Reproducible benchmarks for state-space, attention, and hybrid sequence models on NVIDIA GPUs. Measures latency, memory bandwidth, compute utilization, and energy per token to quantify the GPU-SSM mismatch.
deep-learning cuda pytorch attention-mechanism ssm mamba nvidia-gpu memory-bandwidth state-space-models hybrid-models flash-attention jamba ml-inference mamba-2 gpu-benchmarking roofline-analysis mamba-3 throughput-per-watt arithmetic-intensity hardware-algorithm-co-design
-
Updated
Oct 13, 2026 - Python