Evaluation time is too long #265
Replies: 2 comments 6 replies
-
|
We are working on it and will let you know when the solution is figured out. |
Beta Was this translation helpful? Give feedback.
-
|
More GPU resources (+300%) will be deployed when Phase 3 starts. Hope this can make the scheduling much faster. |
Beta Was this translation helpful? Give feedback.
-
|
We can finish the evaluation in about 3min on our machine. But it always meet timeout problem on the evaluation platform. Is it a common issue? |
Beta Was this translation helpful? Give feedback.
-
|
It seems that the code is not ran and just queued for 2 hours. |
Beta Was this translation helpful? Give feedback.
-
Yes, CPU resources had already been increased. The current main bottleneck is the GPU resource. |
Beta Was this translation helpful? Give feedback.
-
If you cost 3min in your computer, I think you will timeout on the server. you only test one file in your computer, But there are many file s one the server. I cost 36s in my computer, and i also timeout on server( I need CPU not GPU). Three days before, the same file haven't timeout. |
Beta Was this translation helpful? Give feedback.
-
However, I think that the test time is mainly consumed by constantly cycling through the test data, i.e. the bottleneck is the CPU. The reason is that there is only a small increase in time, not a multiplicative increase, using the five-fold model compared to using a single model. |
Beta Was this translation helpful? Give feedback.
-
Dear organizer：
As the competition reaches the third stage, everyone has entered the sprint stage, and the evaluation time is too long. It is completely impossible to submit three times a day. Are you considering adding servers?
Beta Was this translation helpful? Give feedback.
All reactions