I wonder if you get this error for the same reason the GNU compilers don't
work. If my suspicions are correct, then the Intel compilers don't do all
of their own legwork -- they build on some of the machinery provided by the
GNU compilers (glibc, maybe?). If this is the case, then it's entirely
possible that the Intel compilers have inherited the limitations of your
GNU compilers.
My suggestion (depending on how important having pmemd.cuda is), install an
older GCC. Or, switch OSes to one that doesn't quite live on the
bleeding-edge (with the end goal of getting an older default GCC that is
nvcc-compatible). On my end, I have GCC versions 4.5 (among older and
newer versions), and the Intel compilers, both of which build pmemd.cuda
without issue.
(I've had success with several versions of the Intel compilers versions
11.1.069 and 12.X).
Good luck,
Jason
On Thu, Dec 6, 2012 at 10:37 AM, Thomas Evangelidis <tevang3.gmail.com>wrote:
> I've done that already and I also get an error with serial cuda
> compilation. Serial compilation and MPI work well, but when I add the -cuda
> flag I get errors like the following. Do you have any idea what may be
> wrong?
>
>
> export AMBERHOME=/home/thomas/Programs/amber12
> export PATH=$AMBERHOME/exe:$PATH
> export LD_LIBRARY_PATH=$AMBERHOME/lib:$LD_LIBRARY_PATH
> source /home/thomas/Programs/Intel_Compilers/bin/compilervars.sh intel64
> export MPI_HOME=/usr/local/mpich-1.4.1p1_icc
> export MKL_HOME=/home/thomas/Programs/Intel_Compilers/mkl
> export CUDA_HOME=/usr/local/cuda
> export SSE_TYPES=SSE4.1,SSE4.2
> ./configure -cuda intel
> make install
> ...
> ...
> /usr/local/cuda/bin/nvcc -use_fast_math -gencode arch=compute_13,code=sm_13
> -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30
> -DCUDA -Duse_SPFP -I/usr/local/cuda/include -IB40C -IB40C/KernelCommon -c
> kForcesUpdate.cu
> ./kForcesUpdate.cu(87): Advisory: Loop was not unrolled, cannot deduce loop
> trip count
> ./kForcesUpdate.cu(127): Advisory: Loop was not unrolled, cannot deduce
> loop trip count
> /usr/local/cuda/bin/nvcc -use_fast_math -gencode arch=compute_13,code=sm_13
> -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30
> -DCUDA -Duse_SPFP -I/usr/local/cuda/include -IB40C -IB40C/KernelCommon -c
> kCalculateLocalForces.cu
> /usr/local/cuda/bin/nvcc -use_fast_math -gencode arch=compute_13,code=sm_13
> -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30
> -DCUDA -Duse_SPFP -I/usr/local/cuda/include -IB40C -IB40C/KernelCommon -c
> kCalculateGBBornRadii.cu
> /usr/local/cuda/bin/nvcc -use_fast_math -gencode arch=compute_13,code=sm_13
> -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30
> -DCUDA -Duse_SPFP -I/usr/local/cuda/include -IB40C -IB40C/KernelCommon -c
> kCalculatePMENonbondEnergy.cu
> /usr/local/cuda/bin/nvcc -use_fast_math -gencode arch=compute_13,code=sm_13
> -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30
> -DCUDA -Duse_SPFP -I/usr/local/cuda/include -IB40C -IB40C/KernelCommon -c
> kCalculateGBNonbondEnergy1.cu
> /usr/local/cuda/bin/nvcc -use_fast_math -gencode arch=compute_13,code=sm_13
> -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30
> -DCUDA -Duse_SPFP -I/usr/local/cuda/include -IB40C -IB40C/KernelCommon -c
> kNLRadixSort.cu
> B40C/radixsort_single_grid.cu: In instantiation of ‘cudaError_t
> b40c::SingleGridRadixSortingEnactor<K,
> V>::EnactSort(b40c::MultiCtaRadixSortStorage<K, V>&) [with int
> LOWER_KEY_BITS = 12; K = unsigned int; V = unsigned int; cudaError_t =
> cudaError]’:
> kNLRadixSort.cu:39:48: required from here
> B40C/radixsort_single_grid.cu:282:1: error: ‘GetWorkDecomposition’ was not
> declared in this scope, and no declarations were found by
> argument-dependent lookup at the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:282:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:282:1: note: use
> ‘this->GetWorkDecomposition’
> instead
> B40C/radixsort_single_grid.cu:291:1: error: ‘PreSort’ was not declared in
> this scope, and no declarations were found by argument-dependent lookup at
> the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:291:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:291:1: note: use ‘this->PreSort’ instead
> B40C/radixsort_single_grid.cu: In instantiation of ‘cudaError_t
> b40c::SingleGridRadixSortingEnactor<K,
> V>::EnactSort(b40c::MultiCtaRadixSortStorage<K, V>&) [with int
> LOWER_KEY_BITS = 16; K = unsigned int; V = unsigned int; cudaError_t =
> cudaError]’:
> kNLRadixSort.cu:46:48: required from here
> B40C/radixsort_single_grid.cu:282:1: error: ‘GetWorkDecomposition’ was not
> declared in this scope, and no declarations were found by
> argument-dependent lookup at the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:282:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:282:1: note: use
> ‘this->GetWorkDecomposition’
> instead
> B40C/radixsort_single_grid.cu:291:1: error: ‘PreSort’ was not declared in
> this scope, and no declarations were found by argument-dependent lookup at
> the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:291:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:291:1: note: use ‘this->PreSort’ instead
> B40C/radixsort_single_grid.cu: In instantiation of ‘cudaError_t
> b40c::SingleGridRadixSortingEnactor<K,
> V>::EnactSort(b40c::MultiCtaRadixSortStorage<K, V>&) [with int
> LOWER_KEY_BITS = 20; K = unsigned int; V = unsigned int; cudaError_t =
> cudaError]’:
> kNLRadixSort.cu:53:48: required from here
