tf.profile should also return byte usage as observed by the texture manager

[annxingyuan]

The TextureManager keeps track of the status of each texture, and therefore has a truer record of byte usage than the engine, because even though we call dispose on tensors, they might still be sitting around, which means the real memory footprint could be larger than what the engine thinks. We should surface the TextureManager's record of byte usage in tf.profile.

Since this refers to the number of bytes allocated physically for webgl use, it should be surfaced with a webgl prefix, e.g. like webglBytesUsed. This field should be produced by the backend and bubbled up in the engine.

Also, numBytes is wrong for packed ops - it should be based on texture shape not logical shape

[gaikwadrahul8]

Hi, @annxingyuan

Thank you for opening this issue for tracking purposes. Since this issue has been open for a long time, the code/debug information for this issue may not be relevant with the current state of the code base.

The TFJs team is constantly improving the framework by fixing bugs and adding new features. We suggest you try the latest TFJs version with the latest compatible hardware configuration which could potentially resolve the issue. We can keep the issue open if it is still relevant. Please confirm if we need to keep the issue open.

Thank you for your support and cooperation.