Hi @comaniac ,
Thank you for your prompt reply. I have updated the question a little bit so that things are more clear. Basically I use the same program and comment/uncomment one line, which is the following: `task[0] = autotvm.task.create(task[0].name, task[0].args, task[0].target, task[0].target_host,'winograd')` The log file matches the execution time when 'winograd' is set, however, for the 'direct' approach, the configuration does not seem to be applied, and the auto-tuning performance matches the fallback one (even though the log shows that there are configurations producing better results). As a sanity check, I will run the code using more iterations as you suggested, and I will let you know if the results change. Thanks again! --- [Visit Topic](https://discuss.tvm.ai/t/relay-conv2d-layer-performance-after-auto-tuning-same-as-fallback/6888/3) 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/68dd7b64bc64a53340d9c1b1f8c654b210599dae1333f2519acf1b2c84e43005).
