ResearchTasksKernel-level Attention Speed and Memory AnalysisFor AI assistantsMarkdownContent indexKernel-level Attention Speed and Memory AnalysisFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedQwen3-8B model dimensions (H=32, Hk=8, d=128, GQA 4:1) on A100 GPU (test)DualKV27.1Forward Pass Time (ms)7May 18, 2026
Qwen3-8B model dimensions (H=32, Hk=8, d=128, GQA 4:1) on A100 GPU (test)DualKV27.1Forward Pass Time (ms)7May 18, 2026