Parallelizing, Vectorizing and Optimizing Memory Access Patterns for Performance


Details
Join us at 7:00 PM on Wednesday, October 4th as Alex Habeger presents "Parallelizing, Vectorizing and Optimizing Memory Access Patterns for Better Performance"! Doors will open at 6:30 so please arrive ahead of time so we can get ready.
This talk will demonstrate how parallelizing with threads, vectorizing with SIMD, and using efficient memory access patterns can dramatically increase performance of even simple code. Alex Habeger has a long history with concurrency and low level programming. He maintained MPI code for the DOE where he began his unhealthy relationship with pthread_create, then went on to write entirely too much Ada in college which he is still recovering from. He then spent 5 years doing safety critical math and project management in the Army. He is currently working at IBM on LLVM optimizations for Power processors.
The talk will take place at Rackspace's Austin office, located on the East side of IH-35 right before Rundberg and after Henna Chevrolet (there is a sign for Rackspace which is visible from the highway). Feel free to park in the same lot as the big Rackspace flag.

Parallelizing, Vectorizing and Optimizing Memory Access Patterns for Performance