Linux下运行cuda SDK中的例子出错

[chenbo888666999]

我用的系统是fedora11，成功装好了驱动和cuda tool及cuda sdk，执行命令nvidia-xconfig -query-gpu-info结果如下：

[root@localhost release]# nvidia-xconfig -query-gpu-info
Number of GPUs: 1

GPU #0:
Name : GeForce 8400M GS
PCI BusID : PCI:1:0:0

Number of Display Devices: 1

Display Device 0 (DFP-0):
EDID Name : LPL
Minimum HorizSync : 49.323 kHz
Maximum HorizSync : 49.323 kHz
Minimum VertRefresh : 60 Hz
Maximum VertRefresh : 60 Hz
Maximum PixelClock : 69.300 MHz
Maximum Width : 1280 pixels
Maximum Height : 800 pixels
Preferred Width : 1280 pixels
Preferred Height : 800 pixels
Preferred VertRefresh : 60 Hz
Physical Width : 300 mm
Physical Height : 190 mm

也在/root/.bash_profile文件中设置好了环境变量：
export PATH=$PATH:/usr/local/cuda/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib

问题：但是我在 /root/NVIDIA_CUDA_SDK/projects/matrixMu下执行make时，出现很多错误，而且这些错误基本上都是一样的：

/usr/local/cuda/bin/../include/math_functions.h:71: 错误：内联函数‘double fabs(double)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:73: 错误：内联函数‘float fabsf(float)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:159: 错误：内联函数‘double floor(double)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:161: 错误：内联函数‘float floorf(float)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:249: 错误：内联函数‘long int lrint(double)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:251: 错误：内联函数‘long int lrintf(float)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:254: 错误：内联函数‘long long int llrint(double)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:256: 错误：内联函数‘long long int llrintf(float)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:264: 错误：内联函数‘double ceil(double)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:266: 错误：内联函数‘float ceilf(float)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:400: 错误：内联函数‘int __finite(double)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:404: 错误：内联函数‘int __signbit(double)’不能声明为有弱链接
/usr/local/cuda/bin/../include/math_functions.h:409: 错误：内联函数‘int __signbitf(float)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:179: 错误：内联函数‘double ceil(double)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:182: 错误：内联函数‘double fabs(double)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:185: 错误：内联函数‘double floor(double)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:196: 错误：内联函数‘int __finite(double)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:326: 错误：内联函数‘long int lrint(double)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:327: 错误：内联函数‘long long int llrint(double)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:350: 错误：内联函数‘int __signbit(double)’不能声明为有弱链接
/usr/include/bits/mathcalls.h:179: 错误：内联函数‘float ceilf(float)’不能声明为有弱链接


请问高手，这是什么原因？

[无心人_过过小日子]

下面是google到的:
Some flavors of gcc 4.1.0 and 4.1.1 seems to have trouble understanding
weak function definitions.Calls to function from the same file where it is
defined as weak _may_ get inlined, even when there is a non-weak definition of
the function elsewhere.

[无心人_过过小日子]

下面是google到的:
Some flavors of gcc 4.1.0 and 4.1.1 seems to have trouble understanding
weak function definitions.Calls to function from the same file where it is
defined as weak _may_ get inlined, even when there is a non-weak definition of
the function elsewhere.

[chenbo888666999]

非常感谢你的热心帮助，我现在的想法是暂时不换gcc的版本，哦，我现在的版本是gcc4.4.0.暂时不运行sdk自带的例子，而是自己学着写一些简单的cuda程序（不使用makefile文件），直接使用nvcc进行编译，看行不行。

[无心人_过过小日子]

建议:
1)不用emu模式编译.
2)所有GPU上的函数和启动内核的代码,写在一个cu文件中,不用多个cu文件.
也许可以避免weak问题.

[fallenjh]

纯顶

[kugooe]

dfsfsafafasfasf

[JONKO]

我也来顶顶，哈哈哈

[hankanling123]

你可以安装window7 吗？

[babycry627]

也来 顶顶

[zzmoutmans]

很好

[陌上花花]

学习学习

[a7558147]

顶个！！！！！！！！！！！

[suntinghu]

学习学习啦~~~~

[bbscts]

学习下下～！！我没怎么用过

[bbscts]

？？？？？？？？？？？？？？？？？？？？？？？？？？？？？

[GuanMingSoft]

Mark

[zuytok]

顶顶顶

[大雷子老师]

学习下下～！！我没用过

[观弈道童]

学习下下～！！我没用过

[qzcwx]

好贴，学习了！