Installing Linpack Benchmark (HPL) on a Rocks Cluster
- Performed on a Rocks Cluster built on this version:
- area51+base+bio+condor+ganglia+hpc+java+kernel+kvm+os+perl+python+service-pack+sge+web-server+zfs-linux-6.1.x86_64.disk1
As ROOT:
- Add a new user (useradd linpacker) and set a password for newly added user (passwd linpacker)
- Sync users for ROCKS (rocks sync users)
Download LinX - Put your CPU under the scope to detect any potential hardware malfunctions with the help of this lightweight but powerful application. Derivative of arctan. HPC Challenge benchmark consists of 7 tests: (1) HPL - the Linpack TPP benchmark which measures the floating point rate of execution for solving a linear system of equations. (2) DGEMM - measures the floating point rate of execution of double precision real matrix-matrix multiplication.
Performance Summaryを項目を見ると大体33GFlopsから37GFlops位出ているようだ。 スパコンと比べてみると。 実際に、京速コンピューターと較べてみる。 手元のmac book proが約37GFlops,こちらによると、京のLINPACKの実効性能(Rmax)は、10510TFlopsだそうだ。.
As New User (linpacker):
GoToBLAS Portion
- Download GotoBLAS from TACC website http://www.tacc.utexas.edu/tacc-projects/gotoblas2
- Untar it into /export/home/linpacker/ (tar -xvf GotoBLAS2-1.13.tar.gz)
- Change directory to the GoToBLAS2 directory (cd /export/home/linpacker/GotoBLAS2)
- Build GoToBLAS2 (/export/home/linpacker/GotoBLAS2/quickbuild.64bit)
- Run the MAKE command (make)
HPL (Linpack Portion)
- Change to home directory (cd /export/home/linpacker)
- Download HPL from NetLib website http://www.netlib.org/benchmark/hpl/hpl-2.1.tar.gz
- Untar it into /export/home/linpacker/ (tar -xvf hpl-2.1.tar.gz)
- Create a Symbolic Link for the HPL directory (ln -s /export/home/linpacker/hpl-2.1 /export/home/linpacker/hpl)
- Copy the Make.Linux_PII_FBLAS from the HPL setup directory to the HPL root (cp /export/home/linpacker/hpl/setup/Make.Linux_PII_FBLAS /export/home/linpacker/hpl)
- Edit the Make.Linux_PII_FBLAS file
(vim /export/home/linpacker/hpl/Make.Linux_PII_FBLAS) with the following changes (the examples below show the original commented line and the modded one below):- #TOPdir = $(HOME)/hpl
- TOPdir = /export/home/linpacker/hpl
- #MPdir = /usr/local/mpi
- MPdir = /opt/openmpi/
- #MPlib = $(MPdir)/lib/libmpich.a
- MPlib = $(MPdir)/lib/libmpi.a
- #LAdir = $(HOME)/netlib/ARCHIVES/Linux_PII
- LAdir = /export/home/linpacker/GotoBLAS2
- #LAlib = $(LAdir)/libf77blas.a $(LAdir)/libatlas.a
- LAlib = $(LAdir)/libgoto2.a -lm -L/usr/lib/gcc/i386-redhat-linux/4.1.2
- Psp root folder download. #CC = /usr/bin/gcc
- CC = /opt/openmpi/bin/mpicc
- #LINKER = /usr/bin/g77
- LINKER = /opt/openmpi/bin/mpif77
- Save and Quit the Make.Linux_PII_FBLAS file (If using VI/VIM :wq)
- Change to the HPL directory (cd hpl)
- Run the MAKE command specifying the architecture (make arch=Linux_PII_FBLAS)
- Make a backup of the HPL.dat file found in /export/home/linpacker/hpl/bin/Linux_PII_FBLAS/HPL.dat
(cp /export/home/linpacker/hpl/bin/Linux_PII_FBLAS/HPL.dat /export/home/linpacker/hpl/bin/Linux_PII_FBLAS/HPL.dat.original) - Tune the HPL.dat to suit your cluster
- Create a file that contains the nodes, separated by newlines, that will be subjected to the HPL benchmark (vim /export/home/linpacker/machines)
- Change to the directory containing the HPL.dat and xhpl executable
(cd /export/home/linpacker/hpl/bin/Linux_PII_FBLAS/) - Run the Benchmark!
