Publications

Conference Papers


A Toolkit for Profiling and Call Graph Analysis for RISC architectures based on Program Execution Traces

Published in 2025 lEEE International Conference on Cloud Computing Technology and Science (CloudCom), 2025

This paper proposes a novel matrix multiplication optimization for Huawei Ascend NPUs that offloads narrow MatMul computations from the underutilized Cube Unit to the Vector Unit using AscendC instructions. Applied to MLA inference in DeepSeek-V3, the method achieves a 20% mean performance gain in single-token processing by overlapping AIV and AIC execution.

Download Paper

A Toolkit for Profiling and Call Graph Analysis for RISC architectures based on Program Execution Traces

Published in XI International Conference. "Engineering and Telecommunications โ€“ En&T 2024", 2025

The paper solves the problem of analyzing the behavior of programs for RISC architectures based on binary execution traces. As part of the work, the profile generation algorithm was improved, and its subsequent visualization was supported using the KCachegrind application with an accuracy of up to linear sections of code. As a result of the work, accurate application execution profiles were obtained on the SPEC CPU 2017 performance benchmarks.

Download Paper

Detailed profile generation and visualization for RISC architectures based on program execution traces

Published in Proceedings of the 65th All-Russian Scientific Conference of MIPT in honor of the 115th anniversary of L.D. Landau, 2023

The paper is devoted to the development of an algorithm for generating a profile based on a binary execution trace under the RISC architecture. The output format of the Callgrind application was taken as a basis, so that it was possible to visualize the execution profile using KCachegrind.

Recommended citation: Shurygin A., Petushkov I. (2023). "Detailed profile generation and visualization for RISC architectures based on program execution traces." Proceedings of the 65th All-Russian Scientific Conference of MIPT in honor of the 115th anniversary of L.D. Landau, April 3โ€“8, 2023. Radio engineering and computer technology..
Download Paper | Download Slides

Verification of static binary optimizing translation for RISC architecture

Published in Proceedings of the 64th All-Russian Scientific Conference of MIPT, 2021

The paper is devoted to static verification of applications optimized by BOLT, which ensured the detection of errors on SPEC CPU 2017 when writing new optimizations.

Recommended citation: Lisitsyn S., Shurygin A. (2021). "Verification of static binary optimizing translation for RISC architecture" Proceedings of the 64th All-Russian Scientific Conference of MIPT. November 29 โ€“ December 03, 2021 Radio Engineering and Computer Technologies.
Download Paper