docs: add details in gpu selection

[nicolasboisseau]

Added an example of GPU labelling in the documentation to make it more explicit. (It can help snakemake newcomers)

Summary by CodeRabbit

• Documentation
  • Updated GPU configuration guidance: use the machine-specific label nvidia_gpu (e.g., nvidia_gpu=1).
  • Noted that on n1-* machines nvidia_gpu=1 defaults to an NVIDIA Tesla T4.
  • Added guidance and examples for selecting a different GPU by specifying a machine-compatible GPU type (for example, nvidia-tesla-v100).

_{✏️ Tip: You can customize this high-level summary in your review settings.}

[coderabbitai]

📝 Walkthrough

Walkthrough

Docs updated in docs/further.md to reference the machine label nvidia_gpu (replacing gpu_nvidia), note that nvidia_gpu=1 on n1-* machines defaults to nvidia-tesla-t4, and show how to override the GPU type with a machine-compatible label (e.g., nvidia_gpu='nvidia-tesla-v100').

Changes

Cohort / File(s)	Summary
Documentation Updates docs/further.md	Replaced gpu_nvidia reference with nvidia_gpu; added note that nvidia_gpu=1 on n1-* defaults to nvidia-tesla-t4; added guidance to specify GPU type via machine-compatible label (e.g., nvidia_gpu='nvidia-tesla-v100')

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'docs: add details in gpu selection' directly aligns with the main change: updating GPU documentation with explicit notes and guidance on GPU selection.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@docs/further.md`:
- Around line 114-115: Correct the grammar in the sentence on line 114 by
removing the extra "the" and hyphenating "machine-compatible": change "using the
a machine compatible label" to "using a machine-compatible label" while keeping
the example parameter `nvidia_gpu='nvidia-tesla-v100' unchanged.

📜 Review details

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 2daff72 and 0714e96.

📒 Files selected for processing (1)

• docs/further.md

🧰 Additional context used

🪛 LanguageTool

docs/further.md

[grammar] ~114-~114: Use a hyphen to join words.
Context: ...he gpu type directly using the a machine compatible label: e.g. nvidia_gpu='nvidi...

(QB_NEW_EN_HYPHEN)

🔇 Additional comments (1)

docs/further.md (1)

112-112: Good addition for clarity.

This explicitly documents the default GPU type for n1-* machines, which will help users understand the behavior when they specify gpu_nvidia=1.

_{✏️ Tip: You can disable this entire section by setting review_details to false in your review settings.}

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@docs/further.md`:
- Around line 110-115: Update the incorrect parameter name used in the
documentation from gpu_nvidia to nvidia_gpu so it matches the codebase; replace
occurrences of "gpu_nvidia=1" with "nvidia_gpu=1" in the example text and ensure
the wording aligns with the usage in executor.py and the hello-world-gpu
Snakefile, preserving the note about machine type and the machine-compatible
label example (nvidia_gpu='nvidia-tesla-v100').

🧹 Nitpick comments (2)

docs/further.md (2)

112-112: Correct grammar, capitalization, and punctuation.

Line 112 has several style issues:

• Use "an" instead of "a" before "n1-*" (vowel sound)
• Capitalize "GPU" for consistency with technical documentation standards
• Add period at the end of the sentence

📝 Proposed fix

-On a n1-* family machine type, gpu_nvidia=1 will trigger a "nvidia-tesla-t4" gpu by default.
+On an n1-* family machine type, gpu_nvidia=1 will trigger an "nvidia-tesla-t4" GPU by default.

---

114-114: Capitalize "GPU" for consistency.

For consistency with technical documentation standards, "gpu type" should be "GPU type".

📝 Proposed fix

-It's possible to change the gpu type directly using a machine-compatible label:
+It's possible to change the GPU type directly using a machine-compatible label:

📜 Review details

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 0714e96 and 5e26c91.

📒 Files selected for processing (1)

• docs/further.md

_{✏️ Tip: You can disable this entire section by setting review_details to false in your review settings.}

[coderabbitai]

Actionable comments posted: 1

🤖 Fix all issues with AI agents

In `@docs/further.md`:
- Line 112: Update the sentence containing "a n1-*" and the `nvidia_gpu=1`
claim: change "a n1-*" to "an n1-*", remove the incorrect assertion that
`nvidia_gpu=1` defaults to "nvidia-tesla-t4", and replace it with a clear
statement that Batch does not default a GPU type when only the GPU count is
set—you must explicitly specify the GPU type or use an accelerator-optimized
machine type; reference the exact tokens `nvidia_gpu=1` and "nvidia-tesla-t4"
when editing so the incorrect claim is removed and the corrected wording is
inserted.

📜 Review details

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 5e26c91 and a6116c7.

📒 Files selected for processing (1)

• docs/further.md

🔇 Additional comments (2)

docs/further.md (2)

110-110: LGTM! Parameter name corrected.

The documentation now correctly uses nvidia_gpu=1 instead of the previous incorrect gpu_nvidia. This aligns with the codebase usage.

---

114-115: No changes needed. The documented syntax nvidia_gpu='nvidia-tesla-v100' is valid and correctly supported by the executor implementation. The codebase confirms that the nvidia_gpu parameter accepts both numeric values (for GPU count, with default type "nvidia-tesla-t4") and string values (for GPU model type specification).

_{✏️ Tip: You can disable this entire section by setting review_details to false in your review settings.}