SCAR: Scheduling Multi-Model AI Workloads on Heterogeneous Multi-Chiplet Module Accelerators
By Mohanad Odema, Luke Chen, Hyoukjun Kwon, Mohammad Abdullah Al Faruque
University of California, Irvine, USA
Emerging multi-model workloads with heavy models like recent large language models significantly increased the compute and memory demands on hardware. To address such increasing demands, designing a scalable hardware architecture became a key problem. Among recent solutions, the 2.5D silicon interposer multi-chip module (MCM)-based AI accelerator has been actively explored as a promising scalable solution due to their significant benefits in the low engineering cost and composability. However, previous MCM accelerators are based on homogeneous architectures with fixed dataflow, which encounter major challenges from highly heterogeneous multi-model workloads due to their limited workload adaptivity.
Therefore, in this work, we explore the opportunity in the heterogeneous dataflow MCM AI accelerators. We identify the scheduling of multi-model workload on heterogeneous dataflow MCM AI accelerator is an important and challenging problem due to its significance and scale, which reaches O(1018) scale even for a single model case on 6x6 chiplets. We develop a set of heuristics to navigate the huge scheduling space and codify them into a scheduler with advanced techniques such as inter-chiplet pipelining. Our evaluation on ten multi-model workload scenarios for datacenter multitenancy and AR/VR use-cases has shown the efficacy of our approach, achieving on average 35.3% and 31.4% less energy-delay product (EDP) for the respective applications settings compared to homogeneous baselines.
To read the full article, click here
Related Chiplet
- 12nm EURYTION RFK1 - UCIe SP based Ka-Ku Band Chiplet Transceiver
- Interconnect Chiplet
- Bridglets
- Automotive AI Accelerator
- Direct Chiplet Interface
Related Technical Papers
- Inter-Layer Scheduling Space Exploration for Multi-model Inference on Heterogeneous Chiplets
- Communication Characterization of AI Workloads for Large-scale Multi-chiplet Accelerators
- Multi-Objective Hardware-Mapping Co-Optimisation for Multi-DNN Workloads on Chiplet-based Accelerators
- AIG-CIM: A Scalable Chiplet Module with Tri-Gear Heterogeneous Compute-in-Memory for Diffusion Acceleration
Latest Technical Papers
- Die-Level Transformation of 2D Shuttle Chips into 3D-IC for Advanced Rapid Prototyping using Meta Bonding
- STAMP-2.5D: Structural and Thermal Aware Methodology for Placement in 2.5D Integration
- MCMComm: Hardware-Software Co-Optimization for End-to-End Communication in Multi-Chip-Modules
- FoldedHexaTorus: An Inter-Chiplet Interconnect Topology for Chiplet-based Systems using Organic and Glass Substrates
- ChipletQuake: On-die Digital Impedance Sensing for Chiplet and Interposer Verification