observers: use clamp instead of min/max in calculate_qparams

[vkuzo]

Stack from ghstack:

• observers: use torch.all to check for valid min and max values #43151 observers: use torch.all to check for valid min and max values
• observers: use clamp instead of min/max in calculate_qparams #43150 observers: use clamp instead of min/max in calculate_qparams
• observers: make eps a buffer #43149 observers: make eps a buffer
• quant bench: update observer configs #42956 quant bench: update observer configs

Summary:

The current logic was expensive because it created tensors on CUDA.
Switching to clamp since it can work without needing to create tensors.

yields a ~20% latency improvement on the CUDA microbenchmark for small tensors
(prev diff: P139074571, this diff: P139074706)

Test Plan:

benchmarks

Reviewers:

Subscribers:

Tasks:

Tags:

Differential Revision: D23170427

[dr-ci]

💊 CI failures summary and remediations

As of commit 9724ed7 (more details on the Dr. CI page):

---

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

---

ci.pytorch.org: 1 failed

• Failed: pr/pytorch-linux-xenial-rocm3.5.1-py3.6

---

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 or post in the (internal) Dr. CI Users group.

See how this bot performed.

This comment has been revised 1 time.

[facebook-github-bot]

This pull request has been merged in 3264ba0.