linux - nvidia cuda 和驱动程序版本不足

Question

我正在尝试使用 CUDA 技术，但遇到了一些问题

greymachine ~/NVIDIA_CUDA-5.0_Samples/1_Utilities/deviceQuery $ ./deviceQuery 
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

cudaGetDeviceCount returned 35
-> CUDA driver version is insufficient for CUDA runtime version

greymachine ~/NVIDIA_CUDA-5.0_Samples/1_Utilities/deviceQuery $

那是我的问题。

我的配置:

$ nvidia-settings -q NvidiaDriverVersion
  Attribute 'NvidiaDriverVersion' (greymachine.localdomain:0.0): 310.19

$ uname -r
3.7.1-un-def-alt2.1

$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2012 NVIDIA Corporation
Built on Fri_Sep_21_17:28:58_PDT_2012
Cuda compilation tools, release 5.0, V0.2.1221 





$ ./deviceQueryDrv 
./deviceQueryDrv Starting...

CUDA Device Query (Driver API) statically linked version 
Detected 1 CUDA Capable device(s)

Device 0: "GeForce GT 430"
  CUDA Driver Version:                           5.0
  CUDA Capability Major/Minor version number:    2.1
  Total amount of global memory:                 1024 MBytes (1073283072 bytes)
  ( 2) Multiprocessors x ( 48) CUDA Cores/MP:    96 CUDA Cores
  GPU Clock rate:                                1400 MHz (1.40 GHz)
  Memory Clock rate:                             800 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 131072 bytes
  Max Texture Dimension Sizes                    1D=(65536) 2D=(65536,65535) 3D=(2048,2048,2048)
  Max Layered Texture Size (dim) x layers        1D=(16384) x 2048, 2D=(16384,16384) x 2048
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 32768
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  1536
  Maximum number of threads per block:           1024
  Maximum sizes of each dimension of a block:    1024 x 1024 x 64
  Maximum sizes of each dimension of a grid:     65535 x 65535 x 65535
  Texture alignment:                             512 bytes
  Maximum memory pitch:                          2147483647 bytes
  Concurrent copy and kernel execution:          Yes with 1 copy engine(s)
  Run time limit on kernels:                     Yes
  Integrated GPU sharing Host Memory:            No
  Support host page-locked memory mapping:       Yes
  Concurrent kernel execution:                   Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support:                        Disabled
  Device supports Unified Addressing (UVA):      Yes
  Device PCI Bus ID / PCI location ID:           1 / 0
  Compute Mode:
     < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) > 





$ lsmod | grep nvidia
nvidia               9381500  39 
i2c_core               30993  3 i2c_i801,nvidia,videodev


dmesg:
[   28.548939] NVRM: loading NVIDIA UNIX x86_64 Kernel Module  310.19  Thu Nov  8 00:52:03 PST 2012
[   29.065356] NVRM: GPU at 0000:01:00: GPU-5a5ce500-f7fd-ab9d-64d7-cc0d1fe26ff1
[   29.065360] NVRM: Your system is not currently configured to drive a VGA console
[   29.065361] NVRM: on the primary VGA device. The NVIDIA Linux graphics driver
[   29.065362] NVRM: requires the use of a text-mode VGA console. Use of other console
[   29.065363] NVRM: drivers including, but not limited to, vesafb, may result in
[   29.065364] NVRM: corruption and stability problems, and is not supported.
 1682.331776] NVRM: GPU at 0000:01:00: GPU-5a5ce500-f7fd-ab9d-64d7-cc0d1fe26ff1

$ ldd ./deviceQuery 
        linux-vdso.so.1 (0x00007fffd4dff000)
        libcudart.so.5.0 => /usr/local/cuda-5.0/lib64/libcudart.so.5.0 (0x00007f5e90c26000)
        libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x00007f5e90922000)
        libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f5e9070c000)
        libc.so.6 => /lib64/libc.so.6 (0x00007f5e90362000)
        libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f5e90145000)
        libdl.so.2 => /lib64/libdl.so.2 (0x00007f5e8ff40000)
        librt.so.1 => /lib64/librt.so.1 (0x00007f5e8fd38000)
        libm.so.6 => /lib64/libm.so.6 (0x00007f5e8fa3e000)
        /lib64/ld-linux-x86-64.so.2 (0x00007f5e90eaa000) 

我已经安装了从 nvidia 网站下载的 cuda 工具包，我已经从我的发行版 (Alt Linux) 预编译了驱动程序，但是 libcuda.so 没有附带它们，所以我从原始的 nvidia 驱动程序复制了那个库。编译没问题。我还用 304.51 测试了 2.6.32 内核:得到相同的消息，但这是可以理解的，cuda 带有 304.54 驱动程序。

AFAIU，如果我有比 cuda 工具包附带的更新的驱动程序，那没关系。但正如您所见，有些事情是错误的。

那么，内核驱动程序是否可以比原始驱动程序更新(即我用 cuda 获得的驱动程序)也许我应该自己编译模块？但是为什么，为什么？我的发行版模块运行良好。

谢谢

Answer 1

这似乎是由严重损坏的 CUDA 安装引起的。通常的解决方案是卸载所有内容，安装受支持的主机工具链，然后重新安装驱动程序和工具包。每个工具包都包含详细的安装说明和系统要求。如果您阅读并遵循这些内容，则很少会出现此类问题。

[此答案根据评论汇总并添加为社区 wiki 条目以将其从未回答的问题列表中移除]