[Custom OP] Add contiguous api support to paddle::Tensor

[BeingGod]

PR Category

Operator Mechanism

PR Types

New features

Description

We find some limitation of paddle custom op. The target of this PR is to improve custom op capability. Specifically as following:

1. Add contiguous api support to paddle::Tensor

[paddle-bot[bot]]

你的PR提交成功，感谢你对开源项目的贡献! 请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。 Your PR has been submitted. Thanks for your contribution! Please wait for the result of CI firstly. See Paddle CI Manual for details.

[CLAassistant]

All committers have signed the CLA.

[BeingGod]

PTAL. Thx ! @HydrogenSulfate

[BeingGod]

@HydrogenSulfate This PR is ready. Can you help me merge it ?

[BeingGod]

@zyfncg This PR is ready. Can you help me merge it ?

[BeingGod]

To custom operators the input tensors are contiguous so we don't need it temporary.

[paddle-ci-bot[bot]]

Sorry to inform you that 4dc892c's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

[BeingGod]

To custom operators the input tensors are contiguous so we don't need it temporary.

In custom operator user may call paddle::experimental::transpose (stride kernel) it will cause tensor becomes non-contiguous. So it's necessary to implement is_contiguous and contiguous function to paddle::Tensor class.