Interconnection Networks¶

Network-on-Chip¶

Challenge: The bandwidth limitations and low communication efficiency faced by traditional bus architectures in on-chip multi-core & many-core systems.

Solution: NoC provides a flexible and high-bandwidth communication infrastructure for heterogeneous chiplets; enables efficient data movement and processing in heterogeneous many-core architectures.

Wafer-Scale¶

Year	Venue	Authors	Title	Tags	P	E	N
2024	SC	THU	Switch-Less Dragonfly on Wafers: A Scalable Interconnection Architecture based on Wafer-Scale Integration	four-level topology structure; minimal routing algorithm on dragonfly for VC vumber reduction
2024	TCAS	SYSU	CINOC: Computing in Network-On-Chip With Tiled Many-Core Architectures for Large-Scale General Matrix Multiplications	computable input buffers; thread execution free from fine-grained instruction control; data-aware thread execution
2025	ISCA	THU	PD Constraint-aware Physical/Logical Topology Co-Design for Network on Wafer	mesh-switch physical topology; dual-granularity logical topology	4	3	2
2025	ISCA	THU	WSC-LLM: Efficient LLM Service and Architecture Co-exploration for Wafer-scale Chips	optimal resource partition algorithm; optimal KV cache placement algorithm	4	3	2
2025	ISCA	Georgia Tech‌	FRED: A Wafer-scale Fabric for 3D Parallel DNN Training	microswitch based distributed on-wafer topology; conflict-free collective routing algorithm; congestion-aware device placement policy	3	3	2

Topology¶

Challenge: Current NoC topologies are often rigid and not adaptable to the specific needs of heterogeneous chiplets.

Year	Venue	Authors	Title	Tags
2021	HPCA	GWU	Adapt-NoC: A Flexible Network-on-Chip Design for Heterogeneous Manycore Architectures	mux based adaptable router architecture; adaptable link design; reinforcement learning based subNoC optimization algorithm
2022	HPCA	Huawei	Application Defined On-chip Networks for Heterogeneous Chiplets: An Implementation Perspective	bufferless multi-ring NoC design; application-architecture-physical co-design method; architecture expressiveness; deadlock resolution SWAP mechanism
2024	MICRO	THU	Ring Road: A Scalable Polar-Coordinate-based 2D Network-on-Chip Architecture	Ring Road topology based on isolated cycles and trees; polar coordinate DOR(dimension-order-routing); inter/intra-chip decouple routing algorithm
2024	arXiv	WSU	Atleus: Accelerating Transformers on the Edge Enabled by 3D Heterogeneous Manycore Architectures	heterogeneous 3D NoC; pipeline design across heterogeneous resources; crossbar-wise quantization
2024	ISLPED	WSU	HeTraX: Energy Efficient 3D Heterogeneous Manycore Architecture for Transformer Acceleration	3D integration; distinct planar tiers where each tier is tailor-made for either MHA or the FF network; alleviate memory bottlenecks while preventing frequent rewrites on ReRAM crossbars

Interconnect¶

Year	Venue	Authors	Title	Tags
2012	SIGCOMM	CMU	On-Chip Networks from a Networking Perspective: Congestion and Scalability in Many-Core Interconnects	congestion control mechanism for bufferless NoC; interval-based congestion control algorithm; simple injection throttling algorithm
2023	ICCAD	UCF	ARIES: Accelerating Distributed Training in Chiplet-based Systems via Flexible Interconnects	directional bypassing link; ARIES link with transistor; ARIES all-reduce optimization algorithm
2023	MICRO	THU	Heterogeneous Die-to-Die Interfaces: Enabling More Flexible Chiplet Interconnection Systems	heterogeneous interface hetero-PHY and hetero-channel; hetero-channel routing algorithm; application-aware scheduling

Processing on NoC¶

Challenge: The need for efficient data processing and computation on-chip to reduce data movement and improve performance.

Year	Venue	Authors	Title	Tags
2017	ISVLSI	Ruhr-Universität Bochum	Data Stream Processing in Network-on-Chip	data stream processing unit(DSPU); operation mode based DSPU programming framework
2019	HPCA	TAMU	Active-Routing: Compute on the Way for Near-Data Processing	active-routing tree; vector processing in cache block for regular access pattern; data prefetch for irregular access pattern
2020	HPCA	Drexel University	SnackNoC: Processing in the Communication Layer	communication fabric quantification; central packet manager for instruction flit; router compute unit as dataflow pe

Traffic Control¶

Challenge: The need for efficient traffic control to manage network traffic and reduce congestion and power consumption.

