How deep is your learning?

Recently, we've had some hands-on time with NVIDIA's new TITAN V graphics card. Equipped with the GV100 GPU, the TITAN V has shown us some impressive results in both gaming and GPGPU compute workloads.

However, one of the most interesting areas that NVIDIA has been touting for GV100 has been deep learning. With a 1.33x increase in single-precision FP32 compute over the Titan Xp, and the addition of specialized Tensor Cores for deep learning, the TITAN V is well positioned for deep learning workflows.

In mathematics, a tensor is a multi-dimensional array of numerical values with respect to a given basis. While we won't go deep into the math behind it, Tensors are a crucial data structure for deep learning applications.


NVIDIA's Tensor Cores aim to accelerate Tensor-based math by utilizing half-precision FP16 math in order to process both dimensions of a Tensor at the same time. The GV100 GPU contains 640 of these Tensor Cores to accelerate FP16 neural network training.

It's worth noting that these are not the first Tensor operation-specific hardware, with others such as Google developing hardware for these specific functions.

Test Setup

  PC Perspective Deep Learning Testbed
Processor AMD Ryzen Threadripper 1920X
Motherboard GIGABYTE X399 AORUS Gaming 7
Memory 64GB Corsair Vengeance RGB DDR4-3000 
Storage Samsung SSD 960 Pro 2TB
Power Supply Corsair AX1500i 1500 watt
OS Ubuntu 16.04.3 LTS
Drivers AMD: AMD GPU Pro 17.50
NVIDIA: 387.34

For our NVIDIA testing, we used the NVIDIA GPU Cloud 17.12 Docker containers for both TensorFlow and Caffe2 inside of our Ubuntu 16.04.3 host operating system.

AMD testing was done using the hiptensorflow port from the AMD ROCm GitHub repositories.

For all tests, we are using the ImageNet Large Scale Visual Recognition Challenge 2012 (ILSVRC2012) data set.

Continue reading our look at deep learning performance with the NVIDIA Titan V!!

AMD Radeon Vega Frontier Edition Air and Liquid-Cooled GPUs Now Available for Pre-Order

Subject: Graphics Cards | June 23, 2017 - 02:21 AM |
Tagged: vega frontier edition, Vega, radeon, pre-order, gpu, amd

AMD promised “late June” availability for its Radeon Vega Frontier Edition, and it looks like the company will almost hit that mark. The latest high-end prosumer and workstation GPU from AMD is now available for pre-order, with an expected ship date of July 3rd.

Update [2017-06-24]: The initial pre-order stock at both Newegg and Amazon has sold out. It's unknown if AMD will make additional units available in time for the launch.


The Radeon Vega Frontier Edition helps drive the new digital world. It nurtures creativity. It is your gateway to parts unknown. Expand the boundaries of what's possible and witness the impossible. With the new "Vega" GPU architecture at its core, you will have no barriers or compromises to what you want to achieve. Take advantage of the massive 16GB of cutting-edge, second-generation high-bandwidth memory to create expansive designs and models. Crunch and manipulate datasets using the sixty-four Next-Gen Compute Units (nCUs - 4096 stream processors) at your disposal. Unleash your imagination to develop games, CGI or VR content leveraging the latest features found on the "Vega" GPU architecture and witness the breathtaking power of "Vega" course through your system.

The Radeon Vega Frontier Edition is available in both air and AIO liquid-cooled designs, and the product page clarifies the following specs. Note, however, that specific core and memory clocks are not listed, which is especially interesting given the liquid-cooled varient's increased TDP.

AMD Radeon Vega Frontier Edition (Air Cooled)


  • Memory: 16GB High Bandwidth Cache
  • Memory Bandwidth: 483 GB/s
  • Compute Units: 64
  • Stream Processors: 4096
  • Single Precision Compute (FP32): 13.1 TFLOPS
  • Half Precision Compute (FP16): 26.2 TFLOPS
  • Display Output: 3 x DisplayPort 1.4, 1 x HDMI 2.0
  • TDP: 300W
  • Price: $1,199.99 (Newegg | Amazon)

AMD Radeon Vega Frontier Edition (Liquid Cooled)


  • Memory: 16GB High Bandwidth Cache
  • Memory Bandwidth: 483 GB/s
  • Compute Units: 64
  • Stream Processors: 4096
  • Single Precision Compute (FP32): 13.1 TFLOPS
  • Half Precision Compute (FP16): 26.2 TFLOPS
  • Display Output: 3 x DisplayPort 1.4, 1 x HDMI 2.0
  • TDP: 375W
  • Price: $1,799.99 (Newegg | Amazon)

Before you pre-order, however, there’s one big caveat. Although AMD touts the card as ideal for “innovators, creators, and pioneers of the world,” the Radeon Vega Frontier Edition will lack application certification, a factor that is crucial to many who work with content creation software and something typically found in high-end professional GPUs like the Quadro and FirePro lines.

For those hoping for Vega-based professional cards sporting certification, the Vega Frontier Edition product page teases the launch of the Vega-powered Radeon Pro WX in Q3 2017.