add qnnpack path for hardtanh

[vkuzo]

Stack from ghstack:

• add qnnpack path for hardtanh #35779 add qnnpack path for hardtanh

Summary:

Adds a QNNPack path for the clamp kernel, which is useful for
hardtanh.

Test Plan:

python test/test_quantized.py TestQNNPackOps.test_hardtanh

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: D20778588

[dr-ci]

💊 Build failures summary and remediations

As of commit a5d4275 (more details on the Dr. CI page):

---

• 2/2 failures possibly* introduced in this PR
  • 1/2 non-CircleCI failure(s)

---

🕵️ 1 new failure recognized by patterns

The following build failures do not appear to be due to upstream breakages:

pytorch_linux_backward_compatibility_check_test (1/1)

Step: "Test" (full log | pattern match details | 🔁 rerun)

May 04 19:13:47 The PR is introducing backward incompatible changes to the operator library. Please contact PyTorch team to confirm whether this change is wanted or not.

May 04 19:13:47 processing existing schema:  aten::_weight_norm_cuda_interface_backward(Tensor grad_w, Tensor saved_v, Tensor saved_g, Tensor saved_norms, int dim) -> (Tensor, Tensor) 
May 04 19:13:47 processing existing schema:  aten::_weight_norm_differentiable_backward(Tensor grad_w, Tensor saved_v, Tensor saved_g, Tensor saved_norms, int dim) -> (Tensor, Tensor) 
May 04 19:13:47 processing existing schema:  aten::zeros_like(Tensor self, *, int? dtype=None, int? layout=None, Device? device=None, bool? pin_memory=None, int? memory_format=None) -> (Tensor) 
May 04 19:13:47 processing existing schema:  aten::_sparse_sum(Tensor self) -> (Tensor) 
May 04 19:13:47 processing existing schema:  aten::_sparse_sum.dtype(Tensor self, *, int dtype) -> (Tensor) 
May 04 19:13:47 processing existing schema:  aten::_sparse_sum.dim(Tensor self, int[1] dim) -> (Tensor) 
May 04 19:13:47 processing existing schema:  aten::_sparse_sum.dim_dtype(Tensor self, int[1] dim, *, int dtype) -> (Tensor) 
May 04 19:13:47 processing existing schema:  aten::rsub.Tensor(Tensor self, Tensor other, *, Scalar alpha=1) -> (Tensor) 
May 04 19:13:47 processing existing schema:  aten::rsub.Scalar(Tensor self, Scalar other, Scalar alpha=1) -> (Tensor) 
May 04 19:13:47 processing existing schema:  aten::_sparse_addmm(Tensor self, Tensor sparse, Tensor dense, *, Scalar beta=1, Scalar alpha=1) -> (Tensor) 
May 04 19:13:47 The PR is introducing backward incompatible changes to the operator library. Please contact PyTorch team to confirm whether this change is wanted or not.  
May 04 19:13:47  
May 04 19:13:47 Broken ops: [ 
May 04 19:13:47 	aten::as_tensor.float(float t, *, int? dtype=None, Device? device=None) -> (Tensor) 
May 04 19:13:47 	aten::as_tensor.int(int t, *, int? dtype=None, Device? device=None) -> (Tensor) 
May 04 19:13:47 	aten::as_tensor.bool(bool t, *, int? dtype=None, Device? device=None) -> (Tensor) 
May 04 19:13:47 	aten::tensor.float(float t, *, int? dtype=None, Device? device=None, bool requires_grad=False) -> (Tensor) 
May 04 19:13:47 	aten::tensor.int(int t, *, int? dtype=None, Device? device=None, bool requires_grad=False) -> (Tensor) 
May 04 19:13:47 	aten::tensor.bool(bool t, *, int? dtype=None, Device? device=None, bool requires_grad=False) -> (Tensor) 
May 04 19:13:47 	_caffe2::GatherRanges(Tensor data, Tensor ranges, Tensor[]? _caffe2_preallocated_outputs=None) -> (Tensor, Tensor) 
May 04 19:13:47 	aten::channel_shuffle(Tensor self, int groups) -> (Tensor) 

---

ci.pytorch.org: 1 failed

• Failed: pr/py3.6-clang7-rocmdeb-ubuntu16.04

---

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions on the GitHub issue tracker.

See how this bot performed.

This comment has been revised 39 times.

[supriyar]

Can we also compare the results against the server implementation? I think we can combine the test for server and mobile for this op.

[vkuzo]

sure, we can reuse the body of the test, it probably does have to live in a separate wrapper though (check out the specs on class TestQNNPackOps for motivation)

[vkuzo]

seems like only a few lines would be able to be reused, so left it as is for now (seemed overkill), but flexible - lmk!

[z-a-f]

LGTM

[facebook-github-bot]

This pull request has been merged in faad00a.