The Winners Golden Award

NVIDIA Tesla M40

  • Tesla M40 provides single-precision floating-point operations of 7 TFLOPS and up to 24GB ram for significant acceleration of deep learning. Comparing to dual x96 processor systems, the product equipped built-in 8 M40 GPUs to shorten 17 times as long as the computing time.
  • Tesla M40 provides sufficient software functions, including CUDA deep learning library, GPU REST Engine for website applications, FFmpeg acceleration, imaging processing engine, Mesos, etc.
  • Based on the cooperation of NVIDIA and Facebook, the Big Sur deep learning machine is established by Tesla M40 has passed the practice verification of super data centers.

  • Optimized for Machine Learning
    Fastest accelerator for deep learning training (up to 7 TFLOPS SP, purpose-built to dramatically (up to 17x) reduce training time and capacity for more complex model (24 GB).
  • Scale-out performance and 24/7 reliability
    Facebook Big Sur (OCP) : 8 x M40s and DCGM

NVIDIA Hong Kong Holdings Ltd. Taiwan Branch