Thermal Comparison between Monolithic and Chiplet ASIC Design
As Moores law approaches the physical limit of the critical size on chip, the need for the use of domain-specific accelerators (DSA) to meet power and performance has been increased. Advanced packaging is used to achieve higher integration densities and performance. However, managing package thermal density is becoming challenging for accelerators and ASIC. In order to understand the thermal density trade off for die disaggregation advanced package solutions (2D, 2.5D, 3D), this paper selects a ASIC die with compute and SRAM as an example. This talk will focus on understanding the change in thermal density of a package from monolithic die to advanced packaging with chiplet die with 2D, 2.5D, and 3D package options and partner with ecosystem to understand the thermal solution (lid vs. lidless). The thermal performance among the different package options as well as the thermal risks and solutions are discussed in this paper.
Related Chiplet
- Optical Chiplet
- Reverse Gearbox Chiplet
- FlexGen Multi-Die Smart Network-on-Chip (NoC) IP
- Ncore Multi-Die Interconnect IP
- Integrated voltage regulator (IVR) chiplet
Related Videos
- Optimizing Chiplet architectures of RISC-V clusters, GPU and DNN using System-Level IP
- Scale-out Chiplet Based Systems: Design, Architecture and Pathfinding
- DAC 2025: Cadence and Its Ambition to Jumpstart the Chiplet Marketplace
- Unpacking 3D IC microarchitecture: Challenges, solutions, and the future of chiplet design
Latest Videos
- The European Chip Ecosystem
- Scaling up AI Compute, Memory and Connectivity
- CURIE: A Cycle-Accurate and Synthesizable UCIe Interface for the Agile Design and Co-Simulation of Chiplets
- Federated Chiplet Design and Simulation
- Custom HBM as a System: Co-designing the Base Die, Die to Die Power, and Package Thermals for AI Accelerators