01
AI Acceleration
Divide neural network inference across processors while accounting for accuracy, latency, energy, and shared memory.
Research topicColorado School of Mines
High Performance Systems and Software Lab
We study how CPUs, GPUs, and specialized accelerators share work, memory, and energy.
Our group at Colorado School of Mines builds models and runtime systems for applications from neural network inference to autonomous robots.
Computer Science
High performance computing,
architecture and AI systems
AI AccelerationHeterogeneous computingAutonomous systemsArchitecture and security
Research areas
02
Heterogeneous and edge computing
Predict resource contention and decide where applications should run across devices, edge servers, and cloud systems.
Research topic
03
Autonomous systems
Plan robot motion and computational schedules together under timing, energy, and perception constraints.
Research topic
04
Architecture and security
Study information leakage through shared memory and the memory bottlenecks of emerging processor architectures.
Research topic
Workforce Development for Semiconductor Industry
We develop tools that connect computer vision, interactive software, and semiconductor manufacturing education.
SMILE
Interactive videos, 360° cleanroom tours, and augmented reality guidance for learning equipment and procedures.
Try the video demoCoRtEx
A Mines certificate pathway that combines coursework, laboratory training, and a semiconductor industry capstone.
Certificate requirementsRecent publications
ICCAD2026
DRIFT: Dynamic Intra-Frame Parallelization for Fast Continuous Object Detection on Multi-Accelerator SoCs
Partition and schedule continuous object detection across multiple accelerators.
IROS2026
Computing is Physical: Planning for Shared Resources in Robot Motion and Heterogeneous Computational Scheduling
Plan robot motion and computational schedules under shared physical resource limits.
ACL2026
Closing the Spatial Execution Gap in Digital Whiteboards via Verifiable Reinforcement Learning
Train models to perform spatially precise whiteboard operations with verifiable rewards.
ISPASS2026
Hibiscus: End-to-end Architectural Simulation Framework for Hybrid SFQ/CMOS-Memory Compute Systems (Tool or Benchmark Paper)
Simulate hybrid superconducting compute systems and CMOS memory end to end.
For prospective students
Student research
We welcome research interest from Ph.D., master’s, and undergraduate students. Explore a project, read a paper, and tell us what you would like to investigate.
Research with usLed by Mehmet E. Belviranli, Associate Professor of Computer Science.
News
ArchiveJuly 2026
NNME Southwest supports the Mines semiconductor education effort. See the CoRtEx certificate and project.
July 2026
DRIFT, on multi-accelerator object detection, is accepted at ICCAD 2026.
June 2026
Computing is Physical, on joint robot motion planning and computational scheduling, is accepted at IROS 2026.
April 2026
Mehmet Belviranli is promoted to Associate Professor with tenure.