Zhengrong Wang

Zhengrong Wang

Auto-generated profile, based on publicly available information

Zhengrong Wang

University of California, Los Angeles, CA, USA

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 October 2023
Verified
Conference Paper
Open access
ZWZhengrong Wang
Zhengrong Wang
CLChristopher Liu
Christopher Liu
NBNathan Beckmann
Nathan Beckmann
TNTony Nowatzki
Tony Nowatzki
To mitigate the data movement bottleneck on large multicore systems, the near-data computing paradigm (NDC) offloads computation to where the data resides on-chip. The benefit of NDC heavily depends on spatial affinity, where all relevant data are in the same location, e.g. same cache bank. However, existing NDC works lack a general and systematic solution: they either ignore the problem and abort NDC when there is no spatial affinity, or rely on error-prone manual data placement. Our insight is that the essential affinity relationship, i.e. data A should be close to data B, is orthogonal to microarchitecture details and input sizes. By co-optimizing the data structure and capturing this general affinity information in the data allocation interface, the allocator can automatically optimize for data affinity and load balance to make NDC computations truly near data. With this insight, we propose affinity alloc, a general framework to optimize data layout for near-data computing. It comprises an extended allocator runtime, co-optimized data structures, and lightweight extensions to the OS and microarchitecture. Evaluated on parallel workloads across broad domains, affinity alloc achieves 2.26 × speedup and 1.76 × energy efficiency over a state-of-the-art near-data computing technique with 72% traffic reduction.

31 August 2022 in IEEE Computer Architecture Letters
Verified
Journal Article
Closed access
ZWZhengrong Wang
Zhengrong Wang
CLChristopher Liu
Christopher Liu
TNTony Nowatzki
Tony Nowatzki
Although in-memory computing is promising to alleviate the data movement bottlenecks by parallelizing computation across memory bitlines, key challenges from its unique execution model remain unsolved: Automatically parallelizing sequential programs; Dynamically managing and aligning data in transposed layout required for bit-serial logic; Mixing in/near-memory computing. These challenges should be solved transparently to maintain portability without exposing hardware details to programmers. In this work, we introduce a novel intermediate representation – tensor dataflow graph (tDFG) – with tensor nodes representing the spatially unrolled data across bitlines, and explicit move nodes to align operands in the same bitline, which helps the compiler optimize for massive parallelism and data layout. To maintain transparency and portability, we directly embed tDFG in the ISA, which is lowered into bit-serial operations at runtime to hide the hardware details. Evaluated on cycle-accurate simulator across various data-processing workloads, our approach achieves 4.5× speedup and 52% traffic reduction over a state-of-the-art near-memory computing technique.

20 November 2024 in ACM Transactions on Architecture and Code Optimization
Verified
Journal Article
Open access
SMSiyuan Ma
Siyuan Ma
KMKaustubh Mhatre
Kaustubh Mhatre
JWJian Weng
Jian Weng
BHBagus Hanindhito
Bagus Hanindhito
ZWZhengrong Wang
Zhengrong Wang
AAAman Arora
Aman Arora
Bit-serial Processing-In-Memory (PIM) is an attractive paradigm for accelerator architectures, for parallel workloads such as Deep Learning (DL), because of its capability to achieve massive data parallelism at a low area overhead and provide orders-of-magnitude data movement savings by moving computational resources closer to the data. While many PIM architectures have been proposed, improvements are needed in communicating intermediate results to consumer kernels, for communication between tiles at scale, for reduction operations, and for efficiently performing bit-serial operations with constants. We present PIMSAB, a scalable architecture that provides a spatially aware communication network for efficient intra-tile and inter-tile data movement and provides efficient computation support for generally inefficient bit-serial compute patterns. Our architecture consists of a massive hierarchical array of compute-enabled SRAMs (CRAMs), which is codesigned with a compiler to achieve high utilization. The key novelties of our architecture are (1) in providing efficient support for spatially aware communication by providing local H-tree network for reductions, by adding explicit hardware for shuffling operands, and by deploying systolic broadcasting, as well as (2) by taking advantage of the divisible nature of bit-serial computations through adaptive precision and efficient handling of constant operations. These innovations are integrated into a tensor expressions-based programming framework (including a compiler for easy programmability) that enables simple programmer control of optimizations for mapping programs into massively parallel binaries for millions of PIM processing elements. When compared against a similarly provisioned modern Tensor Core GPU (NVIDIA A100), across common DL kernels and end-to-end DL networks (Resnet18 and BERT), PIMSAB outperforms the GPU by 4.80×, and reduces energy by 3.76×. We compare PIMSAB with similarly provisioned state-of-the-art SRAM PIM (Duality Cache) and DRAM PIM (SIMDRAM), and observe a speedup of 3.7× and 3.88×, respectively.

14 November 2024 in IEEE Computer Architecture Letters
Verified
Journal Article
Closed access
SZShizhuo Zhu
Shizhuo Zhu
ISIllia Shkirko
Illia Shkirko
JLJacob Levinson
Jacob Levinson
ZWZhengrong Wang
Zhengrong Wang
TNTony Nowatzki
Tony Nowatzki
Communication is a critical bottleneck for GPUs, manifesting as energy and performance overheads due to network-on-chip (NoC) delay and congestion. While many algorithms exhibit locality among thread blocks and accessed data, modern GPUs lack the interface to exploit this locality: GPU thread blocks are mapped to cores obliviously. In this work, we explore a simple extension to the conventional GPU programming interface to enable control over the spatial placement of data and threads, yielding new opportunities for aggressive locality optimizations within a GPU kernel. Across 7 workloads that can take advantage of these optimizations, for a 32 (or 128) SM GPU: we achieve a 1.28× (1.54×) speedup and 35% (44%) reduction in NoC traffic, compared to baseline non-spatial GPUs.

25 April 2025 in IEEE Computer Architecture Letters
Verified
Journal Article
Closed access
AAArteen Abrishami
Arteen Abrishami
ZWZhengrong Wang
Zhengrong Wang
TNTony Nowatzki
Tony Nowatzki
Vendors are increasingly adopting chiplet-based designs to manage cost for large-scale multi-cores. While near-data computing, a paradigm involving offloading computation near where data is located in memory, has been studied in the context of monolithic chip designs – its applications to chiplets remain unexplored. In this letter, we explore how the paradigm extends to chiplets in a system where computation is offloaded to accelerators collocated within the last-level-cache structure. We explore both shared and private last-level-cache designs across a variety of different workloads, both large-scale graph computations and more regular-access workloads, in order to understand how to optimize the cache and topology design for near-data workloads. We find that with a mesh chiplet architecture with shared last-level-cache (LLC), near-data optimization can achieve an 8.70× speedup on graph workloads, providing an even greater benefit than in traditional systems.