Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Complexity and Throughput Analysis on Hardware Efficiency Benchmark H100 Configuration
Loading...
93.8
H100 Throughput (tokens/s)
GDN-2 (matrix-gated)
22.04
40.67
59.3
77.93
Jun 25, 2026
H100 Throughput (tokens/s)
Training FLOPs Complexity
Training Memory Complexity
Inference Complexity per Token
Parallel Depth
Updated 1mo ago
Evaluation Results
Method
Method
Links
H100 Throughput (tokens/s)
Training FLOPs Complexity
Training Memory Complexity
Inference Complexity per Token
Parallel Depth
GDN-2 (matrix-gated)
Hardware=Single H100,...
2026.06
93.8
-
-
-
-
CARVE
Hardware=Single H100,...
2026.06
93.4
-
-
-
-
Content-aware exact recurrent
Hardware=Single H100,...
2026.06
24.8
-
-
-
-
Linear Attention
Hardware=Single H100,...
2026.06
-
2
2
2
-
Delta Rule
Hardware=Single H100,...
2026.06
-
2
2
2
-
Feedback
Search any
task
Search any
task