lijiansong

Build TF 2.0 from source

Step 1:

从anaconda的base环境中创建新的环境：

$ conda create -n tf2-source python=3.6
$ conda activate tf2-source

lijiansong

"""TensorFlow 2.0 implementation of vanilla Autoencoder."""
import numpy as np
import tensorflow as tf

__author__ = "Abien Fred Agarap"

np.random.seed(1)
tf.random.set_seed(1)
batch_size = 128
epochs = 10

lijiansong

import math
import datetime

from tqdm import tqdm

import pandas as pd
import numpy as np

import tensorflow as tf

lijiansong

lijiansong

The Computer Language Benchmarks Game - Mandelbrot

Source code: https://github.com/mrange/benchmarksgame/tree/master/src/mandelbrot

1. Update 2017-06-25 - Decided I could do a bit better with F# so I added an improved F# program that uses the .NET SSE
2. Update 2017-07-01 - Reduced the overhead of bitmap allocation saving 9ms for 16000x16000 bitmaps
3. Update 2017-07-06 - Improved the fast F# program by removing overy redundancy

Recently I discovered The Computer Language Benchmarks Game which intrigued me, especially the mandelbrot version.

lijiansong

--- lua-5.1.4/Makefile  2008-08-12 00:40:48.000000000 +0000
+++ lua-5.1.4-shared-lib/Makefile       2009-01-13 22:01:33.000000000 +0000
@@ -43,7 +43,7 @@
 # What to install.
 TO_BIN= lua luac
 TO_INC= lua.h luaconf.h lualib.h lauxlib.h ../etc/lua.hpp
-TO_LIB= liblua.a
+TO_LIB= liblua.a liblua.so
 TO_MAN= lua.1 luac.1

lijiansong

--- src/Makefile.original2015-02-27 10:53:53.130125907 -0500
+++ src/Makefile 2015-02-27 10:53:24.674126798 -0500
@@ -7,7 +7,7 @@
 PLAT= none

 CC= gcc -std=gnu99
-CFLAGS= -O2 -Wall -Wextra -DLUA_COMPAT_5_2 $(SYSCFLAGS) $(MYCFLAGS)
+CFLAGS= -O2 -Wall -Wextra -DLUA_COMPAT_5_2 $(SYSCFLAGS) $(MYCFLAGS) -fPIC
 LDFLAGS= $(SYSLDFLAGS) $(MYLDFLAGS)
 LIBS= -lm $(SYSLIBS) $(MYLIBS)

lijiansong

GDB commands by function - simple guide
---------------------------------------
More important commands have a (*) by them.

Startup 
% gdb -help         	print startup help, show switches
*% gdb object      	normal debug 
*% gdb object core 	core debug (must specify core file)
%% gdb object pid  	attach to running process
% gdb        		use file command to load object 

lijiansong

# INSTALL INSTRUCTIONS: save as ~/.gdbinit
#
# DESCRIPTION: A user-friendly gdb configuration file.
#
# REVISION : 7.3 (16/04/2010)
#
# CONTRIBUTORS: mammon_, elaine, pusillus, mong, zhang le, l0kit,
#               truthix the cyberpunk, fG!, gln
#
# FEEDBACK: https://www.reverse-engineering.net

lijiansong

# Instrument binaries, pgo data to /data/pgo, serial make is important to not confuse the pgo generator
env CXXFLAGS='-march=native -fprofile-dir=/data/pgo -fprofile-generate=/data/pgo' cmake .. -DCMAKE_BUILD_TYPE=Release                   
make -j 1

# Run instrumented program, generate and write pgo data
./runIt

# Use profile data and feed into gcc, correct for threading counter noise, serial make is important to not confuse the pgo generator
env CXXFLAGS='-march=native -fprofile-dir=/data/pgo -fprofile-use=/data/pgo -fprofile-correction' cmake .. -DCMAKE_BUILD_TYPE=Release                   
make -j 1