Binzhang Fu
Research Subjects
Research Interests
Binzhang Fu has a total of 113 co-authors. Here are the top co-authors by number of shared publications:
- Dennis Cai
Alibaba Cloud, Beijing, China
6 shared publications - Jiaqi Gao
Alibaba Cloud, Beijing, China
6 shared publications - Ennan Zhai
Alibaba Group, Hangzhou, China
5 shared publications - Jiamin Cao
Alibaba Group, Beijing, China
5 shared publications - Jianbo Dong
Alibaba Cloud, Beijing, China
5 shared publications - Kun Qian
Alibaba Cloud, Hangzhou, China
3 shared publications
Latest Publications
- Open Access
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 368-395
EPIC: Abstraction and Polymorphism of In-Network Collectives on Ethernet
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 368-395
Open AccessEPIC: Abstraction and Polymorphism of In-Network Collectives on Ethernet
- Open Access
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 550-563
Balancing and Beyond: Communication-Centric Optimizations in Expert Parallelism
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 550-563
Open AccessBalancing and Beyond: Communication-Centric Optimizations in Expert Parallelism
- Open Access
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 822-836
Anytest: Localizing the Root Cause of Hardware Transport Performance Anomalies
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 822-836
Open AccessAnytest: Localizing the Root Cause of Hardware Transport Performance Anomalies
- Open Access
- 27 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 645-662
SyCCL: Exploiting Symmetry for Efficient Collective Communication Scheduling
- 27 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 645-662
Open AccessSyCCL: Exploiting Symmetry for Efficient Collective Communication Scheduling
- Open Access
- 27 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 453-466
Alibaba Stellar: A New Generation RDMA Network for Cloud AI
- 27 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 453-466
Open AccessAlibaba Stellar: A New Generation RDMA Network for Cloud AI
- Top Cited PapersOpen Access
- 4 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-15
Crux: GPU-Efficient Communication Scheduling for Deep Learning Training
Deep learning training (DLT), e.g., large language model (LLM) training, has become one of the most important services in multitenant cloud computing. By deeply studying in-production DLT jobs, we observed that communication contention among different DLT jobs seriously influences the overall GPU computation utilization, resulting in the low efficiency of the training cluster. In this paper, we present Crux, a communication scheduler that aims to maximize GPU computation utilization by mitigating the communication contention among DLT jobs. Maximizing GPU computation utilization for DLT, nevertheless, is NP-Complete; thus, we formulate and prove a novel theorem to approach this goal by GPU intensity-aware communication scheduling. Then, we propose an approach that prioritizes the DLT flows with high GPU computation intensity, reducing potential communication contention. Our 96-GPU testbed experiments show that Crux improves 8.3% to 14.8% GPU computation utilization. The large-scale production trace-based simulation further shows that Crux increases GPU computation utilization by up to 23% compared with alternatives including Sincronia, TACCL, and CASSINI.
- 4 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 1-15
Top Cited PapersOpen AccessCrux: GPU-Efficient Communication Scheduling for Deep Learning Training
Deep learning training (DLT), e.g., large language model (LLM) training, has become one of the most important services in multitenant cloud computing. By deeply studying in-production DLT jobs, we observed that communication contention among different DLT jobs seriously influences the overall GPU computation utilization, resulting in the low efficiency of the training cluster. In this paper, we present Crux, a communication scheduler that aims to maximize GPU computation utilization by mitigating the communication contention among DLT jobs. Maximizing GPU computation utilization for DLT, nevertheless, is NP-Complete; thus, we formulate and prove a novel theorem to approach this goal by GPU intensity-aware communication scheduling. Then, we propose an approach that prioritizes the DLT flows with high GPU computation intensity, reducing potential communication contention. Our 96-GPU testbed experiments show that Crux improves 8.3% to 14.8% GPU computation utilization. The large-scale production trace-based simulation further shows that Crux increases GPU computation utilization by up to 23% compared with alternatives including Sincronia, TACCL, and CASSINI.
- Top Cited PapersOpen Access
- 4 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 691-706
Alibaba HPN: A Data Center Network for Large Language Model Training
This paper presents HPN, Alibaba Cloud's data center network for large language model (LLM) training. Due to the differences between LLMs and general cloud computing (e.g., in terms of traffic patterns and fault tolerance), traditional data center networks are not well-suited for LLM training. LLM training produces a small number of periodic, bursty flows (e.g., 400Gbps) on each host. This characteristic of LLM training predisposes Equal-Cost Multi-Path (ECMP) to hash polarization, causing issues such as uneven traffic distribution. HPN introduces a 2-tier, dual-plane architecture capable of interconnecting 15K GPUs within one Pod, typically accommodated by the traditional 3-tier Clos architecture. Such a new architecture design not only avoids hash polarization but also greatly reduces the search space for path selection. Another challenge in LLM training is that its requirement for GPUs to complete iterations in synchronization makes it more sensitive to singlepoint failure (typically occurring on ToR). HPN proposes a new dual-ToR design to replace the single-ToR in traditional data center networks. HPN has been deployed in our production for more than eight months. We share our experience in designing, and building HPN, as well as the operational lessons of HPN in production.
- 4 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 691-706
Top Cited PapersOpen AccessAlibaba HPN: A Data Center Network for Large Language Model Training
This paper presents HPN, Alibaba Cloud's data center network for large language model (LLM) training. Due to the differences between LLMs and general cloud computing (e.g., in terms of traffic patterns and fault tolerance), traditional data center networks are not well-suited for LLM training. LLM training produces a small number of periodic, bursty flows (e.g., 400Gbps) on each host. This characteristic of LLM training predisposes Equal-Cost Multi-Path (ECMP) to hash polarization, causing issues such as uneven traffic distribution. HPN introduces a 2-tier, dual-plane architecture capable of interconnecting 15K GPUs within one Pod, typically accommodated by the traditional 3-tier Clos architecture. Such a new architecture design not only avoids hash polarization but also greatly reduces the search space for path selection. Another challenge in LLM training is that its requirement for GPUs to complete iterations in synchronization makes it more sensitive to singlepoint failure (typically occurring on ToR). HPN proposes a new dual-ToR design to replace the single-ToR in traditional data center networks. HPN has been deployed in our production for more than eight months. We share our experience in designing, and building HPN, as well as the operational lessons of HPN in production.
- Open Access
- 17 August 2019
- Published by Association for Computing Machinery (ACM)
- p. 57-63
Towards Stateless RNIC for Data Center Networks
- 17 August 2019
- Published by Association for Computing Machinery (ACM)
- p. 57-63
Open AccessTowards Stateless RNIC for Data Center Networks