nvdla delegate with custom input and weight formats

[Jas Math]

Hi,

I am looking to create a new delegate for a NVDLA backend, however operations like a convolution require the input (feature data format C'WHC) and weights (C'KWHC) to be in a propriety format. Therefore the data need to be converted before I can invoke the delegate operations. I'm trying determine how I could feed the delegate the custom format? I would like to avoid having to do an on the fly transformation as this will have a impact on performance could negate using NVDLA.

Thanks

[Thai Nguyen]

Hi Jas,

As you can see in our guide (https://www.tensorflow.org/lite/performance/implementing_delegate), there are `Init`, `Prepare` and `Eval` in the delegate interface. So if it is the constant data, then you can convert it inside `Init` or `Prepare`. Otherwise, I am afraid that it needs to be converted on the fly (Ex: in our GL delegate).

Best Regards,

Nguyen Khac Thai

--
You received this message because you are subscribed to the Google Groups "TensorFlow Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tflite+un...@tensorflow.org.
To view this discussion on the web visit https://groups.google.com/a/tensorflow.org/d/msgid/tflite/74c605ed-65a6-4959-a650-772c7028bde0n%40tensorflow.org.