未验证 提交 2f1487c2 编写于 作者: W weixing 提交者: GitHub

Merge pull request #10518 from weixing02/v2_deadlinks

V2 deadlinks
...@@ -5,7 +5,7 @@ ...@@ -5,7 +5,7 @@
充分展现英特尔平台的优势,有效提升PaddlePaddle在英特尔架构上的性能。 充分展现英特尔平台的优势,有效提升PaddlePaddle在英特尔架构上的性能。
<div align="center"> <div align="center">
<img src="image/overview.png"><br/> <img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/v2/images/overview.png"><br/>
Figure 1. PaddlePaddle on IA Figure 1. PaddlePaddle on IA
</div> </div>
...@@ -42,16 +42,43 @@ Figure 1. PaddlePaddle on IA ...@@ -42,16 +42,43 @@ Figure 1. PaddlePaddle on IA
MKL,MKLML以及MKL-DNN三者关系如下表: MKL,MKLML以及MKL-DNN三者关系如下表:
| Name | Open Source | License | Descriptions | <table>
| :---------- | :--------------- | :---------- | :------------ | <thead>
| MKL | No | Proprietary | Accelerate math processing routines | <tr>
| MKLML | No | Proprietary | Small package of MKL, especially for Machine Learning | <th>Name</th>
| MKL-DNN | Yes | Apache 2.0 | Accelerate primitives processing routines especially for Deep Neural Networks | <th>Open Source</th>
<th>License</th>
<th>Descriptions</th>
</tr>
</thead>
<tbody>
<tr>
<td>MKL</td>
<td>No</td>
<td>Proprietary</td>
<td>Accelerate math processing routines</td>
</tr>
<tr>
<td>MKLML</td>
<td>No</td>
<td>Proprietary</td>
<td>Small package of MKL, especially for Machine Learning</td>
</tr>
<tr>
<td>MKL-DNN</td>
<td>Yes</td>
<td>Apache 2.0</td>
<td>Accelerate primitives processing routines especially for Deep Neural Networks</td>
</tr>
</tbody>
</table>
MKLML可以与MKL-DNN共同使用,以此达到最好的性能。 MKLML可以与MKL-DNN共同使用,以此达到最好的性能。
<div align="center"> <div align="center">
<img src="image/engine.png"><br/> <img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/v2/images/engine.png"><br/>
Figure 2. PaddlePaddle with MKL Engines Figure 2. PaddlePaddle with MKL Engines
</div> </div>
...@@ -103,7 +130,7 @@ MKL-DNN的库目前只有动态库`libmkldnn.so`。 ...@@ -103,7 +130,7 @@ MKL-DNN的库目前只有动态库`libmkldnn.so`。
所以我们定义了一个`MKLDNNMatrix`用于管理MKL-DNN数据的不同格式以及相互之间的转换。 所以我们定义了一个`MKLDNNMatrix`用于管理MKL-DNN数据的不同格式以及相互之间的转换。
<div align="center"> <div align="center">
<img src="image/matrix.png"><br/> <img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/v2/images/matrix.png"><br/>
Figure 3. MKLDNNMatrix Figure 3. MKLDNNMatrix
</div> </div>
...@@ -113,7 +140,7 @@ Figure 3. MKLDNNMatrix ...@@ -113,7 +140,7 @@ Figure 3. MKLDNNMatrix
子类只需要使用定义好的接口,实现具体的函数功能即可。 子类只需要使用定义好的接口,实现具体的函数功能即可。
<div align="center"> <div align="center">
<img src="image/layers.png"><br/> <img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/v2/images/layers.png"><br/>
Figure 4. MKLDNNLayer Figure 4. MKLDNNLayer
</div> </div>
...@@ -150,7 +177,7 @@ Figure 4. MKLDNNLayer ...@@ -150,7 +177,7 @@ Figure 4. MKLDNNLayer
所以整体上,在实现每个子类的时候就不需要关心分支的事情了。 所以整体上,在实现每个子类的时候就不需要关心分支的事情了。
<div align="center"> <div align="center">
<img src="image/gradients.png"><br/> <img src="https://raw.githubusercontent.com/PaddlePaddle/Paddle/develop/doc/v2/images/gradients.png"><br/>
Figure 5. Merge Gradients Figure 5. Merge Gradients
</div> </div>
......
digraph G{
subgraph cluster_timestep0 {
label="recurrent timestep i-1"
bgcolor=lightgray
node [style=filled,color=white]
fc0_0 [label="fc 0"]
fc0_1 [label="fc 1"]
fc0_2 [label="fc 2"]
fc0_0 -> fc0_1
fc0_1 -> fc0_2
}
subgraph cluster_timestep1 {
label="recurrent timestep i"
node [style=filled];
fc1_0 [label="fc 0"]
fc1_1 [label="fc 1"]
fc1_2 [label="fc 2"]
color=blue
fc1_0 -> fc1_1
fc1_1 -> fc1_2
}
subgraph cluster_timestep2 {
label="recurrent timestep i+1"
bgcolor=lightgray
node [style=filled,color=white]
fc2_0 [label="fc 0"]
fc2_1 [label="fc 1"]
fc2_2 [label="fc 2"]
fc2_0 -> fc2_1
fc2_1 -> fc2_2
}
fc0_1 -> fc1_1 [style="dotted" constraint=false]
fc1_1 -> fc2_1 [style="dotted" constraint=false]
}
\ No newline at end of file
digraph G{
subgraph cluster_timestep0 {
label="recurrent timestep i-1"
bgcolor=lightgray
node [style=filled,color=white]
fc0_0 [label="fc 0"]
fc0_1 [label="fc 1"]
fc0_2 [label="fc 2"]
m0 [label="memory"]
fc0_0 -> fc0_1
fc0_1 -> fc0_2
fc0_1 -> m0
m0 -> fc0_1
}
subgraph cluster_timestep1 {
label="recurrent timestep i"
node [style=filled];
fc1_0 [label="fc 0"]
fc1_1 [label="fc 1"]
fc1_2 [label="fc 2"]
m1 [label="memory"]
color=blue
fc1_0 -> fc1_1
fc1_1 -> fc1_2
fc1_1 -> m1
m1 -> fc1_1
}
subgraph cluster_timestep2 {
label="recurrent timestep i+1"
bgcolor=lightgray
node [style=filled,color=white]
fc2_0 [label="fc 0"]
fc2_1 [label="fc 1"]
fc2_2 [label="fc 2"]
m2 [label="memory"]
fc2_0 -> fc2_1
fc2_1 -> fc2_2
fc2_1 -> m2
m2 -> fc2_1
}
m0 -> m1 [style="dotted" constraint=false]
m1 -> m2 [style="dotted" constraint=false]
}
\ No newline at end of file
digraph G {
rankdir=LR;
subgraph cluster_t0 {
a [label="4"]
b [label="5"]
c [label="2"]
}
subgraph cluster_t1 {
d [label="0"]
e [label="9"]
}
subgraph cluster_t2 {
f [label="8"]
g [label="1"]
h [label="4"]
}
a -> b;
b -> c;
c -> d [constraint=false];
d -> e;
e -> f [constraint=false];
f -> g;
g -> h;
}
\ No newline at end of file
digraph G {
rankdir=LR;
a [label="4"]
b [label="5"]
c [label="2"]
d [label="0"]
e [label="9"]
f [label="8"]
g [label="1"]
h [label="4"]
a -> b;
b -> c;
c -> d;
d -> e;
e -> f;
f -> g;
g -> h;
}
\ No newline at end of file
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册