onnx-preview1 #844
Replies: 3 comments 1 reply
-
|
之前测试的时候，TRT构建引擎的时候，这可以明显缩短时间 对紧凑架构似乎优化没有太大效率影响，我在2080上观察不到差别，可能需要弱点的gpu 关掉这些额外优化好像也可以 |
Beta Was this translation helpful? Give feedback.
-
|
我测试的时候 3 级和 5 级优化有明显区别。现在是 5 级优化，以后会提供选项来修改优化等级。我认为对于 TensorRT，性能应放在第一位，构建时间长点也是可以接受的，而且引擎会被缓存，一个模型通常只需要构建一次。 以后会在构建引擎/编译着色器时展示一个 UI，防止用户以为缩放失败了。 |
Beta Was this translation helpful? Give feedback.
-
|
I am using the onnx build for real-time antialiasing. Some models I have created only allow for static onnx builds instead of the dynamic required for magpie. These models have a better image quality with acceptable speed. Would it be possible to add support for static models in the future? |
Beta Was this translation helpful? Give feedback.
-
This discussion was created from the release onnx-preview1.
Beta Was this translation helpful? Give feedback.
All reactions