Smarter, faster, greener AI with 3D IC chiplet advanced packaging

The Electronics and Telecommunications Research Institute (ETRI) is a government research institute focuses on IT and semi-conductor components.

In 2019, the AI NPU processor design group at ETRI’s National AI Research Institute developed the first Korean AI processor, the AB9. An FP16 16-bit floating-point artificial neural network computing processor, the AB9 has been successfully adopted by several commercial companies for their AI computing servers.

Coming off that success, the AI NPU processor design team sought to push the boundaries of AI even further by employing a chiplet-based design strategy to create their next-generation of AI processors – the ABS1 neural processing unit (NPU). ETRI wanted to take advantage of 3D IC packaging technology’s capability to dramatically increase bandwidth and performance while reducing the power and cost of their highly-complex, high-performance ASIC designs.

ETRI’s drive to innovation operates within a larger context in which the semiconductor industry faces a considerable technology inflection point as higher cost, lower yield, and reticle size limitations compel the demand for viable alternatives to traditional monolithic system-on-chip (SoC) architec- tures. This is driving an emerging trend to disaggregate device functions into solid, fabricated IP blocks implemented at the appropriate process node and technology, otherwise known as chiplets. Multiple chiplets, memory, and an optional, custom ASIC device can be mounted and interconnected into a single package using high-speed, high-bandwidth chiplet-to-chiplet interfaces, such as those using the Bunch of Wires (BoW) and Universal Chiplet Interconnect Express (UCIe) specifications. The resulting 3D IC heterogeneously integrated assembly can deliver greater performance at a reduced cost and higher yield, with only a slightly larger footprint than a traditional monolithic SoC.

These chiplet architectures provide the increased compute power, memory access, and storage that ETRI’s AI processors require. Employing this truly groundbreaking chiplet technology, ETRI is equipped to create next-generation, ultra-performance NPUs.

The ABS1 Neural Processor Chiplet

The ABS1 NPU chiplet is designed for a one quadrillion floating-point operations per second (petaflops) AI processor. As one petaflops equals 1000 teraflops, this is equivalent to the performance of 10,000 application processors crammed into a single AI processor.

ETRI also wanted the high-speed design to boast advanced energy efficiency and higher AI performance. To achieve these goals, the ABS1 integrates two AI NPU chiplets with eight HBM3 memory stacks. Ten of these dies are integrated on top of a large-scale redistribution-layer (RDL) interposer. The result will be the first petaflops AI processor that has a very high energy efficiency.

The ABS1 neural processing unit consists of two AI NPU chiplets with eight HBM3 memory stacks.

Thus, the ETRI team faced the highly challenging task of integrating ten chiplets side-by-side, while maintaining lowered warpage imbalance, higher thermal stability, and large-scale interposer manufacturability, along with other requirements. To help them meet these requirements, ETRI partnered with Amkor Technology Korea as their OSAT and design partner for the ABS1. Amkor‘s RDL interposer lends itself to large-scale interposer manufacturing, supports low cost chiplet integration, and meet ETRI’s specified requirements for manufacturability.

Ideal Partners Bring Ideal Tools

Due to the long, successful relationship between Amkor and Siemens and the quality and capabilities of the Siemens’ advanced packaging solutions, Amkor and ETRI selected Xpedition™ Package Designer (XPD), Xpedition™ Substrate Integrator (xSI), and Calibre® 3DSTACK, all from Siemens DISW, to help them establish a systematic approach to designing their chiplet-based AI NPU.

Amkor Korea utilized XPD for its excellent design capacity and performance, which they needed to handle their high-density advanced package designs, such as the RDL interposer. Further, for the design of the RDL interposer itself, XPD contains special functionalities that reduce design cycle times.

Amkor had originally adopted XPD for their Silicon Wafer Integrated Fan-out Technology (SWIFT®), which enables the creation of advanced 3D structures, such as the RDL interposers and advanced package designs.

Other features that XPD had in its favor were its “physical reuse circuit” function, which plays an especially important role in reducing total design cycle time, and XPD’s true GDS direct-export functionality. Its high-performance dynamic plane shapes capability was a big part of why they chose XPD for RDL interposer and large size package substrate design.

As well, Amkor has a great deal of experience using xSI for bump connection visualization and optimization during multi-die integration. xSI has proven to show superior design capacity even when dealing with bump counts in the hundreds of thousands.

xSI is packed with features that enable Amkor to integrate 3DSTACK level design structures for both the RDL interposer and the package substrate. xSI can handily perform the bump connection optimizations required to achieve a better quality RDL interposer design, which indeed was successfully done by xSI on the interposer. Amkor has been successfully running Calibre 3DSTACK verification for high-density advanced package designs, so they knew it was a good choice for ETRI’s AI NPU processor project.

The ABS1 NPU being designed in Xpedition Package Designer by the Amkor advanced IC packaging design team (source: Amkor Technology).

Applying the Siemens’ 3D IC Solution

ETRI initially focused on constructing a seamless VLSI CAD flow for advanced packaging, using xSI to describe the top-level chiplet architecture of the ABS1 NPU composed of multiple dies, chiplets, and HBM3 stacks. By providing this description of the top chiplet architecture, xSI then allowed them to seamlessly move on to the architecture simulation tool environment, Xpedition Package Designer (xPD), for interposer and package substrate design.

Youngsu Kwon, Ph.D., the principal researcher in charge of the ETRI AI SoC Research Division, recalls, “We successfully integrated a seamless chiplet VLSI CAD design flow with xSI and developed a systematic approach to designing chiplet-based chips using xSI.”

When they started RDL interposer design, netlist information about the NPU die and the HBM die was only in Verilog. So they needed Siemens to provide tools that could visualize bump-to-bump connections and micro-bumps-to-C4-bumps, including the package BGA. Indeed, xSI is capable of not only connection visualization but also optimizing connections. ETRI was very satisfied with how xSI performed these functions.

The final ABS1 device is the first petaflops AI processor with very high energy efficiency (source: Amkor Technology).

Amkor also needed to verify connectivity-checks of the multi-die architecture, the RDL interposer, and the package substrate in ETRI’s hierarchical, high-density advanced packages. All these checks were successfully performed using the xSI-Calibre 3DSTACK verification flow, which generated source level netlist and connectivity-checks between the multi-die structures, the interposer, and the package substrate.

“Amkor has successfully provided package designs from RDL interposer to substrate. We also verified the connectivity and performed DFM with xSI and Calibre 3DSTACK. Siemens IC packaging design tools helped us provide a fast and high-quality design service to our customers even with large package and chiplet package structures,” JaeBeom Shim, the Amkor package design manager in Korea reports.

At the end of the design work, ETRI successfully completed 3DSTACK verification sign-off for both the interposer and the 3D IC package.

Laying the Foundation for 3D IC Architectural Designers

Youngsu Kwon looks beyond this successful collaboration with great optimism.

“The Siemens’ initiative to provide a CAD design environment for interposer and package substrate for advanced packaging has laid a foundational work for architecture designers. At ETRI, we trust Siemens to further improve their xSI environment to better fit the design criteria of chiplets, which we consider to be the next-generation semiconductor architecture.”

“We are pleased to work closely with Siemens, which continues to empower Amkor EDA systems by offering more features that support our seamless advanced packaging design flow. We look forward to collaborating with Siemens to make an even better xSI and Calibre 3DSTACK environment that aligns with our customers, focusing on RDL interposer and substrate designs in advanced packages, including chiplets,” TaeKyeong Hwang, Ph.D., vice president and product development divivion manager at Amkor Korea concludes.