the malloc is only called once during the first run and then the workspace is cached if possible. That is why in standard benchmark need to skip the first iteration(as we did in the time evaluator)
--- [Visit Topic](https://discuss.tvm.ai/t/malloc-occurring-during-run-may-affect-performance/2491/2) to respond. You are receiving this because you enabled mailing list mode. To unsubscribe from these emails, [click here](https://discuss.tvm.ai/email/unsubscribe/c7aa06873d0a467aa8b3a852a92582a56d7fc8e0a146c0ffa66a88cff799ae30). Tianqi Chen, UW, Seattle, WA, 98105, United States http://tracking.discuss.tvm.ai/tracking/unsubscribe?msgid=qTyNZTr2dvBC-OVYDlxBVQ2
