Add NVIDIA RTX 5090 (Blackwell) and H100 GPU data #60
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
This PR adds the latest NVIDIA GPUs to the calculator:
Why This Matters
The RTX 5090 is the first consumer Blackwell GPU. We conducted real-world energy benchmarks on it and discovered an important finding for the Green AI community:
Key Finding: 4-bit quantization increases energy consumption by 26-29% for small models (<3B parameters) on high-performance GPUs, challenging the common assumption that quantization always saves energy.
Data Sources
Related Work
Full benchmark report and calculator available at: https://github.com/hongping-zh/ecocompute-ai
Thank you for maintaining this valuable tool for the Green AI community! 🌱