Scalable Tools Workshop

Binary Instrumentation for the AMD GPU

Hsuan-Heng Wu and Ronak Chauhan

Abstract

abs tbd

Luthier, a Dynamic Binary Instrumentation Framework Targeting AMD GPUs

Matin Raayai-Ardakani, Norman Rubin, and David Kaeli

Abstract

Dynamic Binary instrumentation (DBI) is a widely used technique for collecting detailed, fine-grained information from program execution without requiring recompilation or access to the program’s source code. DBI provides several benefits over static instrumentation, including full code discovery and the ability to enable/disable profiling selectively during runtime. Earlier efforts focused on CPU-based DBI, which included ATOM, Pin and DynamoRio. More recently, we have seen the introduction of NVBit and GTPin have extended DBI capabilities to NVIDIA and Intel GPUs, respectively. Presently, there is no available DBI framework for AMD GPUs. This is primarily due to the unique challenges posed by AMD’s hardware and its ROCm runtime. In this presentation we will cover some of these challenges, and present our approach to enabling DBI on AMD GPUs. In addition, we will report on our efforts to develop user-facing APIs and internal components for this new DBI framework.

Efficient Instrumentation and Tracepoint Insertion for GPU Compute Kernels

Sébastien Darche

Abstract

Reducing instrumentation noise and overhead in software development tools is a major factor in providing insightful performance reports for programmers. This is particularly challenging when tracing highly parallel GPU compute kernels, which poses many challenges for instrumentation, data movement, and trace analysis. Furthermore, while complex kernels usually benefit the most from tracing tools, the instrumentation overhead often strongly correlates with the kernel complexity. Thus, GPU compute kernel tracing is a prime candidate for improvement.

We propose a method for efficient tracepoint placement in GPU compute kernels, by leveraging properties derived from static analysis of the control flow graph (CFG) at compilation time. This is enabled by GPUs relying on stack-based SIMT control flow, allowing for postmortem computation of vector control flow data. Compared to current tracing methods, our approach can reduce the number of instrumentation points, while guaranteeing the same level of detail when processing the trace.

We evaluate the reference implementation of our method on a comprehensive scientific computing benchmark, obtaining on average a reduction of 59% on the total number of inserted tracepoints, thus reducing runtime overhead and total trace size, when tracing a program. The reference implementation is freely available and integrated into a complete GPU tracing tool, ready for use by programmers.