Reducing the barrier to ASICs with chiplets
To unlock the true power of chiplets, several hard problems need to be solved
By Designing Electronics North America
For the last 50 years, the semiconductor industry has been all about monolithic integration, doubling the number of transistors on a single chip every two years. The results have been nothing short of miraculous, with electronics getting both cheaper and better year over year.
However, as we approach the physical (atomic) limits of device scaling, we now need to find new ways of improving the energy and cost efficiency of electronics. While physical scaling may be saturating, fortunately there remains plenty of room at the bottom in terms of cost, energy, and time. Hyper- specialized circuits have demonstrated 100 to 1000x Size-Weight-Area-Power-Cost (SWAP-C) advantages over general purpose processors. Similarly, specialized devices and materials can offer 10 to 100x SWAP-C advantages over CMOS logic processes.
These two promising pathways are currently blocked by two fundamental economic challenges:
- Designing complex specialized circuits in the form of application- specific integrated circuits (ASICs) can cost over $100M and take years to complete.
- Integrating new specialized materials and devices into advanced CMOS processes is economically impractical.
An elegant solution to both of these challenges is the introduction of “chiplets,” which are tiny integrated circuits (ICs) with partial functionality designed to be combined with other chiplets within a single package to create a complete chip, which is also known as a multi-die system.
If implemented correctly, chiplet technology promises to enable custom silicon development with the design cost of PCBs and the SWAP-C of ASICs.Unfortunately, the benefits of chiplet technology have to date been accessible to only a small number of giant semiconductor vendors who leverage chiplets to improve production yields and to reduce NRE costs for multi-SKU product families.
To unlock the true power of chiplets, several hard problems need to be solved as follows
To read the full article, click here
Related Chiplet
- Interconnect Chiplet
- 12nm EURYTION RFK1 - UCIe SP based Ka-Ku Band Chiplet Transceiver
- Bridglets
- Automotive AI Accelerator
- Direct Chiplet Interface
Related News
- DreamBig Semiconductor Announces Partnership with Samsung Foundry to Launch Chiplets for World Leading MARS Chiplet Platform on 4nm FinFET Process Technology Featuring 3D HBM Integration to Solve Scale-up and Scale-out Limitations of AI for the Masses
- Biden-Harris Administration Announces CHIPS Incentives Award with Amkor Technology to Bring End-to-End Chip Production to the U.S.
- Breaking the Memory Wall: How d-Matrix Is Redefining AI Inference with Chiplets
- CDimension Launches to Tackle Hardware Bottlenecks from the Ground Up, Beginning with Commercial 2D Materials Release
Latest News
- Arteris Joins UALink Consortium to Accelerate High-Performance AI Networks Scale Up
- Athos Silicon Chief mSoC™ Architect Francois Piednoel to Present the IEEE World Technology Summit 2025 in Berlin
- Marvell Unveils Industry’s First 64 Gbps/wire Bi-Directional Die-to-Die Interface IP in 2nm to Power Next Generation XPUs
- Xscape Photonics and Tower Semiconductor Unveil the Industry’s First Optically Pumped On-Chip Multi-Wavelength Laser Platform for AI Datacenter Fabrics
- Electrical Engineering and Computer Science Department Chair Alex K. Jones and Professor Bryan Kim Receive NSF Grant to Develop Energy-Efficient Chiplets for Data Centers