https://groups.google.com/d/msgid/emscripten-discuss/CAEX4NpRbacbS7GOhp30ZbCXu-Pta_mmcw6f1X_yVAF%3DD6jWLPA%40mail.gmail.com.

[Alon Zakai]

The issue is an interaction with JavaScript, when a growable SharedArrayBuffer is grown. There can be some overhead there, on the JavaScript side.

If you aren't writing any JavaScript yourself, then the danger of a slowdown is small. Though there are some places Emscripten itself uses JavaScript (when you need to call Web APIs, say to render or download a file), so the danger is not zero.

If you don't see a slowdown (say, comparing the non-pthread version to a pthread version that uses only one thread), then there is nothing to worry about.

But if your users are on new-enough browsers, I would recommend building with

-sGROWABLE_ARRAYBUFFERS=2

to avoid any danger of a slowdown.

To view this discussion visit https://groups.google.com/d/msgid/emscripten-discuss/CAH1xqgnZ6cXbDVpPPY3KFvOqbDFLF6UY5QgLNWtmSfUNREGRNA%40mail.gmail.com.

[John Dallman]

Since I don't have any users yet, I'm not shy about requiring reasonably modern browsers. FireFox 154 should be OK. 

Thanks!

John

To view this discussion visit https://groups.google.com/d/msgid/emscripten-discuss/CAEX4NpR0yN%3DYUmqDj_3qBRvUii53XtSxzwiBT%3DfE0D%3Do%3DLWXww%40mail.gmail.com.