Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Kernel Throughput Evaluation on MoE Models (OLMoE, Qwen3, DSv3, Mixtral) beta=0.5
Loading...
67
Latency
RaMP
-36.92
664.54
1,366
2,067.46
Apr 28, 2026
Latency
Geomean Speedup
Updated 1mo ago
Evaluation Results
Method
Method
Links
Latency
Geomean Speedup
RaMP
Model=DSv3, S=32
2026.04
67
-
RaMP
Model=OLMoE, S=32
2026.04
72
-
Alpha-MoE
Model=DSv3, S=32
2026.04
77
-
Alpha-MoE
Model=OLMoE, S=32
2026.04
87
-
RaMP
Model=DSv3, S=512
2026.04
178
-
Alpha-MoE
Model=DSv3, S=512
2026.04
238
-
RaMP
Model=OLMoE, S=1024
2026.04
319
-
Alpha-MoE
Model=OLMoE, S=1024
2026.04
384
-
RaMP
Model=Mixtral, S=32
2026.04
617
-
Alpha-MoE
Model=Mixtral, S=32
2026.04
658
-
RaMP
Model=Mixtral, S=512
2026.04
1,529
-
Alpha-MoE
Model=Mixtral, S=512
2026.04
2,665
-
RaMP
Model=OLMoE
2026.04
-
1.13
RaMP
Model=Qwen3
2026.04
-
1.09
RaMP
Model=DSv3
2026.04
-
1.2
RaMP
Model=Mixtral
2026.04
-
1.45
Feedback
Search any
task
Search any
task
