Hsinchu, Taiwan , May 15, 2023 (GLOBE NEWSWIRE) — Andes Technology (TWSE:6533), a number one provider of high-efficiency, low-power 32/64-bit RISC-V processor cores and Founding Premier member of RISC-V International, right this moment introduced the launch of latest product line, AndesAIRE™ or Andes AI Runs Everywhere, in the end environment friendly

options designed for edge and end-point inference, together with the primary era of AI/ML {hardware} accelerator mental property (IP), the AndesAIRE™ AnDLA™ I350 (Andes Deep Lincomes Accelerator), and the neural community software program instruments and
runtimes, the AndesAIRE™ NN SDK.

With the explosive development of AI/ML functions, there’s a crucial want for top

performance-efficient deep studying options that may function on the edge and end- level, the place the strict energy and vitality constraints make it troublesome to solely depend on CPU structure. The AnDLA™ I350 is particularly designed to handle this problem by delivering industry-leading effectivity, low energy, and small space, making it excellent to be used in a variety of edge inference functions, from sensible IoT gadgets and sensible cameras to sensible dwelling home equipment and robotics.

The AndesAIRE™ AnDLA™ I350 is constructed on prime of Andes power on compute

acceleration from the confirmed CPU applied sciences and experiences accrued over the previous 18 years, offering a excessive performance-efficient platform for AI/ML
acceleration. It helps widespread deep studying frameworks, comparable to TensorFlow Lite, PyTorch, and ONNX, and performs versatile neural community operations comparable to convolution, fully-connect, element-wise, activation, pooling, channel padding, upsample, concatenation, and many others. within the int8 knowledge kind. It additionally options an inside Direct Memory Access (DMA) and native reminiscence, using the perfect computing energy of the {hardware} engines. Operation fusion methods are additionally adopted within the AnDLA™ I350 to carry out most typical operator sequences extra effectively. The key configurable parameters of AnDLA™ I350 embody the MAC quantity from 32 to 4096, and SRAM measurement from 16KB to 4MB, and supply versatile computing energy from 64 GOPS to eight TOPS (at 1 GHz) for a variety of functions.

The AndesAIRE™ NN SDK is a complete set of software program instruments and runtimes for end-to-end growth and deployment. It consists of the next elements:

• AndesAIRE™ NNPilot™: a neural community optimization instrument suite
• AndesAIRE™ TFLM for AnDLA™: an AnDLA™-optimized inference framework operating on a number based mostly on TensorFlow Lite for Microcontrollersk
  

• AnDLA™ driver and runtime

The NNPilot™ can robotically analyze enter NN fashions, apply mannequin pruning and quantization, and generate AnDLA™ executable based mostly on its configuration to carry out inference along with the TFLM framework. The NNPilot™ additionally generates pattern host C code to invoke the AnDLA™ driver within the naked metallic surroundings.

To tackle the evolving and quickly advancing AI applied sciences, Andes’ imaginative and prescient is an

extensible AI subsystem, which seamlessly integrates AndesAIRE™ AnDLA™,

AndesCore™ RISC-V CPU, and Andes Custom Extension™ (ACE). In such a subsystem, most structural and time-consuming components of the AI workloads might be computed
effectively within the AnDLA™ whereas much less structural computations comparable to non-linear capabilities might be offloaded to the highly effective and versatile RISC-V CPUs with DSP/SIMD or Vector extensions. The ACE performs a key position for environment friendly knowledge motion between the CPU and the AnDLA™ to cut back important reminiscence bandwidth and energy consumption whereas rising {hardware} utilization. The ACE can facilitate even sooner processing by producing custom-made directions for domain-specific functions, comparable to knowledge pre- or post-processing. In addition to the extensibilities from {hardware} IP’s, Andes commits to steady development of AndesAIRE™ NN SDK and AndesAIRE™ NN Library for the mass-production SoCs to undertake the long run AI
algorithms. Andes had added a couple of hundred compute library APIs yearly since 2021, and we’ll hold optimizing and including new capabilities into NN SDK and NN library sooner or later.

“Andes Technology is thrilled to introduce our new product line for AI/ML solutions, the AndesAIRE™ AnDLA™ I350 and the AndesAIRE™ NN SDK, an exceptional
performance-efficient hardware and an end-to-end software solution for edge and end-point inference,” mentioned Dr. Charlie Su, CTO and President of Andes Technology. “With the AndesAIRE™, we are empowering developers and innovators to create extensible and future-proof AI/ML SoCs and applications.”

“AndesAIRE™ product line projects our vision for the AI/ML market,” mentioned Simon TC Wang, Senior Technical Marketing Manager of Andes Technology. “By combining the advantages from RISC-V CPUs, the AnDLA™, and the ACE into an AI subsystem, the performance, power consumption, and area can be well-balanced for customers to deliver competitive solutions. Furthermore, the flexibility is guaranteed by RISC-V CPUs and NN software stack, and the extensibility is affluent for customers to
differentiate their unique value for target AI/ML applications.”

The AndesAIRE™ AnDLA™ I350 and the AndesAIRE™ NN SDK can be found for

licensing now in an early-adopter program. For extra data, please go to the Andes Technology web site at “http://www.andestech.com/en/products-

solutions/andesaire-ai/”.

About Andes Technology

Eighteen years in enterprise and a Founding Premier member of RISC-V International, Andes is a publicly-listed firm (TWSE: 6533; SIN: US03420C2089; ISIN:

US03420C1099) and a number one provider of high-performance/low-power 32/64-bit

embedded processor IP options, and the driving power in taking RISC-V mainstream. Andes’ fifth-generation AndeStar™ structure adopted the RISC-V as the bottom. Its V5 RISC-V CPU households vary from tiny 32-bit cores to superior 64-bit Out-of-Order processors with DSP, FPU, Vector, Linux, superscalar, and/or multi/many-core
capabilities. By the top of 2022, the cumulative quantity of Andes-Embedded™ SoCs has surpassed 12 billion. For extra data, please
go to https://www.andestech.com. Follow Andes on LinkedIn, Facebook, Twitter, Bilibili and YouTube!