device: NVIDIA GB200
shape: experts=8, tokens/expert=4096, M=32768, K=8192, N=4096
dtype: torch.bfloat16
torch: 2.12.0a0+5aff3928d8.nv26.05
transformer_engine: 2.15.0+42b84005
nvidia-cudnn-frontend: 1.23.0
nvidia-cutlass-dsl: 4.4.1

summary:
activation   cudnn_ms      te_ms   te/cudnn
swiglu          1.544      1.912      1.24x
dswiglu         0.885      1.718      1.94x
srelu           0.740      1.041      1.41x
dsrelu          0.822      1.373      1.67x
geglu           1.575      1.901      1.21x
dgeglu          0.893      1.808      2.03x
