[lammps-users] CUDA on GPU - COMPILATION ERROR

Hi,

Problem compiling LAMMPS gpu library on TESLA card, using the LAMMPS-25Aug09 source code.

Red-Hat x86_64 kernel 2.6.18-128.el5
nvcc --version
nvcc: NVIDIA ® Cuda compiler driver
Copyright © 2005-2007 NVIDIA Corporation
Built on Wed_Dec__3_16:25:17_PST_2008
Cuda compilation tools, release 2.1, V0.2.1221

nvcc -I/usr/local/cuda/include -DUNIX -O3 -DDEBUG -Xptxas -v --use_fast_math -maxrregcount 128 -D_SINGLE_SINGLE -o pair_gpu_nbor.cu_o -c pair_gpu_nbor.cu
nvc_memory.h(98): error: identifier “cudaHostAlloc” is undefined
nvc_memory.h(98): error: identifier “cudaHostAllocWriteCombined” is undefined
detected during instantiation of “void NVC_Host::safe_alloc_w(size_t) [with numtyp=int]”
pair_gpu_nbor.cu(47): here
2 errors detected in the compilation of “/tmp/tmpxft_00005fa1_00000000-4_pair_gpu_nbor.cpp1.ii”.
make: *** [pair_gpu_nbor.cu_o] Error 255

Thank you.
Ritwik

Hi,

Problem compiling LAMMPS gpu library on TESLA card, using the
LAMMPS-25Aug09 source code.

Red-Hat x86_64 kernel 2.6.18-128.el5
nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2007 NVIDIA Corporation
Built on Wed_Dec__3_16:25:17_PST_2008
Cuda compilation tools, release 2.1, V0.2.1221

this is your problem. cudaHostAlloc doesn't exist
in CUDA 2.1. please update to CUDA 2.2 or 2.3.

axel.