Get Started
Home
Topics
Search
Library
Research questionHow can TPU kernels be generated and optimized without scarce hardware expertise?Custom accelerator kernels must satisfy functional requirements while exploiting hardware-specific performance opportunities. Effective tuning depends on compiler feedback, profiling, and specialized knowledge that may not be readily available.
AI
AI Agents
Code Generation & Program Synthesis
Evaluation & Benchmarks
Inference Optimization
Multi-agent Systems
Research Paper
Technology
Latest papersRecent research connected to this question, newest first.MaxKernel: Agentic Kernel Generation for TPUsThe evidence concerns MaxKernel’s human-guided and autonomous agent configurations for TPU kernel development. It covers 50 JaxBench tasks and workloads from open-source models, with reported comparisons to expert hand-tuned baselines; the input provides no evidence for other accelerator types.research paper · Sep 3, 2026
Related questions
How can LLM-generated GPU kernels remain performant across hardware, inputs, and programming models?How can CPU-oriented multiobjective evolutionary algorithms be tensorized without changing their optimization behavior?How can GPUs efficiently compute unnormalized Gaussian kernel sums with signed weights without custom kernels?How should GPU implementations be compared fairly when their optimization effort is asymmetric?