Skip to content

Latest commit

 

History

History
193 lines (138 loc) · 7.38 KB

quickstart_cachesim.md

File metadata and controls

193 lines (138 loc) · 7.38 KB
Raw

cachesim user guide

cachesim is a tool provided by libCacheSim to quickly run some cache simulations, it supports

  • a variety of eviction algorithms such as FIFO, LRU, LFU, ARC, SLRU, LeCaR, CACHEUS, Hyperbolic, LHD, TinyLFU, Belady, LRB and GLCache.
  • a variety of admission algorithms such as size, bloomFilter and adaptSize.
  • text, csv trace as well as binary traces.
  • automatic multi-threaded simulations.

Meanwhile, cachesim has high-performance with low resource usages.

Installation

First, build libCacheSim. After building libCacheSim, cachesim should be in the build directory.

Basic Usage

./cachesim trace_path trace_type eviction_algo cache_size [OPTION...]

use ./cachesim --help to get more information.

Run a single cache simulation

Run the example vscsi trace with LRU eviction algorithm and 1GB cache size. Note that vscsi is a trace format, we also support csv traces.

# Note that no space between the cache size and the unit, unit is not case sensitive
./cachesim ../data/trace.vscsi vscsi lru 1gb

Run multiple cache simulations

# Note that there is no space between the cache sizes
./cachesim ../data/trace.vscsi vscsi lru 1mb,16mb,256mb,8gb

# Or you can quote the cache sizes
./cachesim ../data/trace.vscsi vscsi lru "1mb, 16mb, 256mb, 8gb"

# besides absolute cache size, you can also use fraction of working set size
./cachesim ../data/trace.vscsi vscsi lru 0.001,0.01,0.1,0.2

# besides using byte as the unit, you can also treat all objects having the same size, and the size is the number of objects
./cachesim ../data/trace.vscsi vscsi lru 1000,16000 --ignore-obj-size 1

# new feature: you can run a few algorithms in parallel by concatenating the algorithms
./cachesim ../data/trace.vscsi vscsi fifo,lru,arc,qdlp 0.01 --ignore-obj-size 1

# run 4*4 simulations in parallel (no more than n_thread at the same time)
./cachesim ../data/trace.vscsi vscsi fifo,lru,arc,qdlp 0.01,0.05,0.1,0.2 --ignore-obj-size 1

Auto detect cache sizes

cachesim can detect the working set of the trace and automatically generate cache sizes at 0.0001, 0.0003, 0.001, 0.003, 0.01, 0.03, 0.1, 0.3 of the working set size. You can enable this feature by setting cache size to 0 or auto.

./cachesim ../data/trace.vscsi vscsi lru auto

Use different eviction algorithms

cachesim supports the following algorithms:

You can just use the algorithm name as the eviction algorithm parameter, for example

./cachesim ../data/trace.vscsi vscsi lecar auto
./cachesim ../data/trace.vscsi vscsi hyperbolic auto
./cachesim ../data/trace.vscsi vscsi lhd auto
./cachesim ../data/trace.vscsi vscsi glcache auto

# belady and beladySize require oracle trace
./cachesim ../data/trace.oracleGeneral oracleGeneral beladySize auto

Use different trace types

We have demonstrated the use of cachesim with vscsi trace. We also support csv traces. To use a csv trace, we need to provide the column of time, obj-id, and obj-size. Both time and size are optional, but many algorithms rely on time and size to work properly. The column starts from 1, the first column is 1, the second is 2, etc. Besides the column information, a csv reader also requires the delimiter and whether the csv file has a header. cachesim builds in a simple delimiter and header detector, if the detected result is not correct, you can provide the correct information using delimiter=,, has-header=true.

# note that the parameters are separated by comma and quoted
./cachesim ../data/trace.csv csv lru 1gb -t "time-col=2, obj-id-col=5, obj-size-col=4"

# if object id is numeric, then we can pass obj-id-is-num=true to speed up
./cachesim ../data/trace.csv csv lru 1gb -t "time-col=2, obj-id-col=5, obj-size-col=4, obj-id-is-num=true"


# note that csv trace does not support UTF-8 encoding, only ASCII encoding is supported
./cachesim ../data/trace.csv csv lru 1gb -t "time-col=2, obj-id-col=5, obj-size-col=4, delimiter=,, has-header=true"

Besides csv trace, we also support txt trace and binary trace.

# txt trace is a simple format that stores obj-id in each line
./cachesim ../data/trace.txt txt lru 1gb

# binary trace, format is specified using format string similar to Python struct
./cachesim ../data/trace.vscsi binary lru 1gb -t "format=<IIIHHQQ,obj-id-col=6,obj-size-col=2"

# oracleGeneral is a binary format that stores time, obj-id, size, next-access-time (in reference count)
./cachesim ../data/trace.oracleGeneral.bin oracleGeneral lru 1gb

We recommend using binary trace because it can be a few times faster than csv trace and uses less DRAM resources.

Advanced usage

cachesim supports many advanced features, you can use ./cachesim --help to get more information. Here we give some examples.

Setting parameters for eviction algorithms

Some eviction algorithms have parameters, you can set the parameters by using -e "k1=v1,k2=v2" or --eviction-params "k1=v1,k2=v2" format.

# run SLRU with 4 segments
./cachesim ../data/trace.vscsi vscsi slru 1gb -e n-seg=4

# print the default parameters for SLRU
./cachesim ../data/trace.vscsi vscsi slru 1gb -e print

Admission algorithm

cachesim supports the following admission algorithms: size, probabilistic, bloomFilter, adaptSize. You can use -a or --admission to set the admission algorithm.

# add a bloom filter to filter out objects on first access
./cachesim ../data/trace.vscsi vscsi lru 1gb -a bloomFilter

Prefetching algorithm

cachesim supports the following prefetching algorithms: OBL, Mithril, PG (and AMP is on the way). You can use -p or --prefetch to set the prefetching algorithm.

# add a mithril to record object association information and fetch objects that are likely to be accessed in the future
./cachesim ../data/trace.vscsi vscsi lru 1gb -p Mithril

Advanced features

# change number of threads 
./cachesim ../data/trace.vscsi vscsi lru 1gb --num-thread=4

# cap the number of requests read from the trace
./cachesim ../data/trace.vscsi vscsi lru 1gb --num-req=1000000

# change output 
./cachesim ../data/trace.vscsi vscsi lru 1gb -o my-output

# ignore object size, each object has size one
./cachesim ../data/trace.vscsi vscsi lru 1gb --ignore-obj-size=true

# ignore object metadata size, different algorithms have different metadata size, this option will ignore the metadata size
./cachesim ../data/trace.vscsi vscsi lru 1gb --consider-obj-metadata=false

# use part of the trace to warm up the cache
./cachesim ../data/trace.vscsi vscsi lru 1gb --warmup-sec=86400

# Use TTL
./cachesim ../data/trace.vscsi vscsi lru 1gb --use-ttl=true