Kadle11

Setting up ParMETIS and Python3 Support

1. Install CMake and MPIC++
2. Download and extract the ParMETIS shared object

wget http://glaros.dtc.umn.edu/gkhome/fetch/sw/parmetis/parmetis-4.0.3.tar.gz 
gunzip parmetis-4.0.3.tar.gz 
tar -x parmetis-4.0.3.tar
cd parmetis-4.0.3 

Kadle11

Generate /tmp/perf-<PID>.map files for OpenResty-Nginx Docker Container

# Lua JIT generates /tmp/perf-<PID>.map files that perf needs to resolve symbols
# We need to map PIDs from Container to Host and copy the corrosponding files so 
# perf can resolve the symbols

# Compile LuaJIT in DockerFile with --with-luajit-xcflags="-DLUAJIT_USE_PERFTOOLS"

#!/bin/bash

Kadle11

Ubuntu Setup Notes for Dell 7559 with GTX 960M

scott@ksf16u:~$ lspci | grep -i vga
00:02.0 VGA compatible controller: Intel Corporation Skylake Integrated Graphics (rev 06)
scott@ksf16u:~$ lspci | grep -i 3D
02:00.0 3D controller: NVIDIA Corporation GM107M [GeForce GTX 960M] (rev a2)

Objective is to allow NVidia GTX 960M to work as accelerated GPU with Tensorflow 1.8 (current as of 6/10/18). This supports Eager execution for dynamic graphs.

Kadle11

Tools to analyze the Performance

gProf

The GNU gProf profiler can be used to get a detailed profile of the program. It gives a an easy to read output which shows how much time is being spent in each function.

Usage

gcc -pg TestProf.c -o TestGprof.out
./TestGprof.out