Add TensorExpression

[hedaoyuan]

A new TensorExpression implementation. And there is a documentation , to assist in code review. This document is not complete, at present, do not consider merger into the master.

[byzhang]

I LOVE this PR.

[reyoung]

添加文档

[gangliao]

change to WITH_GPU

[byzhang]

Just FYI, https://github.com/ddemidov/vexcl

Thanks,
-B

On Sun, Oct 30, 2016 at 8:51 PM, gangliao [email protected] wrote:

@gangliao commented on this pull request.

In CMakeLists.txt
#239 (review):

@@ -117,6 +125,9 @@ endif(NOT WITH_TIMER)
if(WITH_AVX)
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${AVX_FLAGS}")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${AVX_FLAGS}")

• if(AVX_FOUND)

change to WITH_GPU

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#239 (review), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AA1O1X2rkkpss7Yd78Glo3K3fzDTvqtUks5q5WWogaJpZM4Ke62g
.

[reyoung]

抱歉，之前看的时候没注意到有文档。

[backyes]

@hedaoyuan @reyoung 这个PR先不要merge吧，为了确保下一个milestone的release能比较stable，这个PR升级比较多，可以考虑放到下下个milestone，你们觉得呢？
另外， 因为GPU编译未使能的avx的bug https://github.com/baidu/Paddle/issues/282， 是否可以单独fix？ @hedaoyuan

[reyoung]

@hedaoyuan says This document is not complete, at present, do not consider merger into the master.

[hedaoyuan]

Merge最新的develop代码，并且对tests中的一些conflicts进行修改。

[reyoung]

LGTM

[reyoung]

虽然不是特别严重的事情，不过 operator =似乎应该返回自身的引用。这样才能写a=b=c。不过我们也可以说，不能这么写code。