Hugging Face's logo

Datasets:
witcheer
/
rtx-5090-benchmarks

Tasks:
Text Generation
Modalities:
Image
Tabular
Text
Formats:
csv
Size:
< 1K
Tags:
benchmark
inference
llm
nvidia
rtx-5090
llama-cpp
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
witcheer commited on
Commit
a081f1b
·
verified ·
1 Parent(s): 7bc8d48

Upload benchmarks.csv with huggingface_hub

Browse files
Files changed (1) hide show
  1. benchmarks.csv +2 -0
benchmarks.csv CHANGED
@@ -316,3 +316,5 @@ Ornith 1.5 9B,Dense (hybrid attn),8.95,Q6_K,6.85,llama.cpp,CUDA,RTX 5090,32,pp51
316
  Ornith 1.5 9B,Dense (hybrid attn),8.95,Q6_K,6.85,llama.cpp,CUDA,RTX 5090,32,tg128,187.71,0.23,2026-08-23
317
  Qwopus3.8-27B-Flash,Dense (hybrid attn),27.32,Q6_K,20.89,llama.cpp,CUDA,RTX 5090,32,pp512,3148.02,65.99,2026-09-07
318
  Qwopus3.8-27B-Flash,Dense (hybrid attn),27.32,Q6_K,20.89,llama.cpp,CUDA,RTX 5090,32,tg128,62.96,0.05,2026-09-07
 
 
 
316
  Ornith 1.5 9B,Dense (hybrid attn),8.95,Q6_K,6.85,llama.cpp,CUDA,RTX 5090,32,tg128,187.71,0.23,2026-08-23
317
  Qwopus3.8-27B-Flash,Dense (hybrid attn),27.32,Q6_K,20.89,llama.cpp,CUDA,RTX 5090,32,pp512,3148.02,65.99,2026-09-07
318
  Qwopus3.8-27B-Flash,Dense (hybrid attn),27.32,Q6_K,20.89,llama.cpp,CUDA,RTX 5090,32,tg128,62.96,0.05,2026-09-07
319
+ Qwen3.8-Flash-Next,MoE (10B active; experts of 28 layers in RAM),125,UD-Q2_K_XL,73.45,llama.cpp,CUDA,RTX 5090,32,pp512,770.14,33.56,2026-09-08
320
+ Qwen3.8-Flash-Next,MoE (10B active; experts of 28 layers in RAM),125,UD-Q2_K_XL,73.45,llama.cpp,CUDA,RTX 5090,32,tg128,52.58,0.22,2026-09-08