Support qcs6490 v68 arch for QNN backend

[amo33]

🚀 The feature, motivation and pitch

Inside qnn delegate, HTP hardware architecture available for HTP backend options starting from V68 to V75.
We can see that supported Chipset are snapdragon versions as below.

Supported Chipset

• Snapdragon 8 Gen 1
• Snapdragon 8 Gen 1+
• Snapdragon 8 Gen 2
• Snapdragon 8 Gen 3

However, we like to utilize QCS6490 whose DSP Hexagon Arch is V68.

Is there any chance by just adding QCS6490 to utils.py , and qc_compiler_spec.fbs would work in executorch?

I have successfully used QCS6490 in the tutorial_for_llama2 provided by QPM (Qualcomm Package Manager). The key question is: would the results be the same when using tutorial_for_llama2 versus Executorch?

Alternatives

No response

Additional context

No response

RFC (Optional)

No response

[cccclai]

It should be similar. If you're interested in contributing and add the SoC, feel free to submit a PR and add the reviewers listed in https://github.com/pytorch/executorch/tree/main/backends/qualcomm#pull-requests for review.

#6986 is another example PR to add the SoC