Skip to content

Add RTX 5090 numbers to the README - #76

Merged
praveenperera merged 4 commits into
masterfrom
docs/rtx5090-readme
Oct 23, 2026
Merged

praveenperera merged 4 commits into
masterfrom
docs/rtx5090-readme

Conversation

@praveenperera

@praveenperera praveenperera commented Oct 23, 2026 •

Copy link
Copy Markdown
Member

Adds RTX 5090 results to the README.

  • VoxConverse dev table: RTX 5090 cuda 2922x and cuda-fast 5633x, with the hardware they ran on
  • All-datasets table: RTX 4090 and RTX 5090 RTFx side by side; DER is the same on both GPUs
  • Tested NVIDIA GPUs: RTX 5090 at 2500x with cuda and 2501x with cuda-rtx50

The tested-GPU row uses the same method as the other rows: the first 10 VoxConverse test files (6700.9 s of audio), process wall time including start-up and model loading, 16 CPUs, median of 3 alternating rounds after one warm-up per build. Both builds wrote byte-identical RTTMs in every round.

Summary by CodeRabbit

  • Documentation
    • Added RTX 5090 VoxConverse benchmark results for CUDA and CUDA-fast, including diarization error rates and processing speeds.
    • Expanded cross-dataset benchmark comparisons to show RTX 4090 and RTX 5090 results for both CUDA modes, with updated runtime summaries.
    • Added RTX 5090 results to the tested-GPU table and updated notes on tuned GPU profiles.

Show the RTX 5090 next to the RTX 4090 in the VoxConverse dev table and
the all-datasets table, and add its row to the tested GPUs table. The
tested row is the median of 3 process-wall runs on the first 10
VoxConverse test files, pinned to 16 CPUs, with identical RTTMs from
the cuda and cuda-rtx50 builds.
@coderabbitai

coderabbitai Bot commented Oct 23, 2026 •

Copy link
Copy Markdown

Review in Change Stack →

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration
  • Source mode: Auto
  • Repository: Unavailable
  • Branch: Unavailable
  • File: No configuration file from this repository applied
  • Commit: Unavailable
  • Layers: defaults
  • Review profile: CHILL
  • Plan: Advanced
  • Run ID: 435f7768-b926-486b-9cd8-845b8acdbe97
📥 Commits

Reviewing files that changed from the base of the PR and between 6537347 and 21237d9.

📒 Files selected for processing: 2
  • README.md
  • src/lib.rs
🚧 Files skipped from review as they are similar to previous changes: 2
  • README.md
  • src/lib.rs

Included review availability: This review used your included allowance. Your plan provides up to 2 included reviews per hour; 0 remain after this review.

📝 Walkthrough

The benchmark documentation adds RTX 5090 VoxConverse dev results and compares RTX 4090 and RTX 5090 performance across CUDA modes and datasets. It also updates aggregate processing-time estimates and the tested-GPU table.

Changes: GPU benchmark results

Layer / File(s) Summary
VoxConverse benchmark results
README.md, src/lib.rs
The VoxConverse dev results add RTX 5090 throughput and DER alongside RTX 4090 results. The documentation reports RTX 5090 benchmark details.
Cross-dataset and tested-GPU results
README.md, src/lib.rs
The all-datasets tables compare RTX 4090 and RTX 5090 results for cuda and cuda-fast, and update aggregate runtime estimates. The tested-GPU tables add RTX 5090 throughput and update tuned kernel-profile notes.

Priority: ⬇️ Low

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Other

Possibly related PRs

  • avencera/speakrs#75: Adds RTX 5090 CUDA kernels, device selection, and inference changes associated with the documented results.
Merge Risk: ⚪ Minimal · up to 21237

The RTX 5090 benchmark documentation is consistent across both files and is ready to merge after normal checks.

Pre-merge checks: Passed 5
Check name Status Explanation
Description Check Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check Passed The title clearly and concisely describes the main change: adding RTX 5090 benchmark results to the README.
Docstring Coverage Passed No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 1…
Linked Issues check Passed Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check Passed Check skipped because no linked issues were found for this pull request.
Full details: Docstring Coverage

Explanation

No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check. Docstring coverage is scoped to functions touched by this diff. Analyzed 0 functions across 1 files. (1 skipped: 1 unsupported.)

✨ Finishing Touches
📝 Generate docstrings
  • Commit to this branch
  • Create a new PR
🧪 Generate unit tests (beta)
  • Commit to this branch
  • Create a new PR
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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

Comment @coderabbitai help to get the list of available commands.

Signed-off-by: Praveen Perera <me@praveenperera.com>
@greptile-apps

greptile-apps Bot commented Oct 23, 2026 •

Copy link
Copy Markdown

RetriggerView in GreptileConfidence Score: 5/5

[Low impact] This documentation-only PR appears safe to merge.

Summary

Adds RTX 5090 benchmark results to README.md and the matching crate docs in src/lib.rs.

  • VoxConverse results now show RTX 5090 speed and accuracy.
  • The dataset table compares RTX 4090 and RTX 5090 speeds.
  • The tested GPU table now includes RTX 5090 results.

Reviews (1) · Last reviewed commit: "Add RTX 5090 numbers to the README" · Reviewed by Greptile

@praveenperera
praveenperera merged commit beb3073 into master Oct 23, 2026
25 checks passed
@praveenperera
praveenperera deleted the docs/rtx5090-readme branch October 23, 2026 20:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant