2805 Bowers Ave, Santa Clara, CA 95051 | 408-730-2275
sales@colfax-intl.com
My Colfax  

Colfax servers and workstations based on the NEC SX-Aurora TSUBASA vector architecture give developers, scientists and engineers a platform to accelerate HPC and AI application performance using vectorization.

The platforms provide both unprecedented performance and power efficiency, and are ready for out-of-the-box deployment with pre-configured NEC Software Development Kit for Vector Engine.


Inquire

Download SX-Aurora TSUBASA Brochure

Applications

Successful use cases of the NEC Vector Engine architecture include, fluid dynamics/analysis (CFD) for aerospace, weather/climate, earth science, seismic imaging/reservoir simulations for O&G industry, monte carlo based finance risk analysis simulations, statistical machine learning, numpy, data frame, recommendation engine, demand/price prediction, and deep learning

Find the Right Fit for Your Needs



Colfax ProEdge™ VX100 Workstation

  • Up to 4x NEC Vector Engine Processor Type 10C
  • 2x Intel® Xeon® Scalable Processors
  • 16x DIMMs
  • 8x HDDs/SSDs
  • NEC Software Development Kit for Vector Engine
  • NEC MPI Library
  • CentOS

Configure



Colfax CX2400a-VE8 Server

  • Up to 4x NEC Vector Engine Processor Type 20B
  • 1x AMD EPYC™ 7002 Processor
  • 8x DIMMs
  • 8x HDDs/SSD
  • NEC Software Development Kit for Vector Engine
  • NEC MPI Library
  • CentOS

Configure

NEC SX-Aurora TSUBASA Architecture

Vector processor + x86/Linux architecture - The NEC SX-Aurora TSUBASA architecture contains the Vector Engine (VE) and Vector Host (VH). The VE executes complete applications while the VH mainly provides OS functions for connected VEs. The VE is connected to the VH, a standard x86/Linux node, through PCIe. This architecture, which executes an entire application on the VE and the OS on the VH, combines highest sustained performance, in a well-known x86/Linux environment.

NEC Vector Engine (VE) Processor

The NEC Vector Engine (VE) processor integrates eight vector-cores and 48 GB of high bandwidth memory (HBM2) providing a peak performance of up to 3.07 TeraFLOPS. The computational efficiency is achieved by the outstanding memory bandwidth of up to 1.53 TB/s per CPU and by the latency-hiding effect of the vector architecture.

NEC SX-Aurora TSUBASA Software Environment

The NEC Vector Engine product family is designed for computing workloads that follow standard (CPU-oriented) programming paradigms based on task- and data-parallelism. Compatibility with standards is provided by a NEC compiler with automatic vectorization capability, which allows users with no special programming skills to enjoy the benefits of vector computing.

In addition to the software development stack for compiled languages, the Frovedis middleware implemented for NEC SX-Aurora TSUBASA provides the Apache SPARK API for machine learning acceleration with a Python interface.

Pre-installed Software

The following software are pre-installed:

  • VEOS
    - VEOS Application Runtime / Application Development
  • NEC Software Development Kit (SDK) for Vector Engine
    - C/C++ compiler
    - Fortran compiler
    - binutils
    - NEC Numeric Library Collection (ASL, BLAS, LAPACK, ScaLAPACK)
    - Tuning tool
    • PROGINF/FTRACE
    • NEC Ftrace Viewer
    - NEC Parallel Debugger
  • NEC MPI

NEC Vector Engine (VE) Processor Models

NEC Vector Engine

Type 10C

Type 20B

Core Specifications

Clock speed (GHZ)

1.4

1.6

Peak performance (GigaFLOPS) DP

268.8

307.2

Average memory bandwidth (GB/s)

94

192

Processor Specifications

# of cores / processor

10

8

Peak performance (TeraFLOPS) DP

2.15

2.45

Memory bandwidth (TB/s)

0.75

1.53

Cache capacity (MB)

16

16

Memory capacity (GB)

24

48

What can we help you with?

As a NEC Partner, we have established a reputation for solving the most complex problems and securing outstanding results for enterprises across different sectors.

Get your project rolling with quick answers from technical sales to address your unique challenges.

Tell us about your challenge

All product, brand, or trade names used on this page are the trademarks or registered trademarks of their respective owners.