Balazs Gerofi

Dr. Balazs Gerofi


Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

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

Representative Publications

View All Publications
29 January 2022 in ACM Transactions on Storage
Verified
Journal Article
Open access
JLJun Li
Jun Li
XXXiaofei Xu
Xiaofei Xu
ZCZhigang Cai
Zhigang Cai
JLJianwei Liao
Jianwei Liao
KLKenli Li
Kenli Li
YIYutaka Ishikawa
Yutaka Ishikawa
This article proposes a pattern-based prefetching scheme with the support of adaptive cache management, at the flash translation layer of solid-state drives ( SSDs ). It works inside of SSDs and has features of OS dependence and uses transparency. Specifically, it first mines frequent block access patterns that reflect the correlation among the occurred I/O requests. Then, it compares the requests in the current time window with the identified patterns to direct prefetching data into the cache of SSDs. More importantly, to maximize the cache use efficiency, we build a mathematical model to adaptively determine the cache partition on the basis of I/O workload characteristics, for separately buffering the prefetched data and the written data. Experimental results show that our proposal can yield improvements on average read latency by 1.8 %– 36.5 % without noticeably increasing the write latency, in contrast to conventional SSD-inside prefetching schemes.

14 December 2023 in ACM Transactions on Architecture and Code Optimization
Verified
Journal Article
Open access
JDJens Domke
Jens Domke
EVEmil Vatai
Emil Vatai
BGBalazs Gerofi
Balazs Gerofi
YKYuetsu Kodama
Yuetsu Kodama
MWMohamed Wahib
Mohamed Wahib
SMSatoshi Matsuoka
Satoshi Matsuoka
Over the last three decades, innovations in the memory subsystem were primarily targeted at overcoming the data movement bottleneck. In this paper, we focus on a specific market trend in memory technology: 3D-stacked memory and caches. We investigate the impact of extending the on-chip memory capabilities in future HPC-focused processors, particularly by 3D-stacked SRAM. First, we propose a method oblivious to the memory subsystem to gauge the upper-bound in performance improvements when data movement costs are eliminated. Then, using the gem5 simulator, we model two variants of a hypothetical LARge Cache processor (LARC), fabricated in 1.5 nm and enriched with high-capacity 3D-stacked cache. With a volume of experiments involving a broad set of proxy-applications and benchmarks, we aim to reveal how HPC CPU performance will evolve, and conclude an average boost of 9.56× for cache-sensitive HPC applications, on a per-chip basis. Additionally, we exhaustively document our methodological exploration to motivate HPC centers to drive their own technological agenda through enhanced co-design.

1 May 2018
Verified
Conference Paper
Closed access
BGBalazs Gerofi
Balazs Gerofi
RRRolf Riesen
Rolf Riesen
MTMasamichi Takagi
Masamichi Takagi
TBTaisuke Boku
Taisuke Boku
KNKengo Nakajima
Kengo Nakajima
RWRobert W. Wisniewski
Robert W. Wisniewski
Multi-kernels leverage today's multi-core chips to run multiple operating system (OS) kernels, typically a Light Weight Kernel (LWK) and a Linux kernel, simultaneously. The LWK provides high performance and scalability, while the Linux kernel provides compatibility. Multi-kernels show the promise of being able to meet tomorrow's extreme-scale computing needs while providing strong isolation, yielding high performance and scalability needed by classical HPC applications. McKernel and mOS started as independent research initiatives to explore the above potential. Previous work described their design and architecture advantages. This paper deploys the two LWKs and presents results from running them on a 2,048-node system with Intel Xeon Phi processors (KNL) connected by Intel Omni-Path Fabric. We compare the performance of McKernel, mOS, and Linux. Although the two multi-kernel efforts approached the problem from different angles, the results show a median performance improvement of 9% with some applications as high as 280% validating the efficacy of the multi-kernel approach. We provide insight into the performance improvements and discuss the strengths of the two different multi-kernel approaches.

