No description, website, or topics provided.
Assembly C++ C CMake
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
asmjit @ ab59e35
scratchpad
src
tools
.gitignore
.gitmodules
README.md

README.md

Thundercat

This is a sparse matrix-vector multiplication (SpMV) library that generates SpMV code specialized for a given matrix. Code is generated and executed dynamically for the X86_64 architecture. For technical details see:

Files

  • main.cpp
  • matrix.*: Matrix class that keeps matrix information in CSR format.
  • method.*: Specialization methods.
  • profiler.*: Time measurement support.
  • svmAnalyzer.*: Feature extration from the matrices. Features are used for autotuning (done separately, not integrated here).
  • plaincsr.*: SpMV implementation using the CSR format.
  • mkl.*: SpMV using Intel MKL (calls mkl_dcsrmv after setting the parameters).

Runtime Specialization Methods

  • CSRbyNZ
  • RowPattern
  • GenOSKI <r> <c> (Similar to PBR. GenOSKI44 is for 4x4 block size, GenOSKI55 is for 5x5.)
  • Unfolding
  • CSRWithGOTO
  • UnrollingWithGOTO

See the papers for details. For each method, there exist a corresponding .cpp file.

How to Compile

Thundercat uses asmjit for assembling executable code at runtime. It should be placed under the main thundercat folder. Below are the steps. In these steps, we use cmake to generate build files for Ninja. You may generate build files for other targets as well. In particular, if you don't have Ninja installed, replace -G Ninja below with -G "Unix Makefiles", and replace ninja with make.

  1. Clone this git repository.
    ~ $ git clone https://github.com/ozusrl/thundercat.git
    
  2. Clone asmjit under thundercat
    ~ $ cd thundercat
    ~/thundercat $ git clone https://github.com/asmjit/asmjit.git
    
  3. Build asmjit.
    ~/thundercat $ cd asmjit
    ~/thundercat/asmjit $ mkdir build
    ~/thundercat/asmjit $ cd build
    ~/thundercat/asmjit/build $ cmake -G Ninja -DCMAKE_BUILD_TYPE=Release ..
    ~/thundercat/asmjit/build $ ninja
    ~/thundercat/asmjit/build $ cd ../..
    
  4. Build thundercat.
    ~/thundercat $ mkdir build
    ~/thundercat $ cd build
    ~/thundercat/build $ cmake -G Ninja -DCMAKE_BUILD_TYPE=Release ../src
    ~/thundercat/build $ ninja
    

This will produce a main executable file, named thundercat. The library runs on Mac OS X and Linux.

Setting the CMAKE_BUILD_TYPE variable to Debug is a good idea for a build that you will use for debugging purposes. All the benchmarkings, however, should be done using a build configured as Release.

To force a particular compiler, e.g. icc, do the following:

~/thundercat/build $ cmake -G Ninja -DCMAKE_BUILD_TYPE=Release -DCMAKE_C_COMPILER=`which icc` -DCMAKE_CXX_COMPILER=`which icc` ../src

How to Run

./thundercat <matrixName> <methodName> [optional flags]

<matrixName> is the path to the .mtx file (i.e. the matrix file as downloaded from the Matrix Market or the U. of Florida collection). The name should be provided without the .mtx extention.

The following are recognized as <methodName>:

  • Specialization methods: CSRbyNZ, RowPattern, Unfolding, GenOSKI33, GenOSKI44, GenOSKI55, UnrollingWithGOTO, CSRWithGOTO
  • Non-generative methods: MKL and PlainCSR

Optional flags

  • -num_threads <num_threads>: Number of threads to be used. By default, a single thread is used.
  • -debug: Output vector is printed.
  • -dump_matrix: Prints matrix's rows, cols and vals array (in the format required by the chosen method).
  • -dump_object: Dumps the generated object code to current folder into files named generated_X where X is a number from 0 up to the thread count.
  • -matrix_stats: Prints the svmAnalyzer's results for the current matrix.

Examples

Run for the rajat22 matrix (assuming rajat22.mtx exists in the current directory), using the RowPattern method with 6 threads:

./thundercat rajat22 stencil -num_threads 6

Sample Matrices

There is a small set of matrices at

http://srl.ozyegin.edu.tr/matrices_debug.tar.gz

Matrices used in the GPCE 2014 paper are available at

http://srl.ozyegin.edu.tr/matrices_gpce.tar.gz