Ffwd
This is the public release of the ffwd delegation system, and associated benchmarks from the SOSP'17 paper.
Install / Use
/learn @bitslab/FfwdREADME
================== | ffwd benchmarks|
Sepideh Roghanchi, Jakob Eriksson, Nilanjana Basu (SOSP '17)
This file has instuctions on how to use the benchmarks provided in this repository, for more information, please send an email to srogha2@uic.edu.
The ffwd code and the benchmarks are developed and tested on 4 different machines. To get the best result compatibility the machine specifications and dependencies are provided below:
AMD Opteron 6378
uname -a: Linux 3.8.0-29-generic #42~precise1-Ubuntu SMP Wed Aug 14 16:19:23 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux gcc version: gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 2.4 GHz
Intel Xeon E5-4620
uname -a: Linux 4.0.7 #2 SMP Wed Jul 1 19:50:57 CDT 2015 x86_64 x86_64 x86_64 GNU/Linux gcc version: gcc (Ubuntu 5.4.0-6ubuntu1~16.04.4) 5.4.0 2.20 GHz
Intel Xeon E7-4820
uname -a: Linux 2.6.37.6 #408 SMP Sun Oct 9 22:07:58 PDT 2016 x86_64 x86_64 x86_64 GNU/Linux gcc version: gcc (Ubuntu/Linaro 4.7.3-1ubuntu1) 4.7.3 2.00 GHz
Intel Xeon E5-4660 v4
uname -a: Linux 4.4.0-62-generic #83-Ubuntu SMP Wed Jan 18 14:10:15 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux gcc version: gcc (Ubuntu 5.4.0-6ubuntu1~16.04.4) 5.4.0 2.20 GHz
============ Benchmarks
libnuma and PAPI need to be installed prior to compile the benchmarks.
The include directory contains the ffwd library and a simple sample code called "ffwd_sample.c": to compile: make ffwd_sample to run: ./ffwd_sample -t NUM_THREADS -s NUM_SERVERS -d DURATION_IN_MS
For memcached benchmarks, run './configure.sh' in the current directory for once In order to compile all the benchmarks, run 'make' or run 'make BENCHMARK_NAME' to compile a specific benchmark.
In the scripts directory, you can find scripts to run and plot the results. For each benchmark the results will be stored a directory named "results/" under the benchmarks directory.
Related Skills
node-connect
343.1kDiagnose OpenClaw node connection and pairing failures for Android, iOS, and macOS companion apps
frontend-design
90.0kCreate distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, or applications. Generates creative, polished code that avoids generic AI aesthetics.
openai-whisper-api
343.1kTranscribe audio via OpenAI Audio Transcriptions API (Whisper).
qqbot-media
343.1kQQBot 富媒体收发能力。使用 <qqmedia> 标签,系统根据文件扩展名自动识别类型(图片/语音/视频/文件)。
