Getting error when loading wasm backend

System information

• OS : Windows 7 - 32 bit
• Hardware : PC : Pentium Dual Core CPU E5400 @ 2.70 GHz 2.69 GHz , RAM : 1 GB (also seen in a machine with RAM of 4GB)
• TensorFlow.js installed from (npm or script link): Script
• TensorFlow.js version : 2.3
• Browser version: Chrome Version 85

The functionality(loading of backend ‘wasm’) works fine for most of the machines I am using, but there are few machines where I have noticed that the wasm backend fails to load. The error says : Error: Backend ‘wasm’ has not yet been initialized. Make sure to await tf.ready() or await tf.setBackend() before calling other methods

Although I am pretty sure that I am not executing any other method before promise is resolved as shown in below code.

try {
        BlazeFaceModel = await blazeface.load();
        tf.setBackend('wasm').then(() => callNext());

    } catch (error) {
        console.error(error);
    }

I have seen this issue mostly in machines with low end hardware and when webgl is not available.

Also it will be helpful if you can clarify some of the below:

1. If we request tf to set a certain backend and if it is not available, will there be fallback to other backends?
2. When I have put the command, tf.setBackend(‘wasm’). Why is webgl loading? Is webgl loaded first and then the change in backend happens?
3. I have used a promise based approach (similar to await as mentioned in the console error) to execute next command only after the backend is set. Why does the error ask me to await again? Should I use both await tf.ready() with await tf.setBackend() ?

[rthadur]

you are right it is supported ,let wait for @annxingyuan response. Thank you

[google-ml-butler[bot]]

Closing as stale. Please @mention us if this needs more attention.