9 September 2023 in ACM Transactions on Embedded Computing Systems
Verified
Journal Article
Open access
ZSZhibing Sha
Zhibing Sha
JWJiaojiao Wu
Jiaojiao Wu
JLJun Li
Jun Li
BGBalazs Gerofi
Balazs Gerofi
ZCZhigang Cai
Zhigang Cai
JLJianwei Liao
Jianwei Liao
Solid-State Drives (SSDs) exhibit different failure characteristics compared to conventional hard disk drives. In particular, the Bit Error Rate (BER) of an SSD increases as it bears more writes. Then, Parity-based Redundant Array of Inexpensive Disks (RAID) arrays composed from SSDs are introduced to address correlated failures. In the RAID-5 implementation, specifically, the process of parity generation (or update) associating with a data stripe, consists of read and write operations to the SSDs. Whenever a new update request comes to the RAID system, the related parity must be also updated and flushed onto the RAID component of SSD. Such frequent parity updates result in poor RAID performance and shorten the life-time of the SSDs. Consequently, a DRAM cache is commonly equipped accompanying with the RAID controller, called the parity cache, and used to buffer the parity chunks that are most frequently updated data, for boosting I/O performance. To better improve the use efficiency of the parity cache, this paper proposes a stripe reconstruction approach to minimize the number of parity updates on SSDs, thus boosting I/O performance of the SSD RAID system. When the currently updated stripe has both cold and hot updated data chunks, it will proactively carry out stripe reconstruction if we can find another matched stripe that also includes cold and hot update data chunks on the complementary RAID components. In the reconstruction process, we first group the cold data chunks of two matched stripes, to build a new stripe and flush the parity chunk on the RAID component. After that, the hot data chunks are organized as a new stripe as well, and its parity chunk is buffered in the parity cache. This results in better cache use efficiency, as it can reduce the number of parity updates on RAID components of SSDs, as well as proactively free up cache space for quickly absorbing subsequent write requests. In addition, the proposed method adjusts the target SSD of write requests based on stripe reconstructions through considering the I/O workload balance of all SSDs. Experimental results show that our proposal can reduce the number of parity chunk updates in SSDs by 2.3% and overall I/O latency by 12.2% on average, compared to state-of-the-art parity cache management techniques.

15 November 2025
Verified
Conference Paper
Open access
YLYonatan Levitt
Yonatan Levitt
RBRichard Barella
Richard Barella
SZSam Zeltner
Sam Zeltner
TMThomas Musta
Thomas Musta
LCLance Cheney
Lance Cheney
BGBalazs Gerofi
Balazs Gerofi

1 July 2022
Verified
Book Chapter
Open access
PDF file
AHAtsushi Hori
Atsushi Hori
KOKaiming Ouyang
Kaiming Ouyang
BGBalazs Gerofi
Balazs Gerofi
YIYutaka Ishikawa
Yutaka Ishikawa
Shared memory mechanisms, e.g., POSIX shmem or XPMEM, are widely used to implement efficient intra-node communication among processes running on the same node. While POSIX shmem allows other processes to access only newly allocated memory, XPMEM allows accessing any existing data and thus enables more efficient communication because the send buffer content can directly be copied to the receive buffer. Recently, the shared address space model has been proposed, where processes on the same node are mapped into the same address space at the time of process creation, allowing processes to access any data in the shared address space. Process-in-Process (PiP) is an implementation of such mechanism. The functionalities of shared memory mechanisms and the shared address space model look very similar – both allow accessing the data of other processes –, however, the shared address space model includes the shared memory model. Their internal mechanisms are also notably different. This paper clarifies the differences between the shared memory and the shared address space models, both qualitatively and quantitatively. This paper is not to showcase applications of the shared address space model, but through minimal modifications to an existing MPI implementation it highlights the basic differences between the two models. The following four MPI configurations are evaluated and compared; 1) POSIX Shmem, 2) XPMEM, 3) PiP-Shmem, where intra-node communication is implemented to utilize POSIX shmem but MPI processes share the same address space, and 4) PiP-XPMEM, where XPMEM functions are implemented by the PiP library (without the need for linking to XPMEM library). Evaluation is done using the Intel MPI benchmark suite and six HPC benchmarks (HPCCG, miniGhost, LULESH2.0, miniMD, miniAMR and mpiGraph). Most notably, mpiGraph performance of PiP-XPMEM outperforms the XPMEM implementation by almost 1.5x. The performance numbers of HPCCG, miniGhost, miniMD, LULESH2.0 running with PiP-Shmem and PiP-XPMEM are comparable with those of POSIX Shmem and XPMEM. PiP is not only a practical implementation of the shared address space model, but it also provides opportunities for developing new optimization techniques, which the paper further elaborates on.