/opt/openmpi/bin/mpirun -nolocal -np 6 -machinefile /export/home/linpacker/machines xhpl- Note: The number of processes specified by the -np parameter should correlate to the process grid configured in HPL.dat
Linpack Xtreme 1.1.3
April 22nd, 2020 - What's New- 8.2 MBLinpackXtreme-1.1.3.zip
- 7.5 MBLinpackXtreme-1.1.3.tar.gz
- 321.1 MBLinpackXtreme-1.1.3-Porteus-v4.0-x86_64.iso
- 318.3 MBLinpackXtreme-1.1.3-Porteus-v4.0-i586.iso
Linpack Xtreme 1.1.2
November 26th, 2019 - What's New- 8.2 MBLinpackXtreme_1.1.2.zip
Linpack Xtreme 1.1.1
November 19th, 2018 - What's New- 8.1 MBLinpackXtreme_1.1.1.zip
Linpack Xtreme 1.0.0
November 9th, 2018 - What's New
- 8.1 MB
TPU Edition with more customization options
LinpackXtreme_1.0.0_TPU.zip - 8.1 MBLinpackXtreme_1.0.0.zip
- 7.5 MBLinpackXtreme_1.0.0_Linux.tgz
- 321.1 MBLinpackXtreme_1.0.0_Porteus_v4.0_x86_64.iso
- 318.3 MBLinpackXtreme_1.0.0_Porteus_v4.0_i586.iso
Linpack Xtreme 0.9.5
November 5th, 2018 - What's New- 8.1 MBLinpackXtreme_0.9.5.zip
- 7.5 MBLinpackXtreme_0.9.5_Linux.tgz
- 321.1 MBLinpackXtreme_0.9.5_Porteus_v4.0_x86_64.iso
- 318.3 MBLinpackXtreme_0.9.5_Porteus_v4.0_i586.iso
Linpack Xtreme 0.9.4
November 1st, 2018 - What's NewLinpack Benchmark Test
- 8.1 MBLinpackXtreme_0.9.4.zip
- 7.4 MBLinpackXtreme_0.9.4_Linux.tgz
- 321.1 MBLinpackXtreme_0.9.4_Porteus_v4.0_x86_64.iso
- 318.3 MBLinpackXtreme_0.9.4_Porteus_v4.0_i586.iso
Linpack Xtreme 0.9.3
October 1st, 2018Linpack Windows
- 8.2 MBLinpackXtreme_0.9.3.zip
- 7.4 MBLinpackXtreme_0.9.3_Linux.tgz
- 318.3 MBLinpackXtreme_0.9.3_Porteus_v4.0_i586.iso
- 321.1 MBLinpackXtreme_0.9.3_Porteus_v4.0_x86_64.iso
Downloaded: 28,122 times (641.5 GB)
Linpack is a benchmark and the most aggressive stress testing software available today. Best used to test stability of overclocked PCs. Linpack tends to crash unstable PCs in a shorter period of time compared to other stress testing applications.
Linpack Xtreme is a console front-end with the latest build of Linpack (Intel Math Kernel Library Benchmarks 2018.3.011).
Linpack solves a dense (real*8) system of linear equations (Ax=b), measures the amount of time it takes to factor and solve the system, converts that time into a performance rate, and tests the results for accuracy. The generalization is in the number of equations (N) it can solve, which is not limited to 1000. Linpack uses partial pivoting to assure the accuracy of the results.
This project was created because Prime95 is no longer effective for stress testing like it used to be. LinX, IntelBurnTest, OCCT use outdated Linpack binaries from 2012.
Linpack solves a dense (real*8) system of linear equations (Ax=b), measures the amount of time it takes to factor and solve the system, converts that time into a performance rate, and tests the results for accuracy. The generalization is in the number of equations (N) it can solve, which is not limited to 1000. Linpack uses partial pivoting to assure the accuracy of the results.
This project was created because Prime95 is no longer effective for stress testing like it used to be. LinX, IntelBurnTest, OCCT use outdated Linpack binaries from 2012.
Make sure to keep an eye on the temperatures as Linpack generates excessive amount of stress like never seen before.