Rapidly evolving GPU architectures featuring complex memory hierarchies, matrix units, and varied precision formats continue to widen the gap between theoretical peaks and achievable performance. We design and develop analytical performance models for NVIDIA Blackwell (B200) and AMD CDNA3 (MI300A) grounded in systematic microbenchmark characterization. For Blackwell, the model captures Tensor Memory (TMEM), asynchronous bulk copy (TMA), and 5th-generation tensor cores; for CDNA3, the model captures Infinity Cache hierarchy, VGPR constraints, and occupancy. Validation yields 1.31% MAE on B200 (21 kernels) and 0.09% on MI300A (27 kernels), while naive roofline baselines exceed 95% error on the same kernels. We further validate the models using Rodinia~3.1 and SPEChpc 2021 Tiny.The models are updated with HBM bandwidth, capacity, and cache parameters and applied to H200 (Hopper) and MI250X (CDNA2), indicating no major restructuring of the models are needed. All models and benchmarks will be released as open-source upon acceptance.

翻译：随着GPU架构的快速发展，其复杂的存储层次结构、矩阵运算单元以及多样化的精度格式不断拉大理论峰值与可达性能之间的差距。我们基于系统化的微基准测试特征分析，为NVIDIA Blackwell（B200）和AMD CDNA3（MI300A）架构设计并开发了分析性能模型。针对Blackwell架构，模型捕获了张量存储器（TMEM）、异步批量复制（TMA）以及第五代张量核心；针对CDNA3架构，模型捕获了Infinity Cache层级结构、VGPR约束以及占用率。验证结果显示：在B200上（21个内核）平均绝对误差（MAE）为1.31%，在MI300A上（27个内核）为0.09%，而朴素roofline基线模型在相同内核上的误差超过95%。我们进一步使用Rodinia 3.1和SPEChpc 2021 Tiny基准测试对模型进行了验证。模型通过更新HBM带宽、容量及缓存参数后，成功应用于H200（Hopper）和MI250X（CDNA2）架构，表明无需对模型进行重大重构。所有模型与基准测试将在论文被接收后以开源形式发布。