VividSparks: High Speed Fabless Semiconductor Technology

pl7

What is VividSparks?

VividSparks is a high-speed computer technology firm that specializes in fabless semiconductors and is self-funded. They are focused on resolving fundamental arithmetic circuit design issues. They have libraries and algorithms that are extremely silicon efficient and perform exceptionally well. Their technology at VividSparks is designed to provide excellent performance to consumers while using as little electricity as feasible. They think that technology should be innovative rather than costly. Furthermore, new technology should be accessible and cheap to everybody. VividSparks aspires to build a computing environment that consumes the least amount of power and consumes the least amount of silicon space while delivering world-class performance.

Dr. Vijay Holimath, President & CEO

Dr. Vijay Holimath, the company’s President and CEO, has over 20 years of expertise building and deploying high-performance computing machines and network-on-chip systems for scientific computing and automotive applications. His research has mostly focused on the design and implementation of efficient mobile computing algorithms throughout his career. He has several research publications to his credit. Dr. Holimath has collaborated with several educational institutions in France, Japan, and the United States. He has already created and presented many company proposals to investors.

He has a Doctorate from the University of Santiago de Compostela in Spain, a Masters’s in Computer Science from Newcastle University in the United Kingdom, and a Bachelor’s in Electronic Engineering from Karnataka University in India (India). Dr. Holimath worked as a Research and Development Engineer at the National Institute of Informatics in Tokyo before starting VividSparks (Japan).

VividarithmetiK

VividarithmetiK is a next-generation processor that uses a new number system called POSIT to compute fundamental arithmetic operations including addition, subtraction, multiplication, division, square root, trigonometric functions, compare and convert instructions. VividarithmetiK has a Supersonik, RacEr, FalKon, and Tez processor family with distinct POSIT configurations geared for HPC, Graphics, AI, and Automotive applications, respectively. VividarithmetiK uses a new adder dubbed the Carry Free Adder to prevent carry propagation in adders and multipliers (CFA). CFA not only prevents carry propagation, but also uses extremely little energy. Rather than manufacturing circuits in higher technology nodes, we develop high-performance chips by incorporating super-fast arithmetic libraries into them.

Their Products

VividKompiler, the world’s first native POSIT C, C++, and Fortran compiler, is included with all of the following products :

1. HPC Multicore

64 POSIT cores deliver unrivaled server-class performance in terms of efficiency and compactness.

It possesses the following features :

  • Designed for high-performance computing (HPC) applications that perform complicated calculations.
  • OpenFOAM and GROMACS are supported.
  • High-end mathematical functions are supported.
  • When compared to the Floating Point number system, it produces more accurate results.

2. Automotive multicore

32 POSIT cores in an automotive multicore that doesn’t sacrifice performance Automotive applications are the best fit.

It possesses the following features :

  • All floating-point instructions are supported.
  • VividSparks POSIT compiler is supported.

3. World’s first POSIT GPGPU

It possesses the following features :

  • Up to 512 cores can be configured in this premium GPU architecture.
  • POSIT setups may be parameterized in a flexible design.
  • POSIT operations per watt are higher.
  • Supports POSIT commands that are equivalent in Floating Point.
  • Supports CUDA-equivalent programming paradigms, as well as TensorFlow, Python, and PyTorch.
  • OpenGL/OpenCL are supported.

4. AI Multicore

It possesses the following features:

  • 16 POSIT cores for AI and machine learning
  • Arithmetic units that are very low in weight and are ideally suited for AI applications
  • Matrix multiplications and additions are supported for Neural Networks.
  • Flexible architecture with processing element parameterization Supports OpenGL/OpenCL
Industry Computer Hardware
Company size 51-200 employees
Headquarters Hubli, Karnataka
Type Privately Held
Founded 2015
Website: http://www.vivid-sparks.com

Leave a Reply

Your email address will not be published.