David Medina

July 6

0x001: Programming Padawan

Video games were my inspiration for learning how to program. Excited about making the next Final Fantasy or World of Warcraft, I took one of the "Computer Science 101" classes during my undergraduate studies. Oh man was that a let down (゜-゜).
Read More . . .
August 3

Working at Two Sigma

Started working at the Houston branch
June 1 - July 11

Visiting Researcher at LLNL

Continue working on OCCA (thesis work) and collaborate with LLNL guys in ASQ/CASC.
May 26

PhD Defense

OKL: A Unified Language for Parallel Architectures (PDF)
May 7

Visit to Sandia National Lab: Seminar Talk

OKL: A Unified Language for Parallel Architectures
March 14-15

SIAM CSE '15: Talk + Poster

OCCA: An Extensible Portability Layer for Many-core Programming (PDF)
March 04-05

Rice Oil & Gas 2015: OCCA Tutorial + Talk

OCCA: Portability Layer for Many-core Thread Programming
Links to the: Slides, Source Code
March 03

OCCA Tutorials 02: Unified Virtual Address + Managed Memory + Shared Memory

February 28

OCCA Tutorials 01: Installation + Introduction

February 11

Visit to Chevron: Talk

OKL: A Unified Language for Parallel Architectures
February 04

PhD Proposal

OKL: A Unified Language for Parallel Architectures
2015
June 24

ICOSAHOM '14: Talk

OCCA: A Unified Approach to Multi-Threading Languages
May - August

Summer Internship at Shell

Optimized finite difference codes on multiple architectures (OpenMP, OpenCL, CUDA)
Investigated hybrid kernels with portable performance across varying architectures
April 23

Master Thesis Defense

OCCA: A Unified Approach to Multi-Threading Languages (PDF)
March 06

Rice Oil & Gas Workshop '14: Poster

OCCA: A Unified Approach to Multi-Threading Languages
2014
May - August

Summer Internship at Shell

Worked on numerical methods in parallel architectures (OpenMP, OpenCL, CUDA)
January

PASI: Workshop on Tsunami Modeling (NSF/DOE-Funded Travel Grant)

Attended multiple workshops and talks about tsunami modeling
Lab co-coordinator to teach OpenCL/OpenGL interop for applying and visualizing finite difference simulations
2013

OCCA


OCCA is an open-source (MIT license) library used to program current multi-core/many-core architectures.Devices (such as CPUs, GPUs, Intel's Xeon Phi, FPGAs, etc) are abstracted using an offload-model for application development and programming for the devices is done through a C-based (OKL) or Fortran-based kernel language (OFL).OCCA gives developers the ability to target devices at run-time by using run-time compilation for device kernels.