定点包测试时出现错误消息

嗨，
看看FPU是如何工作的。
首先，真实类型的datea变为非规范化为某个定点表示。
然后执行计算。
最后，固定点结果被标准化，这意味着转换回实数类型的数字。
现在，谁说中间的操作需要一些简单的+  -  * /？
为什么不是复杂的算法？
Xilinx浮点运算符IP核可以执行所需的转换。
因此，您可以将其中一些放在算法的前后，然后使用仅使用定点类型的VHDL设计算法本身。
您可能面临的下一个问题是math_complex库只支持实际数据类型。
所以也许你必须寻找该软件包的定点版本。
如果没有这样的包，请写下来。
（我们都将不胜感激。） 
math_complex_fp.vhd
可能是它的名字。
这是一个起点：
http://code.google.com/p/open-vhdl/wiki/ComplexFixedPointLibrary
所以你不是第一个面临这个问题的人。
有一个很好的综合 
Eilert
附：
即使您打算仅将此代码用于仿真，请记住所有IP核也都有仿真模型。
所以，也有一个很好的模拟。
:-)
在原帖中查看解决方案



以下为原文

Hi,
look how a FPU works.
First the real type datea becomes denormalized to some fixed point representation.
Then the calculation is performed.
In the end the fixed point result becomes normalized, which means converted back into a real type number.
 
Now, who says that the operation in between needs to be some simple +-*/ ?
Why not a complex algorithm?
 
The Xilinx Floating Point Operator IP-core can perform the desired transformations.
So you can put some of these in front and behind of your algorithm, then design the algorithm itself with VHDL using only fixed point types.
 
The next problem you might face then is that the math_complex library just supports real data types.
So maybe you have to look for a fixed point version of that package.
And if there is no such package, write it. (We all would be grateful.)
   math_complex_fp.vhd
 could be the name of it.
And here's a starting point:
http://code.google.com/p/open-vhdl/wiki/ComplexFixedPointLibrary
 
 
So you are not the first one facing this problem.
 
Have a nice synthesis
  Eilert
 
P.S. And even if you are intending to use this code just for simulation, remember that all IP-cores have a simulation model too.
 
So, have a nice simulation too. :-)
View solution in original post

嗨，
是否为所有工具编译了包？
有一个很好的模拟 
Eilert



以下为原文

Hi,
has the package been compiled for all your tools?
 
Have a nice simulation
  Eilert
 

“为所有工具编译”是什么意思？
我只是使用ISim作为我的模拟器。
我是新手，你能告诉我怎么做吗？
我的项目在附件中，您不需要查看所有vhdl文件，只需要查看包文件即可。



以下为原文

What do you mean by "compiled for all your tools"? I just use ISim as my simulator. I am a newbie, can you tell me how to do it? My project is in the Attachment, you don't need to see all the vhdl files, you just need to see the package files.

嗨，
to_fixed是为signed和unsigned的输入类型声明的，但不是real类型。
a.im导致实际类型，因此它不适用于to_fixed函数。
有一个很好的模拟 
Eilert



以下为原文

Hi,
to_fixed is declared for input types of signed and unsigned, but not of type real.
a.im results in a real type, so it does not work with the to_fixed function.
 
 
 
Have a nice simulation
  Eilert

那么我们可以做些什么来将REAL类型数据转换为SFIXED类型数据？您知道真实类型数据可以通过截断和扩展转换为固定类型数据，但我们如何在VHDL中做？



以下为原文

Then what can we do to turn the REAL type data into SFIXED type data?You know the real type data CAN be turn into sfixed type data by truncation and extension, but how can we do in VHDL?

weilings写道：
那么我们可以做什么来将REAL类型数据转换为SFIXED类型数据？您知道真实类型数据可以通过截断和扩展转换为固定类型数据。
需要编写和实现执行此操作的功能。
----------------------------是的，我这样做是为了谋生。



以下为原文

weilings wrote:
Then what can we do to turn the REAL type data into SFIXED type data?You know the real type data CAN be turn into sfixed type data by truncation and extension.

A function to do that needs to be written and implemented. 
----------------------------Yes, I do this for a living.

嗨，
看看FPU是如何工作的。
首先，真实类型的datea变为非规范化为某个定点表示。
然后执行计算。
最后，固定点结果被标准化，这意味着转换回实数类型的数字。
现在，谁说中间的操作需要一些简单的+  -  * /？
为什么不是复杂的算法？
Xilinx浮点运算符IP核可以执行所需的转换。
因此，您可以将其中一些放在算法的前后，然后使用仅使用定点类型的VHDL设计算法本身。
您可能面临的下一个问题是math_complex库只支持实际数据类型。
所以也许你必须寻找该软件包的定点版本。
如果没有这样的包，请写下来。
（我们都将不胜感激。） 
math_complex_fp.vhd
可能是它的名字。
这是一个起点：
http://code.google.com/p/open-vhdl/wiki/ComplexFixedPointLibrary
所以你不是第一个面临这个问题的人。
有一个很好的综合 
Eilert
附：
即使您打算仅将此代码用于仿真，请记住所有IP核也都有仿真模型。
所以，也有一个很好的模拟。
:-)



以下为原文

Hi,
look how a FPU works.
First the real type datea becomes denormalized to some fixed point representation.
Then the calculation is performed.
In the end the fixed point result becomes normalized, which means converted back into a real type number.
 
Now, who says that the operation in between needs to be some simple +-*/ ?
Why not a complex algorithm?
 
The Xilinx Floating Point Operator IP-core can perform the desired transformations.
So you can put some of these in front and behind of your algorithm, then design the algorithm itself with VHDL using only fixed point types.
 
The next problem you might face then is that the math_complex library just supports real data types.
So maybe you have to look for a fixed point version of that package.
And if there is no such package, write it. (We all would be grateful.)
   math_complex_fp.vhd
 could be the name of it.
And here's a starting point:
http://code.google.com/p/open-vhdl/wiki/ComplexFixedPointLibrary
 
 
So you are not the first one facing this problem.
 
Have a nice synthesis
  Eilert
 
P.S. And even if you are intending to use this code just for simulation, remember that all IP-cores have a simulation model too.
 
So, have a nice simulation too. :-)

非常感谢你



以下为原文

Thank you very much

你有这个功能吗？
你能跟我分享一下吗？
或者你知道在哪里可以找到这个功能吗？



以下为原文

Do you have this function? Can you share it with me? Or do you know where to find this function?