Difference between revisions of "Seawulf Items Needed"

From Rizzo_Lab
Jump to: navigation, search
(→‎List of Items needed)
(→‎List of Items needed)
 
(15 intermediate revisions by 3 users not shown)
Line 1: Line 1:
 
== List of Items needed ==
 
== List of Items needed ==
−
*csh
+
*python 2.7 (create module)
−
*python 2.7
+
***ipython
−
**numpy
+
***pandas
−
**matplotlib
+
***cython
−
**scipy
+
***mpi4py
 +
***notebook
 +
***RDKit
 +
****We haven't been using this RDK Toolkit, but it was recommended/used by Michael Gilson and other group, and it seems to be useful and powerful, here's a link to the installation guide page http://www.rdkit.org/docs/Install.html
 
* X11 forwarding
 
* X11 forwarding
 +
*Increase GPU queue max walltime to 12hrs
 +
===AMBER CUDA builds===
 +
AMBER + '''CUDA7.5 or 8.0''' via GCC compilers: OK<br>
 +
module loaded
 +
{|
 +
<pre>1) torque/6.0.2  2) cuda75/toolkit/7.5.18</pre>
 +
|}
 +
AMBER + '''CUDA7.5 or 8.0''' via intel compilers: Fails<br>
 +
modules loaded
 +
{|
 +
<pre>
 +
  1) torque/6.0.2                    3) intel/mkl/64/2017/0.098        5) cuda75/toolkit/7.5.18
 +
  2) intel/compiler/64/2017/17.0.0  4) intel/mpi/64/2017/0.098 </pre>
 +
|}
 +
{|
 +
configure -cuda -noX11 --with-python (path to python) intel; fails with the following error
 +
|<pre>/cm/shared/apps/cuda75/toolkit/7.5.18/bin/nvcc  -c -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_53,code=sm_53  -o core_kernels.o core_kernels.cu                                                   
 +
In file included from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/driver_types.h:70:0,
 +
                from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/builtin_types.h:59,
 +
                from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/cuda_runtime.h:84,
 +
                from <command-line>:0:
 +
/cm/shared/apps/intel/compilers_and_libraries/2017.0.098/compiler/include/limits.h:36:54: error: missing binary operator before token "("defined(__has_include_next) && __has_include_next(<limits.h>)
 +
                                                     
 +
In file included from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/math_functions.h:8791:0,
 +
                from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/common_functions.h:224,
 +
                from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/cuda_runtime.h:112,
 +
                from <command-line>:0:
 +
/cm/shared/apps/intel/compilers_and_libraries/2017.0.098/compiler/include/math.h:108:3: error: #error "This Intel <math.h> is for use with only the Intel compilers!"
 +
# error "This Intel <math.h> is for use with only the Intel compilers!"</pre>
 +
|
 +
|}

Latest revision as of 13:36, 9 March 2017

List of Items needed

  • python 2.7 (create module)
      • ipython
      • pandas
      • cython
      • mpi4py
      • notebook
      • RDKit
        • We haven't been using this RDK Toolkit, but it was recommended/used by Michael Gilson and other group, and it seems to be useful and powerful, here's a link to the installation guide page http://www.rdkit.org/docs/Install.html
  • X11 forwarding
  • Increase GPU queue max walltime to 12hrs

AMBER CUDA builds

AMBER + CUDA7.5 or 8.0 via GCC compilers: OK
module loaded 

1) torque/6.0.2  2) cuda75/toolkit/7.5.18

AMBER + CUDA7.5 or 8.0 via intel compilers: Fails
modules loaded 

   1) torque/6.0.2                    3) intel/mkl/64/2017/0.098         5) cuda75/toolkit/7.5.18
  2) intel/compiler/64/2017/17.0.0   4) intel/mpi/64/2017/0.098
configure -cuda -noX11 --with-python (path to python) intel; fails with the following error 
/cm/shared/apps/cuda75/toolkit/7.5.18/bin/nvcc  -c -gencode arch=compute_20,code=sm_20 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_52,code=sm_52 -gencode arch=compute_53,code=sm_53  -o core_kernels.o core_kernels.cu                                                    
In file included from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/driver_types.h:70:0,
                 from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/builtin_types.h:59,
                 from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/cuda_runtime.h:84,
                 from <command-line>:0:
/cm/shared/apps/intel/compilers_and_libraries/2017.0.098/compiler/include/limits.h:36:54: error: missing binary operator before token "("defined(__has_include_next) && __has_include_next(<limits.h>)
                                                      
In file included from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/math_functions.h:8791:0,
                 from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/common_functions.h:224,
                 from /cm/shared/apps/cuda75/toolkit/7.5.18/bin/..//include/cuda_runtime.h:112,
                 from <command-line>:0:
/cm/shared/apps/intel/compilers_and_libraries/2017.0.098/compiler/include/math.h:108:3: error: #error "This Intel <math.h> is for use with only the Intel compilers!"
 # error "This Intel <math.h> is for use with only the Intel compilers!"
Retrieved from "https://ringo.ams.stonybrook.edu/index.php?title=Seawulf_Items_Needed&oldid=10285"