Evgeny Mankov
Principal Member Of Technical Staff at AMD
Belgrade, Central Serbia
Summary
🤩
Rockstar🎓
Top SchoolEvgeny Mankov is a Principal Member of Technical Staff with 11+ years specializing in GPU runtimes, compilers and heterogeneous computing, currently based in Belgrade. He has led engineering efforts at AMD to port CUDA ecosystems to ROCm/HIP, integrate AMD GPU support into the Apollo autonomous-driving stack, and build the ROCm OpenCL compiler driver that seeded later products. Deeply experienced in LLVM/Clang, CUDA, HIP and toolchain integration, he combines low-level compiler and runtime expertise with pragmatic DevOps and build-system engineering across Windows and Linux. Notably, his open-source work includes adding device-property and cuBLAS support to ROCm HIP and fixing tricky CUDA-on-Windows Clang issues, demonstrating both large-system design and careful bug-level craftsmanship.
11 years of coding experience
19 years of employment as a software developer
specialist, Hardware and software support of electronic computer systems, 5,0, specialist, Hardware and software support of electronic computer systems, 5,0 at Military Space Force Engineering Academy
Русский, English, German, Serbian, Ukrainian
Github Skills (25)
docker10
c-language10
windows10
gpu-programming10
build-system10
c1110
dockers10
c1710
hip10
roc10
compiler-compiler10
cplus10
bazel10
cuda10
compiler10
Programming languages (7)
Github contributions (5)
ROCm/hip
Feb 2016 - Nov 2020
HIP: C++ Heterogeneous-Compute Interface for Portability
Role in this project:
Back-end Developer
Back-end DeveloperContributions:18 reviews, 917 commits, 713 PRs in 4 years 10 months
Contributions summary:Evgeny primarily worked on implementing device properties and related functionalities within the HIP (Heterogeneous-Compute Interface for Portability) project, focusing on CUDA and ROCm. They added support for device properties like `concurrentKernels` and various aspects related to BDFID (BusID/DeviceID/FunctionID) support within the HIP runtime. Additionally, the user added support for several functions from the cuBLAS library.
cudaheterogeneousgpuportabilityhip-runtime
ApolloAuto/apollo
Feb 2022 - Aug 2022
An open autonomous driving platform
Role in this project:Back-end & DevOps Engineer
Back-end & DevOps EngineerContributions:4 reviews, 259 commits, 5 comments in 6 months
Contributions summary:Evgeny's contributions focused on adapting the Apollo autonomous driving platform to support AMD GPUs. This involved extensive code modifications to integrate AMD's HIP and MIOpen libraries, replacing CUDA and cuDNN calls. Furthermore, the user implemented changes to the build system, introducing GPU platform configurations for both NVIDIA and AMD GPUs and modifying the project's bash scripts for Docker builds to include AMD GPU support. These changes improve the platform's hardware compatibility and enhance its development workflow.
autonomousmachine-learningautonomyautonomous-drivingapollo
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.