>This is not an apples-to-apples comparison. Neither is FLOPS. RDNA3 FLOPS numbers are greatly inflated, because they are only applicable to fairly specific VO

>This is not an apples-to-apples comparison. Neither is FLOPS. RDNA3 FLOPS numbers are greatly inflated, because they are only applicable to fairly specific VOPD and WMMA instructions, the former being for two MADs at the same time, and the latter being applicable in the same cases as tensor cores. Besides, it should be possible to use the tensor cores combined with the codec: you can use vector hardware to decode matrix values to fp16/fp32, then yeet those into the tensor cores. Although most of the time will probably be spent on the decoding part, assuming you're doing matrix-vector multiplication and not matrix-matrix (which might be different with MoE models like Mixtral 8x7B?)