Implement approx_tanh for ROCm using OCML tanh function

[phambinhfin]

AMD CDNA3 (MI300X/gfx942) does not have a hardware tanh instruction like NVIDIA's PTX tanh.approx. This implements approx_tanh for ROCm using:

• For f32 (and f16/bf16 via casting): Triton's __triton_hip_fast_tanhf
  which uses a fast exp-based formula: tanh(x) = (exp(2x) - 1) / (exp(2x) + 1)
• For f64: OCML's __ocml_tanh_f64 (AMD's Open Compute Math Library)

Changes:

• Add f64 support to approx_tanh function
• Add ROCm platform detection in _elementwise_inline_asm_lowering
• Add _approx_tanh_rocm_lowering function for ROCm-specific lowering
• Add test_approx_tanh test with f16/bf16/f32/f64 support

Related commit form ROCm/jax PR #614
See: triton-lang/triton#7780

[gemini-code-assist]

Summary of Changes

Hello @phambinhfin, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request enhances the approx_tanh function within JAX's Pallas Triton integration to support AMD's ROCm platform, specifically addressing the lack of a hardware tanh instruction on CDNA3 GPUs. It introduces a specialized lowering mechanism for ROCm, leveraging Triton's fast exponential-based tanh for f32 and AMD's Open Compute Math Library for f64, ensuring functional parity and performance across different floating-point precisions on AMD hardware.

Highlights

• ROCm-specific approx_tanh Implementation: Introduced a specialized lowering mechanism for the approx_tanh function on AMD's ROCm platform, addressing the absence of a hardware tanh instruction on CDNA3 GPUs.
• Extended Data Type Support: The approx_tanh function now supports float64 on ROCm, utilizing OCML's __ocml_tanh_f64. For float32, float16, and bfloat16, it leverages Triton's __triton_hip_fast_tanhf (with casting for f16/bf16).
• Platform-Aware Lowering: Modified the _elementwise_inline_asm_lowering function to detect the ROCm platform and dispatch tanh.approx operations to the new ROCm-specific lowering function, bypassing PTX inline assembly where necessary.
• Comprehensive Testing: Added new unit tests for approx_tanh covering float16, bfloat16, float32, and float64 data types, including conditional skips for unsupported hardware or configurations.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[gemini-code-assist]

Code Review

This pull request implements approx_tanh for ROCm using OCML's tanh function for f64 and Triton's fast tanh function for f32, f16, and bf16. It includes changes to jax/_src/pallas/triton/primitives.py to add f64 support and ROCm platform detection, adds _approx_tanh_rocm_lowering function, and adds a test case in tests/pallas/ops_test.py with f16/bf16/f32/f64 support. The review identifies a potential issue with the library path in _approx_tanh_rocm_lowering.