Yi-chen Lin
SMTS Software Developer Engineer at AMD
Taipei, Taiwan
Summary
👤
Senior🎓
Top SchoolYi-chen Lin is a senior software engineer with 10 years of experience specializing in 3D graphics, rendering and engine development, currently working on ROCm MathLib at AMD where he contributes to hipBLASLt, rocFFT and rocBLAS. He combines deep academic background in image-based rendering with practical engine work across games, AR and cross-platform mobile/desktop engines, having led rendering efforts on Unreal Engine projects and AR spatial mapping at HTC Vive. His open-source contributions to performance-critical GPU libraries like ROCm/Tensile and rocBLAS show a strong focus on correctness and low-level optimization, including packed-int8 GEMM support and refactors for maintainability. Comfortable bridging research and product, he is equally at home tuning shader/lighting pipelines or squeezing performance from BLAS backends, and is open to international opportunities beyond the games industry.
10 years of coding experience
15 years of employment as a software developer
Master Computer Science IBR (Image-Based Rendering) Lab, Master Computer Science IBR (Image-Based Rendering) Lab at National Tsing Hua University
Chinese, English
Github Skills (18)
assembly10
c-language10
performance-analytics10
performance-monitor10
matrix-multiplication10
blas10
performance-measurement10
hip10
gemfire10
roc10
performance-analysis10
gpu10
performance-tuning10
performance-monitoring10
cprogramming-language10
Programming languages (3)
Github contributions (5)
ROCm/Tensile
Apr 2020 - Apr 2021
Stretching GPU performance for GEMMs and tensor contractions.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:69 reviews, 67 commits, 59 PRs in 11 months
Contributions summary:Yi-chen primarily focused on improving the Tensile library's functionality and correctness. They addressed several bug fixes, including incorrect comments, parameter handling, and reading data from CSV files. Additionally, the user added support for features such as host-side assertions and UseBeta for convolution, enhancing the library's capabilities. They also refactored code for better maintenance, including using structures to encapsulate values.
amdpythontensorhipauto-tuning
ROCm/rocBLAS
May 2020 - May 2021
Next generation BLAS implementation for ROCm platform
Role in this project:Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:17 commits, 1 PR, 1 comment in 11 months
Contributions summary:Yi-chen primarily focused on enhancing the rocBLAS library, particularly concerning performance optimizations and the integration of new features for int8 GEMM operations. Their contributions included the implementation of packed-int8x4 support, modifications to the benchmarking framework for improved accuracy, and updates to accommodate diverse architectures and hardware layouts. The user’s work involved modifying core library files and adjusting existing code to achieve high performance in a performant manner.
blasnext-generationrocmhip
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.