Year	Venue	Authors	Title	Tags	P	E	N
2017	ISCA	TAMU	APPROX-NoC: A Data Approximation Framework for Network-On-Chip Architectures	value approximate technique VAXX; encoder/decoder module pair for data compression; approximate value compute logic
2017	ICCD	HIT	ABDTR: Approximation–Based Dynamic Traffic Regulation for Networks–on–Chip Systems	approximate computing based dynamic traffic regulation technique; lightweight design including controller, throttler and approximater
2019	DATE	SCUT	ACDC: An Accuracy- and Congestion-aware Dynamic Traffic Control Method for Networks-on-Chip	quality loss and network congestion modeling; autoregressive model based flow prediction method
2025	arXiv	NTU	Learning Cache Coherence Traffic for NoC Routing Design	cache coherence traffic analyzer; DRL based topology selection and routing design	2	3	2

Fault-Tolerant Communication¶

Year	Venue	Authors	Title	Tags
2014	VLSI	ICT	ZoneDefense: A Fault-Tolerant Routing for 2-D Meshes Without Virtual Channels	fault chains based faulty blocks construction; floor/ceiling rule based defense zone forming; L/F chain routing
2017	TPDS	NTU	Path-Diversity-Aware Fault-Tolerant Routing Algorithm for Network-on-Chip Systems	path diversity analysis; fault-location-based path diversity; PDA-FTR algorithm
2019	DATE	UMich	SiPterposer: A Fault-Tolerant Substrate for Flexible System-in-Package Design	blowing based customized topology; lightweight ECC module based defect tolerance
2022	DATE	Colorado State University	DeFT: A Deadlock-Free and Fault-Tolerant Routing Algorithm for 2.5D Chiplet Networks	virtual network based deadlock freedom; congestion-aware vertical link selection

Router¶

Year	Venue	Authors	Title	Tags
2016	HPCA	KTH	DVFS for NoCs in CMPs: A Thread Voting Approach	thread voting based DVFS machenism; pre-defined region-based V/F adjustment algorithm
2022	HPCA	Chalmers	FastTrackNoC: A NoC with FastTrack Router Datapaths	non-turning hops; direct FastTrack flit path; zero-load latency analysis
2022	HPCA	UToronto	Stay in your Lane: A NoC with Low-overhead Multi-packet Bypassing	FastFlow flow controll method; time-division-multiplexed (TDM) based non-overlapping FastPass-lanes; FastPass for throughput enhancement
2023	HPCA	THU	A Scalable Methodology for Designing Efficient Interconnection Network of Chiplets	interface grouping; hypercube construction algorithm; deadlock-free adaptive routing algorithm; safe/unsafe flow control; network interleaving method
2025	arXiv	SJTU	StreamGrid: Streaming Point Cloud Analytics via Compulsory Splitting and Deterministic Termination	compulsory splitting for reducing on-chip buffer size; deterministic termination for regularizing non-deterministic operations; line buffer optimization for point cloud pipelines; ILP-based buffer size minimization

Remote Procedure Call¶

Year	Venue	Authors	Title	Tags	P	E	N
2023	arXiv	ICT	CXL over Ethernet: A Novel FPGA-based Memory Disaggregation Design in Data Centers	combining CXL and Ethernet for low-latency remote memory access; FPGA-based prototype with cache optimization; switch-independent congestion control algorithm; native memory semantics for transparent access
2024	arXiv	UCSD	Telepathic Datacenters: Fast RPCs using Shared CXL Memory	pointer-passing RPC over CXL; MPK-based sandboxing for RPC safety; Seal/Release mechanism for RPC safety; RDMA fallback for RPC scalability; Lease/Quota shared memory management	4	3	3

RDMA¶

Year	Venue	Authors	Title	Tags	P	E	N
2024	arXiv	UCR	GPUVM: GPU-driven Unified Virtual Memory	GPUVM architecture for on-demand paging; RDMA-capable NIC for GPU memory management; GPU thread-based memory management and page migration; reuse-oriented paged memory for efficient eviction; high-level programming abstraction for GPU memory extension