Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
PrefixLM Attention Performance on Llama2-7B (8k Context, 32/32 Heads)
Loading...
163.7
TFLOPS (PrefixLM Attention)
CuBridge
8.5944
48.8622
89.13
129.3978
May 6, 2026
TFLOPS (PrefixLM Attention)
Updated 27d ago
Evaluation Results
Method
Method
Links
TFLOPS (PrefixLM Attention)
CuBridge
Implementation=CuBridge
2026.05
163.7
FlexAttention
Implementation=PyTorch...
2026.05
142.77
Qimeng Attn
Implementation=Qimeng...
2026.05
122.44
Torch
Implementation=Standar...
2026.05
14.56
Feedback
Search any
task
Search any
task