This directory contains test harness to validate many parts of the code in the src/ directory. I am documenting the most important code in this directory.

On most POSIX systems (including Darwin):

gmake -j5

The object files and exe will be in a OS and build specific directory; the directory name will have the following convention:

./obj-$UNAME-$ARCH-$TYPE/

Where:

• $UNAME: is the lowercase OS name
• $ARCH: is the common arch name (amd64, aarch64, etc.)
• $TYPE: is the release type dbg or rel; where the latter have optimizations enabled.

If you wish to build for benchmarking and full optimization, try:

gmake OPTIMIZE=1 -j5

t_mpmcq.c

Test harness for Multi-producer, Multi-consumer Queue. Uses timestamps to record when items are pushed and pulled. It dumps sorted (ordered) timestamps and "delta" to stdout. Use the output with latplot.py to see variance in latency. Optional arguments: NPRODUCERS NCONSUMERS.

t_spscq.c

Test harness for Single-producer, Single-consumer queue. It verifies consistency of queue operations. It prints a summary of performance results upon test completion.

t_hashbench.c

Benchmark various hash functions by reading tokens (keys) from stdin. Prints the hashing speed to stdout (MB/s and cyc/byte).

t_hashspeed.c

Benchmark various hash functions by using synthetic data of various sizes. The sizes are supplied on the command line. e.g., ./Darwin_objs_rel/t_hashspeed 32 56 64 will print hash function performance for keys of size 32 bytes, 56 bytes and 64 bytes respectively.

t_bloom.c

Test harness and benchmark for Bloom filters. Reads tokens from stdin or the input file provided on command line.

t_mempool.c

Pooled memory allocator test harness.

t_fast-ht.c

Test harness and benchmark for super-fast hash table.

t_fast-ht-basic.c

Simple test harness for the super-fast hash table.

t_arena.c

Test harness and benchmark for object-lifetime based memory allocator.

zbuf_eg.c

Example program to show usage of the zlib.h buffered I/O interface (

I wrote an example program to use the user-space arc4random implementation -- except it turned into a standlone program to wipe disks using parallel I/O (multiple threads). This is collected in 3 files:

dd-wipe-opt.c

Auto-generated command-line processing file from the input dd-wipe-opt.in.

disksize.c

Semi portable implementation of grokking disk-size for various operating systems. THis has code for Darwin and Linux.

mt-dd-wipe.c

Main program to drive the disk-wipe logic. Usage: mt-dd-wipe FILE|DISKNAME Try with mt-dd-wipe --help on Darwin or Linux.

Use this on a live-system at your own risk!

confparse.py

Parse a python-ish config file (white-space indented).

createlog.py

Create multiple "log files" for testing file rotation logic.

genhash.py

Scan one or more paths provided on the command line and find all "#define" tokens; write them to stdout. Good example of using python's multi-processing module to scan files in parallel.

latplot.py

Plot the GET/PUT latencies of the MPMC Queue test harness and show a nice graph. Requires 'matplotlib' and 'numpy'.