Roy Oursler
Senior Software Developer at Intel Corporation
Portland, Oregon Metropolitan Area United States
Summary
🤩
Rockstar🎓
Top SchoolRoy Oursler is a Senior Software Developer in the Portland area with 10 years of experience building high-performance C and assembly systems for data centers and machine learning. At Intel he has delivered extreme-performance work—implementing decompression routines up to 2.5x faster than zlib and optimizing oneDNN kernels and bf16 handling—demonstrating deep expertise in low-level optimization and numerical accuracy. His background in mathematics and a PhD-level grounding informs rigorous algorithmic approaches to hashing, ECC, and compression, while his open-source contributions to projects like isa-l and oneDNN show a commitment to production-grade community software. Colleagues rely on him for squeezing out microarchitectural gains and restructuring core math implementations that improve both speed and correctness.
10 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy - PhD, Mathematics, Doctor of Philosophy - PhD, Mathematics at Arizona State UniversityBachelor of Science - BS, Mathematics w/ minor in Computer Science, Bachelor of Science - BS, Mathematics w/ minor in Computer Science at University of Wyoming
Github Skills (13)
compression-algorithm10
opencl10
hashing-algorithm10
c-language10
assembly10
cprogramming-language10
performance-optimization10
assemble10
assembler10
x8610
jit9
convolution9
c-programming8
Programming languages (2)
Github contributions (5)
intel/isa-l
Mar 2016 - May 2019
Intelligent Storage Acceleration Library
Role in this project:
Back-end Developer
Back-end DeveloperContributions:235 commits, 9 comments in 3 years 3 months
Contributions summary:Roy's commits center on modifications within the Intelligent Storage Acceleration Library (isa-l) project, with a focus on optimizing the Intel IGZIP library. They are modifying and implementing assembly code, primarily focusing on low-level optimizations and improving performance through enhanced algorithms. The user is adding macros and optimized code for low-level instructions like BZHI, SHRX, and SHLX, for older architectures, as well as updating hash functions within the IGZIP compression algorithm. These modifications suggest a focus on performance engineering and optimization of compression algorithms.
compressionaccelerationstorageintelligent
uxlfoundation/oneDNN
Aug 2019 - Jan 2023
oneAPI Deep Neural Network Library (oneDNN)
Role in this project:Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:241 reviews, 379 commits, 93 PRs in 3 years 6 months
Contributions summary:Roy primarily focused on optimizing the performance of the oneAPI Deep Neural Network Library (oneDNN). Their contributions centered around improving the accuracy of verbose printing for bfloat16 (bf16) data types in various components such as CPU, jit, and convolution kernels. They also made significant changes to the library's internal implementations, including the restructuring of modulus calculations and optimizations of the math library, and made code changes in multiple code generation files.
bfloat16sse41avx512openmpopencl
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.