betaFeedback

WorksComputer Science

CANNBot-CAKE: Collaborative Agentic Kernel Evolution for Ascend NPU

Abstract

Efficient custom operators are essential for modern deep learning workloads, yet developing high-performance kernels for Ascend NPUs remains slow, expert-intensive, and difficult to automate. Although AI agents now approach expert performance on GPU kernel benchmarks, current models remain less capable of generating AscendC kernels for NPUs, potentially reflecting the relative scarcity of relevant training data. We thus present CAKE, a CANN-native system for deployable custom operators that combines adaptive multi-DSL routing, separate Generation and Evolution modes, and reviewed knowledge reuse through CAKE Book and a Self-Evolving Agent. On 53 CANN-Bench v1.1 operators on Ascend 950PR, CAKE's aggregated results achieve 100%100\% case pass rate on the standard set (1,060/1,060) and 98.28%98.28\% on the hidden set (4,167/4,240), with an overall mean score of 78.63 out of 100 and an average performance speedup of 1.334×1.334\times over the baseline. Evolution then pushes correct operators further: EngramGateFusion cuts aggregate device time by 46.6%46.6\% over 20 rounds and roughly halves the mean roofline gap. These results offer a practical path toward closing Ascend's AI-for-kernel capability gap.

Paper details

vlnc:W_qbqv7hkm
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
arXiv nonexclusive license
Usage

views · 20

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…