Yang Yang (杨扬)
yangyang [at] virginia [dot] edu

Email  /  LinkedIn  /  GitHub /  Google Scholar

Hi, I'm a PhD student in the Department of Computer Science at University of Virginia, advised by Prof. Adwait Jog.

Prior to that, I received my B.S. degree from Jilin University and was a member of ETECA Lab under Prof. Jingweijia Tan. I was also a visiting student at the State Key Laboratory of Processors at ICT, CAS, under Prof. Guangli Li.

I am seeking Research Internship (Summer 2026) in GPU architecture, memory technologies and confidential computing (TEEs). Feel free to drop me an email :)

I am interested in performance issues in GPU x Trusted Computing x Cryptography x Memory.

Recently, I have focused on understanding and addressing performance issues (e.g., data movement [ISCA'25]) in secure GPU architectures like confidential computing [ISPASS'25, +].

I am also interested in making cryptographic protocols (e.g., PIR and FHE) more practical using GPUs.

[March 2025] NetCrafter got accepted by ISCA'25, see you in Tokyo!
[March 2025] One paper got accepted by ISPASS'25, see you in Ghent!

2023 - Present
Ph.D. in Computer Science, University of Virginia

2019 - 2023
B.S. in Computer Science, Jilin University

Insight Lab, University of Virginia, USA
Aug. 2023 - Present
Advisor: Adwait Jog
Topics: GPU & Trusted-Computing (TEE, Cryptography, etc.) & Memory

State Key Laboratory of Processors, ICT, CAS, P.R.China
Jul. 2022 - Sept. 2023
Advisor: Guangli Li
Topics: Compiler & Profile-Guided Optimization & LLVM

ETECA Lab, Jilin University, P.R.China
Feb. 2022 - Jul. 2023
Advisor: Jingweijia Tan
Topics: GPU Power Modeling & MCM-GPU & Under-Voltage Reliability
Thesis: The Design and Implementation of Binary Code Analysis Framework for NVIDIA GPU

1. (Under Review)
   Performance-efficient GPU-based Confidential Computing: Are we there yet?
   Yang Yang, Adwait Jog
   [Show Brief] 

   Building on our ISPASS'25 results, we propose X. Our evaluation shows that X largely reduces CC overhead, achieving significant speedup (up to 5.1x).



2. (ISCA'25)
   NetCrafter: Tailoring Network Traffic for Non-Uniform Bandwidth Multi-GPU Systems
   Amel Fatima, Yang Yang, Yifan Sun, Rachata Ausavarungnirun, Adwait Jog
   In the Proceedings of ACM International Symposium on Computer Architecture, Tokyo, Japan, June 2025
   [PDF] [Slides.pdf] [Slides.pptx] [Code]  [Show Brief] 

   We present NetCrafter, a set of novel techniques to manage network traffic, especially across low-bandwidth links in multi-GPU systems. NetCrafter reduces the volume of flit traffic by (i) stitching compatible, partially filled flits, (ii) trimming unnecessary flits to avoid redundant transfers, and (iii) sequencing flits so that latency-sensitive ones arrive at their destinations faster.



3. (ISPASS'25)
   Dissecting Performance Overheads of Confidential Computing on GPU-based Systems
   Yang Yang, Mohammad Sonji, Adwait Jog
   In the Proceedings of IEEE International Symposium on Performance Analysis of Systems and Software, Ghent, Belgium, May 2025
   [PDF] [Talk@ISPASS] [Talk@CC-Summit] [Artifact Available] [Show Brief] 

   Confidential computing (CC) is a critical technology for protecting data in use. By leveraging encryption and virtual machine (VM) level isolation, CC allows existing code to run without modification while offering confidentiality and integrity guarantees. However, the performance impact of CC in GPU-based systems can be significant. In this work, we present a comprehensive performance evaluation of CC guided by a simple performance model. Specifically, we start by evaluating CUDA applications with a focus on data transfer, memory management, encryption, kernel launch, and kernel execution. We also present a detailed event-level analysis of these applications, revealing that the execution times of kernels that do not use unified virtual memory (UVM) are mostly unaffected, while associated kernel launch overhead and queuing time increase significantly. On the other hand, the execution time of kernels using UVM increases drastically under CC, in addition to other launch and queuing overheads. We also study CNN training and LLM inference to see how CC overhead would affect them. Finally, we consider several optimization techniques, including kernel fusion, overlapping, and quantization, towards addressing the overheads of CC.

1. [24 Fall] TA for CS 6354: Computer Architecture
2. [25 Fall] TA for CS 4444: Introduction to Parallel Computing

1. [Travel Grant] ISPASS'25 ISCA'25

1. [SIG] Co-organizer of Systems-Interest-Group meetings at UVA.
2. [Artifact Evaluation Committee] IISWC'25

1. MQ-Sim Tutorial: [video][slides]
2. Introduction blogs about Intel SGX: [Part I] [Part II]
3. Basic concepts about Confidential Computing (CC): [Whitepapers]
4. Intel AVX-family extension: [Reference]
5. Intel SGX: [Library]
6. Linux memory management: [Docs]
7. Page mapping: [Codes]
8. How to access Linux physical memory (/dev/mem): [Codes] [Blog]
9. How to perform virtual2physical address mapping: [Kernel] [User 1] [User 2]
10. A note on how to debug GPGPU-Sim: [note]
11. CXL paper reading blog series from Zhihu (Chinese) [Link]
12. CXL cache coherence and why from Zhihu (Chinese) [Link]
13. Latency Numbers Every Programmer Should Know [Link1] [Link2] [Link3]
14. System & Arch Conference Deadline: [Link]
15. Resource Collection of Performance Measurement/Analysis [here]
16. LaTeX Symbols: [Link1] [Link2]
17. Instruction latency: [PDF1] [PDF2]

1. Advice collection by Prof. Tao Xie
2. A nice talk by Prof. Lin Zhong