FinFET ASICs

Taming the Complexity of FinFET ASIC Production

FinFET ASIC design and manufacturing involves the complex interaction of foundry process technology, IP design and integration, EDA tools and design methodologies, physical design, design for test, package design and product engineering. We focus on optimizing for the combination of power, performance, and area (PPA) appropriate for your device’s specific targets.

  • We have access to the leading foundry process technologies to 7/14/16nm through our partnerships with the world’s leading-edge foundries.
  • Through our application-aware IP platforms, we meet your specific needs, be they targeted for high performance or low power. Our IP platforms include:
    • For networking applications:
      • 56G/112G long-reach SerDes
      • High-speed single-port ternary CAM (SP TCAM) compiler
      • High-speed single-port fast cache (FC) compiler
      • High-speed single-port (SP) SRAM compiler
      • High-speed dual-port (DP) SRAM compiler
      • High-speed 2-port asynchronous register file (2PARF) compiler
      • High-speed pseudo 2-port (P2P) SRAM compiler
      • Ultra-high-density (UHD) pseudo 2-port (P2P) SRAM compiler
      • High-speed pseudo 4-port (P4P) SRAM compiler
      • High-speed pseudo quad-port (PQP) SRAM compiler
      • 1024-bit HBM2 PHY
      • High-bandwidth interconnect (HBI)
      • 1.8V oxide 1.8V LVDS I/O library
      • 1.8V oxide 1.8V/2.5V/3.3V general-purpose I/O library
    • For AI applications:
      • Configurable multiply-accumulate (MAC) blocks
      • Single-port SRAM
      • Pseudo two-port and pseudo four-port SRAM
      • Ternary content-addressable memory (TCAM)
      • Pitch match memory
      • GIGA memory
      • WAZPS (word all zero power saving) memory
      • Transpose memory
      • Re-mapper – low power cross-bar
      • Convolution engine
      • 56G SerDes
      • HBM2 PHY
  • Our unique PPA-optimized design methodology is targeted to provide the best power, performance and area for your complex ASIC.

Rich design experience in FinFET ASICs

Through early R&D, our foundry partnerships and our cutting-edge customer base, we have been working in FinFET-class custom ICs and IP from the beginning and have emerged as a leader in the technologies required to design and manufacture a custom FinFET chip: 2.5D, HBM2, TCAM and FinFET-class IP platforms, as demonstrated by our recent tapeout for production of a 2.5D/HBM2 customer ASIC. Our FinFET ASIC experience includes:

  • Large die at high performance (>1GHz), optimized for power
  • Up to reticle-sized devices and stitched interposers
  • High embedded SRAM content: 800Mb, custom SRAM, TCAM
  • 60+ designs with SerDes integration
  • Complex monolithic & 2.5D packaging

High-bandwidth, high-performance customer products enabled by eSilicon FinFET-class ASIC designs and IP

  • 100G/200G/400G Ethernet enterprise switches
  • Multi-terabit-class data center switches
  • Wireless backhaul GSN and NodeB controllers
  • Cognitive deep learning server accelerators
  • Edge aggregation and core service provider routers
  • Converged Ethernet networking adapters 10G/100GE
  • 100G metro and long-haul optical transport
  • Network processor for 5G market

eSilicon FinFET-class IP driving high-volume, high-performance ASICs

Market Tech IP Needs Optimization ASIC Benefit from eSilicon IP Differentiation Memory Subsystem Size
Networking 7nm SerDes, TCAM, HBM2,
fast cache, dense multi-port SRAM with high throughput
Ultra-high speed (UHS) memory optimization Increased speed by 20% and 2.5Gbps HBM2 1.19Gb
Networking 14nm TCAM, HBM2, fast cache, dense multi-port SRAM with high throughput Custom high-speed pseudo 2-Port (P2P) SRAM instances and asynchronous RFs Increase SRAM density by 14% and enabling asynchronous architecture porting and 2Gbps HBM2 828Mb
Networking 14nm High-bandwidth interconnect (HBI), UHS P2P, UHS TCAM and fast cache Custom HBI 16% reduction in unit price and  25% improvement in yield with 2.5Gbps+ HBI 790Mb
Networking 14nm 2R/2W and 4R/4W RFs, HBM2 Custom 4R/4W instances Increased data throughput by 30% 856Mb
Machine Learning 14nm HBM2, dense multi-port SRAM with high throughput Custom pseudo 2-port SRAM instances Increased storage capacity by 10% while saving  40% power ~2 Gb
Networking 16nm TCAM, HBM2, dense SP SRAM, dense multi-port SRAM with high throughput Ultra-high speed memory optimization Increased speed by 20% with increased density 1.22Gb
Machine Learning 28nm TCAM, HBM2, P2P Custom pseudo 2-port SRAM instances 2Gbps HBM2 and
1.0GHz TCAM on 28nm
414Mb

Please contact us at sales@esilicon.com to find out more.