Overview
Azinta GPU-Accelerated Analytics Cloud Services - Overview

Graphical Processing Units (GPUs) are now being used to massively reduce the processing time of large scale data mining, risk management and quantitative finance applications, to name but a few. Typical speed-ups can range from 30x to more than 1000x and in many cases application processing can be reduced from days to minutes or less.

Reduction in processing times of this magnitude delivers near real-time analytics resulting in the rapid testing, exploration and implementation of new ideas and increased business agility.

Business users can now use GPU-empowered tools, such as MATLAB, R and Mathematica, to analyse terabytes of data in a fraction of current processing time and costs, and without the up-front investment using the Azinta GPU-Accelerated Cloud and professional services.

The Azinta GPU-Accelerated Cloud offering includes the following hosting, GPU development and consultancy services:

  • Dedicated GPU Hosting Service: The Azinta GPU-Accelerated Cloud service is powered by PEER 1 Hosting GPU Cloud and is based on NVIDIA Tesla M2050 GPUs. Each hybrid CPU-GPU node contains:
    • 2 x M2050 GPUs (Fermi GPU each with 448 cores and 3GB GDDR5 RAM)
    • 2 x Intel Xeon 5220 CPU (each CPU has 32GB RAM)
    • 2 x 300GB HDD
    • Windows or Linux OS
    • 1Gb/s inter-node network
    Users can specify their own dedicated hosted GPU cluster size based on any number of nodes. Monthly and annual payment plans are available.
  • Matlab, R and Mathematica Acceleration: Algorithms and models developed using Matlab, R, and Mathematica can be accelerated, typically from 30x up to 1000x or more. Azinta can provide algorithm tuning services to help derive the best possible performance.
  • GPU-Accelerated Analytics Service for Large Data Sets: Conducting analytical and data mining processing on large data sets comprising terabytes data can often take many days or hours on CPUs. Substantial speed-ups can be achieved using GPU-acceleration. Azinta provides assistance in selecting the best analytical algorithms and tuning them to deliver significant GPU speed-ups over current processing times.
  • GPU-Accelerated Derivative Pricing and Valuations: Many financial services business units are seeking achieve near real-time valuations of their quantitative pricing and risk management models. Azinta provides a GPU-Accelerated hosting service for Derivative Pricing and Valuations that enables users to quickly get started, and have their own dedicated hosted GPU cluster, without up-front hardware investment or waiting on data center availability.
  • Custom GPU Development Services: Experienced GPU developers are availabe for custom GPU development using Matlab, R, Python, C, C++ for both Windows and Linux environments. Azinta's GPU programmers will work with your domain experts to accelerate the development of new GPU-based applications and platforms.
For more information on the Azinta GPU-Accelerated Analytics Cloud services contact GPU@azinta.com

Azinta Kick-Start GPU-Accelerated Development Kit

Developing GPU-Accelerated solutions requires a different set of programing skills than the traditional approaches to software development. Azinta is now providing a Kick-Start GPU Kit that enables a business unit to quickly jumpstart a GPU-based solution. The Azinta Kick-Start GPU Kit contains the following:
  • Analyse the business problem and determine how GPU-Accelerated Analytics could be used to help solve these problems.
  • Determine the set of products and GPU-empowered tools that will be required to implement the solution.
  • Design the required GPU-Accelerated solutions architecture.
  • Implement the GPU-Accelerated solution.
  • Integrate the results from the GPU-accelerated engine with existing analytical tools, BI and legacy applications
  • Provide consultancy on the integration of the GPU service engines with IT operational systems including GPU cluster management

For further information on Azinta Kick-Start GPU-Accelerated Development Kit email: GPU@azinta.com

 
 
 
 
Copyright Azinta Systems Ltd, 2013 About Us   |   Products   |   Services   |   Blog   |   Contacts