Have you seen anyone productively using TMA on Nvidia or async instructions on AMD? I’m currently looking at a 60% throughput degradation for 2D inputs on H100:

Have you seen anyone productively using TMA on Nvidia or async instructions on AMD? I’m currently looking at a 60% throughput degradation for 2D inputs on H100: https://github.com/ashvardanian/scaling-democracy/blob/a8092613fac1ae107e6a956c5c41ad4994a51735/scaling_democracy.cu#L266 https://github.com/ashvardanian/scaling-democracy/blob/a8092...