GPU

Tools for Top-down Performance Analysis of GPU-Accelerated Applications

Presented our ICS'20 work.

GVProf

We implemented GVProf, the first value profiler that locates value redundancy problems in applications running on GPU-based clusters. Our experiments show that GVProf incurs acceptable overhead and scales to large executions. GVProf provides useful insights to guide performance optimization. Under the guidance of GVProf, we optimized several HPC and machine learning workloads, obtaining speedups up to 1.93x.

A Tool for Top-down Performance Analysis of GPU-accelerated Applications

Presented our GPU performance tool

Tools for Top-down Performance Analysis of GPU-Accelerated Applications

This paper describes extensions to Rice University's HPCToolkit performance tools to support measurement and analysis of GPU-accelerated applications. To help developers understand the performance of accelerated applications as a whole, HPCToolkit's …

Optimizing GPU-accelerated Applications with HPCToolkit

Presented our GPU performance tool

HPCToolkit

Our tool provides a profile view and a trace view for GPU-accelerated applications. The profile view identifies where GPU APIs are invoked in CPU calling context, approximates calling context for GPU execution, and analyzes instruction mix for GPU kernels. The tool traces CPU and GPU activities for a large number of processes and threads with minimal overhead.

A Tool for Performance Analysis of GPU-accelerated Applications

Presented the prototype of our GPU performance tool

A Performance Analysis Framework for Exploiting GPU Microarchitectural Capability

Presented our ICS'17 work.

Deep Learning on Modern Architectures

Discussed how state-of-the-art deep learning libraries optimize computations by utilizing architectural features.

A performance analysis framework for exploiting GPU microarchitectural capability