[Oisf-users] suricata-2.0rc3 'make' error with Nvidia K20/Tesla

[Victor Julien]

On 03/24/2014 07:17 PM, Shant Kassardjian wrote:
> Hello,
> 
> I'm having trouble compiling suricata-2.0rc3 with Nvidia K20 as I get
> the following 'make' error while compiling:
> 
> usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_10 -arch=sm_10 -ptx
> util-mpm-ac-cuda-kernel.cu
> util-mpm-ac-cuda-kernel.cu: In function `SCACCudaSearch32':
> util-mpm-ac-cuda-kernel.cu:79: warning: cast to pointer from integer of
> different size
> ./util-mpm-ac-cuda-kernel.cu(54): Warning: Cannot tell what pointer
> points to, assuming global memory space
> ./util-mpm-ac-cuda-kernel.cu(87): Warning: Cannot tell what pointer
> points to, assuming global memory space
> /usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_11 -arch=sm_11 -ptx
> util-mpm-ac-cuda-kernel.cu
> util-mpm-ac-cuda-kernel.cu: In function `SCACCudaSearch32':
> util-mpm-ac-cuda-kernel.cu:79: warning: cast to pointer from integer of
> different size
> ./util-mpm-ac-cuda-kernel.cu(54): Warning: Cannot tell what pointer
> points to, assuming global memory space
> ./util-mpm-ac-cuda-kernel.cu(87): Warning: Cannot tell what pointer
> points to, assuming global memory space
> /usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_12 -arch=sm_12 -ptx
> util-mpm-ac-cuda-kernel.cu
> util-mpm-ac-cuda-kernel.cu: In function `SCACCudaSearch32':
> util-mpm-ac-cuda-kernel.cu:79: warning: cast to pointer from integer of
> different size
> ./util-mpm-ac-cuda-kernel.cu(54): Warning: Cannot tell what pointer
> points to, assuming global memory space
> ./util-mpm-ac-cuda-kernel.cu(87): Warning: Cannot tell what pointer
> points to, assuming global memory space
> /usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_13 -arch=sm_13 -ptx
> util-mpm-ac-cuda-kernel.cu
> util-mpm-ac-cuda-kernel.cu: In function `SCACCudaSearch32':
> util-mpm-ac-cuda-kernel.cu:79: warning: cast to pointer from integer of
> different size
> ./util-mpm-ac-cuda-kernel.cu(54): Warning: Cannot tell what pointer
> points to, assuming global memory space
> ./util-mpm-ac-cuda-kernel.cu(87): Warning: Cannot tell what pointer
> points to, assuming global memory space
> /usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_20 -arch=sm_20 -ptx
> util-mpm-ac-cuda-kernel.cu
> /usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_21 -arch=sm_21 -ptx
> util-mpm-ac-cuda-kernel.cu
> /usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_30 -arch=sm_30 -ptx
> util-mpm-ac-cuda-kernel.cu
> /usr/bin/nvcc -O2 -o util-mpm-ac-cuda-kernel.ptx_sm_35 -arch=sm_35 -ptx
> util-mpm-ac-cuda-kernel.cu
> nvcc fatal   : Value 'sm_35' is not defined for option 'gpu-architecture'
> make[2]: *** [util-mpm-ac-cuda-kernel.ptx_sm_35] Error 255
> make[2]: Leaving directory `/home/skassardjian/suricata-2.0rc3/src'
> make[1]: *** [all-recursive] Error 1
> make[1]: Leaving directory `/home/skassardjian/suricata-2.0rc3'
> make: *** [all] Error 2
> 
> nvcc --version
> nvcc: NVIDIA (R) Cuda compiler driver
> Copyright (c) 2005-2012 NVIDIA Corporation
> Built on Thu_Apr__5_00:24:31_PDT_2012
> Cuda compilation tools, release 4.2, V0.2.1221
> 
> If I use suricata 1.4.7, I can compile cuda properly with no error
> however I'm missing ac-cuda algo.
> 
> 24/3/2014 -- 18:10:36 - <Info> - This is Suricata version 1.4.7 RELEASE
> 24/3/2014 -- 18:10:36 - <Info> - CPUs/cores online: 40
> 24/3/2014 -- 18:10:38 - <Info> - GPU Device 1: Tesla K20Xm, 14
> Multiprocessors, 732MHz, CUDA Compute Capability 3.5
> CUDA Cards recognized by the suricata CUDA module -
> |-----------------------------------------------------------------------------|
> | Device Id  |     Device Name      |   Multi-   | Clock Rate | Cuda
> Compute  |
> |            |                      | Processors |    (MHz)   |
> Capability    |
> |-----------------------------------------------------------------------------|
> | 0          | Tesla K20Xm          | 14         | 732        |
> 3.5           |
> |-----------------------------------------------------------------------------|
> 
> 
> I'm using Debian wheezy gcc4.7 with cuda toolkit 4.2.9-2 and
> nvidia-kernel-amd64 304.88.
> 
> Any help is appreciated :)

I think you'll have better luck with CUDA 5.x.

-- 
---------------------------------------------
Victor Julien
http://www.inliniac.net/
PGP: http://www.inliniac.net/victorjulien.asc
---------------------------------------------