Andes Joins FDXcelerator™ Program, an Ideal Platform for IoT, Mainstream Mobile, RF Connectivity and Networking Applications
Hsinchu, Taiwan. – October 24, 2017 – Andes Technology Corporation (TWSE: 6533) and GLOBALFOUNDRIES (GF) today jointly announced that Andes’ 32-bit CPU IP cores have been implemented on GF’s 22nm FD-SOI (22FDX®) technology. GF’s 22FDX technology offers the optimum combination of performance, power consumption and cost for IoT, mainstream mobile, RF connectivity and networking applications. GF’s 22FDX is a natural CPU IP solution for Andes Technology, with its focus on small and efficient processors.
“Our 32-bit power-efficient small-footprint cores have been wildly successful for deeply-embedded applications, especially in the growing global market. We are excited to be part of GF’s FDXcelerator Program with silicon proven AndesCore™ on GF’s 22FDX process. Our newest products, N25 32bit and NX25 64bit RISC-V based cores coupled with a mature toolchain, will provide even more value to customers in these advanced nodes by providing both high speed and power efficiency,” said Charlie Su, CTO and VP of Engineering of Andes Technology.
“IoT designs demand the lowest power consumption as well as the lowest possible cost. The combination of GF’s 22FDX process and Andes’ low power architecture and small footprint will deliver the power, performance, area (PPA) advantage to our IoT customers. We are also excited about Andes newest RISC-V CPU IP. Customer demand is pushing this Open Architecture solution and GF is pleased to support Andes’ customer solutions for the IoT and RF Connectivity markets,” said Jai Durgam, Vice President, Customer Design Enablement of GLOBALFOUNDRIES.
About Andes Andes Technology Corporation was founded in Hsinchu Science Park, Taiwan in 2005 to develop innovative high-performance/low-power 32/64-bit processor cores and associated development environment to serve worldwide rapidly growing embedded system applications. The company delivers the best super low power CPU cores with integrated development environment and associated software and hardware solutions for efficient SoC design. To meet the demanding requirements of today’s electronic devices, Andes Technology delivers configurable software/hardware IP and scalable platforms to respond to customers’ needs for quality products and faster time-to-market. Andes Technology’s comprehensive CPU includes entry-level, mid-range, high-end, extensible and security families to address the full range of embedded electronics products, especially for connected, smart and green applications. For more information about Andes Technology, please visit http://www.andestech.com/.
Hsinchu, Taiwan – December 6th, 2017 – Andes Technology Corporation (TWSE: 6533), the first Asia-based major supplier of innovative 32/64-bit processor IP cores and SoC design platform announced a series of 4 AndeStar™ V3-based high efficiency superscalar processors, the AndesCore™ N15/N15F/D15/D15F. The N15 implements integer baseline instructions while the D15 comes with DSP extension instructions; the N15F/D15F add a floating point unit (FPU) to the N15/D15. The series of processors is perfect for a wide range of applications such as sensors, smart meters, biometric, communication, audio/voice processing, image processing, medical devices, industrial applications, GPS, and drones.
The N15(F)/D15(F) are dual-issue processors capable of delivering performance at 5.41 CoreMark/MHz, the highest among similar products in the industry, with the max frequency of 500MHz at the 40LP process. The N15(F)/D15(F) come with various configuration options, including MMU for Linux OS and MPU for secure RTOSes. Instruction/data caches supports up to 64KB and instruction/data Local Memories up to 16MB respectively, all with optional ECC or parity support for overall SoC reliability. The 64-bit data paths for caches, Local Memories and the main bus provide the bandwidth needed for instruction fetch and data accesses to boost performance. The series of processors is designed for diversified performance-driven applications in the embedded Linux, Real-Time OS or bare metal environment.
The D15(F) with over 130 DSP and SIMD instructions, supported by C/C++ compiler and over 200 optimized DSP library functions, facilitates designers to focus on coding without resorting to assembly programming. The 8/16-bit SIMD instructions process four 8-bit and two 16-bit data in a single cycle, including two 16×16 multiplications, and considerably boost the performance of matrix computation, filter, Fourier transform and statistics calculation. The support of fixed-points, saturation, shift-and-round, and 64-bit operations offers developers all options to efficiently implement their algorithms with the desired precision while Zero-Overhead Loops (ZOL) eliminates the penalties of looping branches for tight loops. The N15(F)/D15(F) also include a built-in IEEE-754 compliant FPU which supports single-precision and double-precision floating-points.
Compared with similar superscalar processors in the industry, the N15(F)/D15(F) deliver 32% higher Dhrystones at 3.36 DMIPS/MHz, and feature 20% less area and 45% less power consumption under 28HPM process. In addition, the D15(F) is 97% faster than the competing processor in the function-equivalent DSP libraries and is also 48% faster in the FPU benchmarks.
“The N15(F)/D15(F) series are versatile processors designed based on a 6-stage asymmetrical dual-issue pipeline micro-architecture, where 2 pipelines are not identical,” said Dr. Charlie Su, CTO and Senior VP of R&D of Andes Technology. “The asymmetrical structure leads to less hardware resources and lower power consumption while still capable of delivering superior performance with the highly-optimized compiler as witnessed by the leading results of integer, DSP and FPU benchmarks. They can perform a pair of loads in parallel, complete a 32×32 multiplication in one cycle, and support a fast radix-4 divider, with dynamic branch predication to further boost overall performance. The N15(F)/D15(F) include debug module and trace support as well as the StackSafe™ stack protection to help catch tough bugs. In addition, the N15(F)/D15(F) offer standby control, PowerBrake power throttling, and automatic suspend-to-RAM control to meet different power management requirements.”
“By combining superior superscalar hardware architecture performance, built-in FPU, optimized compiler and efficient DSP library, the N15(F)/D15(F) meet the needs of a wide variety of applications and allow customers to shorten development time and deliver competitive products easily. The D15(F) has been licensed for customers to design in their multimedia chips.“