name,decode_1_H100,decode_1_MI300x,abs_diff Attention,6.854378547570919,7.540308478018733,0.6859299304478137 Llama3RotaryEmbedding,1.81774562552658,2.156621209163042,0.33887558363646186 LogitsProcessor,5.075146841646092,4.6691502316885956,-0.4059966099574961 "MergedColumnParallelLinear(weight=bfloat16[28672, 4096])",38.652132986747134,35.79513292422048,-2.8570000625266516 "QKVParallelLinear(weight=bfloat16[6144, 4096])",9.785950245233131,9.448245445236797,-0.3377047999963345 RMSNorm(weight=bfloat16[4096]) <- LlamaDecoderLayer,3.0304211602626046,5.318987930281327,2.288566770018723 RMSNorm(weight=bfloat16[4096]) <- LlamaForCausalLM,0.0468820505350108,, "RowParallelLinear(weight=bfloat16[4096, 14336])",20.407072199090585,19.164716615141298,-1.242355583949287 "RowParallelLinear(weight=bfloat16[4096, 4096])",8.50892363706815,7.933789312968523,-0.5751343240996274 Sampler,1.7628618245720273,3.2239583845426703,1.461096559970643 SiluAndMul,4.016277846567755,4.590159098921689,0.5738812523539343 others,0.042207035180003474,0.07876288543950605,0.03655585025950257 vocab_embed_ops,,0.08016748437734832,