framework,version,device,op_name,kernel_source,quant_type,num_token,hidden_size,intermediate_size,avg_ms
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,1,7168,2048,0.04410240054130554
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,2,7168,2048,0.04375360012054443
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,4,7168,2048,0.04368639886379242
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,8,7168,2048,0.04589119851589203
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,16,7168,2048,0.046000000834465024
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,32,7168,2048,0.043721601366996765
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,64,7168,2048,0.045542401075363156
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,128,7168,2048,0.04577920138835907
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,256,7168,2048,0.050595200061798094
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,512,7168,2048,0.05815039873123169
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,1024,7168,2048,0.0843999981880188
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,2048,7168,2048,0.12445759773254395
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,4096,7168,2048,0.20109760761260986
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,8192,7168,2048,0.36347839832305906
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,16384,7168,2048,0.6743360042572022
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,32768,7168,2048,1.2969696044921875
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,65536,7168,2048,2.5338687896728516
SGLang,0.5.8.post1,NVIDIA GB200,mlp,deepseek_v3,fp8_block,131072,7168,2048,5.014492797851562
