w*s
发帖数: 7227 | 1 what's your user experience ? | a****l
发帖数: 8211 | 2 再solid的系统也受不了傻瓜的瞎折腾.matlab自己其实就是一个壳,matlab调用的库和
你用c调用的是一样的.
【在 w*s 的大作中提到】
: what's your user experience ?
| w*s
发帖数: 7227 | 3 want to know if there're any issues, that's it.
【在 a****l 的大作中提到】
: 再solid的系统也受不了傻瓜的瞎折腾.matlab自己其实就是一个壳,matlab调用的库和
: 你用c调用的是一样的.
| a****l
发帖数: 8211 | 4 只要你的c程序是正确的，不应该有任何问题。为什么这么说？有matlab这个使用了多
年的大程序再那里，还不够证明吗？
【在 w*s 的大作中提到】
: want to know if there're any issues, that's it.
| k**********g
发帖数: 989 | 5 If you need to handle large data and your application (not OS) is 32-bit,
you will see memory fragmentation (in the form of premature "out-of-memory"
errors soon.)
When MATLAB calls a C extension, the C extension must return all its data (
states) as a MATLAB output variable. A C extension must not try to remember
its states via global variables because this can create hard-to-fix bugs.
Long-running and batch-processing applications involving MATLAB should
implement application checkpoints: http://en.wikipedia.org/wiki/Application_checkpointing The application should periodically write its intermediate result to disk, so that if the application is interrupted (such as from an "out-of-memory" error), the application can be restarted from the most recent checkpoint.
Memory used to back a MATLAB matrix variable must be allocated using MATLAB'
s memory allocator. Do not use the C/C++ standard library "new" or "malloc"
for matrix variables. Conversely, do not use MATLAB's memory allocator for
variables used only in your C code.
If you need to make robust applications (such as industrial-strength
applications), use process isolation. (Perform the work in a separate
running instance of MATLAB.)
Just 2 cents, might have errors. | w*s
发帖数: 7227 | 6 perfect, this is what i want to learn, thanks very much !
"
remember
MATLAB'
"
【在 k**********g 的大作中提到】
: If you need to handle large data and your application (not OS) is 32-bit,
: you will see memory fragmentation (in the form of premature "out-of-memory"
: errors soon.)
: When MATLAB calls a C extension, the C extension must return all its data (
: states) as a MATLAB output variable. A C extension must not try to remember
: its states via global variables because this can create hard-to-fix bugs.
: Long-running and batch-processing applications involving MATLAB should
: implement application checkpoints: http://en.wikipedia.org/wiki/Application_checkpointing The application should periodically write its intermediate result to disk, so that if the application is interrupted (such as from an "out-of-memory" error), the application can be restarted from the most recent checkpoint.
: Memory used to back a MATLAB matrix variable must be allocated using MATLAB'
: s memory allocator. Do not use the C/C++ standard library "new" or "malloc"
| w*s
发帖数: 7227 | 7 BTW, how's the performance of c++ code calling matlab library ?
i never used it so what to ask guru.
we have a hard requirement for performance for c++ real-time transactions,
so ...
also for statistical computation, any other good libs besides matlab ?
"
remember
MATLAB'
"
【在 k**********g 的大作中提到】
: If you need to handle large data and your application (not OS) is 32-bit,
: you will see memory fragmentation (in the form of premature "out-of-memory"
: errors soon.)
: When MATLAB calls a C extension, the C extension must return all its data (
: states) as a MATLAB output variable. A C extension must not try to remember
: its states via global variables because this can create hard-to-fix bugs.
: Long-running and batch-processing applications involving MATLAB should
: implement application checkpoints: http://en.wikipedia.org/wiki/Application_checkpointing The application should periodically write its intermediate result to disk, so that if the application is interrupted (such as from an "out-of-memory" error), the application can be restarted from the most recent checkpoint.
: Memory used to back a MATLAB matrix variable must be allocated using MATLAB'
: s memory allocator. Do not use the C/C++ standard library "new" or "malloc"
|
|
