C++ · MIT
JakeATX/llamAmpere
llama.cpp fork for significantly improved performance on Ampere generation Nvidia cards (especially RTX 3090 / 3090 Ti): example: 110+ tok/s over a 100K-token generation at temperature 1 for qwen3.8, supporting full 262K ctx. Supports low memory (12gb) configs for 200k+ ctx
★ 221
Stars
31 forks · 13 open issues
+3.3/day
Growth · measured
Stars counted between two of our own snapshots roughly a week apart. The repository page gives the exact span.
42.1
Momentum / 100
see the breakdown below
2mo ago
Created
last push 3d ago
Star history
How this score was built
| Component | Weight | Earned | Why |
|---|---|---|---|
| Velocity | 55 | 12.1 | +3.3/day, log-scaled against a 750/day ceiling |
| Acceleration | 15 | 8.3 | last week ran 1.55× the monthly pace |
| Freshness | 15 | 15.0 | last push 3d ago; decays to zero at 180 days |
| Scale | 15 | 6.8 | 221 stars, log-scaled against 150k — stops a 3-star repo topping the board |
| Momentum | 100 | 42.1 | sum of the above |
Detail
| Topics | ampereconsumer-gpucudaggmlggufkv-cache-quantizationllama-cppllmllm-inferencelocal-llmlong-contextmtpmulti-token-predictionquantizationqwenqwen3rtx-3090rtx-3090-tispeculative-decodingturboquant |
| Matched on | LLM & Inference (llm, llm-inference) |
| Short window | +23 stars over 7 days = 3.3/day |
| Long window | +55 stars over 26 days = 2.1/day |
| Lifetime average | 3.75 stars/day over 59 days |
| First seen here | 2026-10-11 |
| Last read | 2d ago |