Submitted proposal For a few months I have been working on and off on a proposal with Kirk for a large power-pack cluster. Yesterday we submitted it. Your Sponsored Research Office or equivalent at your organization has submitted the following proposal to NSF: Title: CI-ADDO-NEW: Bringing Power to the People: Enabling scalable computing energy-efficiency [...]

CS4234 2011 L14 - Networks Download the slides from todays CS4234 lecture here. [...]

PMBS11 Paper accepted A couple of weeks ago we submitted a paper to PMBS11 and yesterday we got the email that it has been accepted. From: PMBS11 Date: Thu, 13 Oct 2011 22:44:07 +0100 Subject: PMBS11 notification for paper 15 Dear Mat, We are pleased to inform you that your paper titled Critical Path-Based Thread Placement for NUMA [...]

PMBS11 Paper submitted I have been working with Chun-Yi all year on optimal thread placement for NUMA systems. We just submitted the first paper on this work to the ACM Performance Modeling, Benchmarking and Simulation of High Performance Computing Systems workshop. The workshop part of Super Computing 2011 in Seattle. [...]

Cluster 2011 Presentation This morning I presented the iso-energy-efficiency paper in the "Technical Papers: System Performance" session of Cluster 2011. The presentation went well and I had lots of people come to me after the session and during lunch to ask questions and make suggestions about measuring power in GPUs. I have uploaded [...]

Cluster 2011 Austin, TX Today I flew out to Austin to attend IEEE Cluster 2011. I am presenting the iso-energy-efficiency paper I wrote with Leon in May. Big change from Blacksburg weather, it is 100 degrees here! [...]

MPI where am I? Several students in parallel computation (CS4234) were having problems getting set numbers of MPI processes to run on a set number of compute nodes. To help I have written a simple MPI program which will print out the hostname of the node it is running on. By printing the hostname [...]

CS4234 2011 L10 - Shared Memory Java Download the slides from todays CS4234 lecture here. [...]

CS4234 2011 L09 - Shared Memory OpenMP Download the slides from todays CS4234 lecture here. [...]

Pthreads race condition demo Here is a clear text version of the Pthreads race condition example. This code is a good base to build the homework from. The commented out lines will prevent the race condition. #include <pthread.h> #include <stdio.h> #include <stdlib.h> #define NTHREADS 100 #define LOOPS [...]

CS4234 2011 L07 - Shared Memory Pthreads Download the slides from todays CS4234 lecture here. [...]

MPI find minimum demo Here is a clear text version of the MPI find minimum example. I have included compilation and execution examples. #include "mpi.h" #include <stdlib.h> #include <stdio.h> #include <string.h> #define MAXSIZE 1000 int main(int argc, char **argv) { int myid, numprocs, data[MAXSIZE], i, x, myresult, result, status; FILE *fp; MPI_Init(&argc,&argv); MPI_Comm_size(MPI_COMM_WORLD,&numprocs); MPI_Comm_rank(MPI_COMM_WORLD,&myid); [...]

CS4234 2011 L05 - MPI Part 2 Download the slides from todays CS4234 lecture here. [...]

Install MPICH2 on a Mac As part of the programming parts of the parallel computation lectures I try and demo an example as part of the lecture. This is a bit of fun but it also proves the homework is doable since I will typically include the simplest version of the homework as my example. I [...]

MPI Hello World If people are having problems copying from the lecture notes here is a clear text version of the example Hello World. I have included compilation and execution examples. hello.c #include <stdio.h> #include <string.h> #include <stddef.h> #include <stdlib.h> #include "mpi.h" main(int argc, char **argv ) { char message[20]; int i,rank, size, type=99; MPI_Status status; [...]

CS4234 2011 L03 - MPI Part 1 Download the slides from todays CS4234 lecture here. [...]

CS4234 2011 L02 - Parallel Computers Download the slides from todays CS4234 lecture here. [...]

Internal Powerpack I have spent the past few weeks helping Kirk write a proposal for a new cluster. Every node will be fully instrumented by Powerpack to measure power consumption. Powerpack at scale has a number of challenges, one of which is how to fit the instrumentation inside the case. We can't [...]

System G Powerpack We are currently building a new fully instrumented Powerpack node for our System G cluster. System G is a cluster of Apple tower computers and Powerpack allows us to measure how much power a computer is using at the component level. Most people only know how much electricity a computer [...]

Cluster paper camera ready I have been working on the final version of our paper for IEEE Cluster 2011. Today I submitted the camera ready version and registered for the conference. I have uploaded a pre-print. [...]