> B40C/radixsort_single_grid.cu:282:1: error: ‘GetWorkDecomposition’ was not
> declared in this scope, and no declarations were found by
> argument-dependent lookup at the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:282:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:282:1: note: use
> ‘this->GetWorkDecomposition’
> instead
> B40C/radixsort_single_grid.cu:291:1: error: ‘PreSort’ was not declared in
> this scope, and no declarations were found by argument-dependent lookup at
> the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:291:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:291:1: note: use ‘this->PreSort’ instead
> B40C/radixsort_single_grid.cu: In instantiation of ‘cudaError_t
> b40c::SingleGridRadixSortingEnactor<K,
> V>::EnactSort(b40c::MultiCtaRadixSortStorage<K, V>&) [with int
> LOWER_KEY_BITS = 24; K = unsigned int; V = unsigned int; cudaError_t =
> cudaError]’:
> kNLRadixSort.cu:60:48: required from here
> B40C/radixsort_single_grid.cu:282:1: error: ‘GetWorkDecomposition’ was not
> declared in this scope, and no declarations were found by
> argument-dependent lookup at the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:282:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:282:1: note: use
> ‘this->GetWorkDecomposition’
> instead
> B40C/radixsort_single_grid.cu:291:1: error: ‘PreSort’ was not declared in
> this scope, and no declarations were found by argument-dependent lookup at
> the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:291:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:291:1: note: use ‘this->PreSort’ instead
> B40C/radixsort_single_grid.cu: In instantiation of ‘cudaError_t
> b40c::SingleGridRadixSortingEnactor<K,
> V>::EnactSort(b40c::MultiCtaRadixSortStorage<K, V>&) [with int
> LOWER_KEY_BITS = 28; K = unsigned int; V = unsigned int; cudaError_t =
> cudaError]’:
> kNLRadixSort.cu:67:48: required from here
> B40C/radixsort_single_grid.cu:282:1: error: ‘GetWorkDecomposition’ was not
> declared in this scope, and no declarations were found by
> argument-dependent lookup at the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:282:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:282:1: note: use
> ‘this->GetWorkDecomposition’
> instead
> B40C/radixsort_single_grid.cu:291:1: error: ‘PreSort’ was not declared in
> this scope, and no declarations were found by argument-dependent lookup at
> the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:291:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:291:1: note: use ‘this->PreSort’ instead
> B40C/radixsort_single_grid.cu: In instantiation of ‘cudaError_t
> b40c::SingleGridRadixSortingEnactor<K,
> V>::EnactSort(b40c::MultiCtaRadixSortStorage<K, V>&) [with int
> LOWER_KEY_BITS = 32; K = unsigned int; V = unsigned int; cudaError_t =
> cudaError]’:
> kNLRadixSort.cu:74:48: required from here
> B40C/radixsort_single_grid.cu:282:1: error: ‘GetWorkDecomposition’ was not
> declared in this scope, and no declarations were found by
> argument-dependent lookup at the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:282:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:282:1: note: use
> ‘this->GetWorkDecomposition’
> instead
> B40C/radixsort_single_grid.cu:291:1: error: ‘PreSort’ was not declared in
> this scope, and no declarations were found by argument-dependent lookup at
> the point of instantiation [-fpermissive]
> B40C/radixsort_single_grid.cu:291:1: note: declarations in dependent base
> ‘b40c::MultiCtaRadixSortingEnactor<unsigned int, unsigned int,
> b40c::MultiCtaRadixSortStorage<unsigned int, unsigned int> >’ are not found
> by unqualified lookup
> B40C/radixsort_single_grid.cu:291:1: note: use ‘this->PreSort’ instead
>
>
>
>
> On 6 December 2012 02:30, Jason Swails <jason.swails.gmail.com> wrote:
>
> >
> >
> > On Dec 5, 2012, at 6:18 PM, Thomas Evangelidis <tevang3.gmail.com>
> wrote:
> >
> > > I am using CUDA 4.2 but changed the
> /usr/local/cuda/include/host_config.h
> > > to not complain with my gcc compiler version. I have posted an older
> > > message about intel compilers failing to install pmemd.cuda, and Ross
> > > recommended the GNU compilers as they are more stable with CUDA.
> >
> > My guess is there is a reason that NVIDIA disabled nvcc with newer GCC
> > versions, and the errors you are seeing are a result of that
> > incompatibility.
> >
> > If you do not have an older GCC version to use, I suggest that you go
> back
> > to the intel compilers. The biggest thing to worry about with those
> > compilers is to make sure that you set up your environment correctly with
> > them. Intel provides sourceable scripts for both bash and csh to do this
> > for you.
> >
> > Good luck.
> >
> > Jason
> >
> > --
> > Jason M. Swails
> > Quantum Theory Project,
> > University of Florida
> > Ph.D. Candidate
> > 352-392-4032
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
>
>
>
> --
>
> ======================================================================
>
> Thomas Evangelidis
>
> PhD student
> University of Athens
> Faculty of Pharmacy
> Department of Pharmaceutical Chemistry
> Panepistimioupoli-Zografou
> 157 71 Athens
> GREECE
>
> email: tevang.pharm.uoa.gr
>
> tevang3.gmail.com
>
>
> website: https://sites.google.com/site/thomasevangelidishomepage/
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Candidate
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Dec 06 2012 - 09:00:02 PST
