Back to Table of contents

Primeur weekly 2019-02-25

Special

Astronomers publish new sky map detecting hundreds of thousands of previously unknown galaxies ...

LOFAR and APERTIF telescope systems tooled with machine learning and FPGA/GPU technology make cosmic explosions more visible to astronomers ...

Focus

Over 700 million euro of budget allocated to e-Infrastructure related projects in H2020 - close to 600 million euro to HPC related projects ...

German Government reconfirms its support for supercomputing development on a national and European scale ...

Quantum computing

Dutch King opens Microsoft Quantum Lab on TU Delft campus ...

ISARA Corp. unveils tools to simplify and accelerate quantum-safe cryptography roll-out ...

Exploring the global landscape of quantum technology research ...

Focus on Europe

International research collaboration computes climate past, present, and future ...

Supercomputing Frontiers Europe again in Poland ...

Middleware

Open Compute Project announces updated market forecast ...

Excelero appoints HPC industry visionary Sven Breuner as Field CTO ...

Hardware

vScaler to perform ResNet benchmarks on the NVIDIA DGX-2 server ...

Cray elects Stephen Gold to Board of Directors ...

Atos helps CALMIP achieve leading calculation results with BullSequana supercomputer ...

Indian Prime Minister Shri Narendra Modi launches PARAM Shivay at the Indian Institute of Technology ...

HDR 200G InfiniBand sets new performance records, accelerating multiple high-performance computing and Artificial Intelligence platforms around the world ...

Supermicro expands Intelligent Edge product portfolio to address emerging AI and 5G technologies ...

Applications

HLRS Visualization supports Stuttgart 21 Construction ...

Lorentz-eScience competition 2020 to issue Call for Applications ...

Supercomputers target bovine tuberculosis ...

NESAP hosts first NERSC-9 GPU hackathon for application developers ...

Artificial Intelligence allows for identification of new cancer genes ...

Novel material properties of hybrid perovskite nanostructures for next-generation non-linear electronic devices ...

Clarkson University awarded National Science Foundation Research Experience for Undergraduate Site on High Performance Computing ...

The Cloud

Hyperion Research releases Cloud Application Assessment Tool ...

vScaler to perform ResNet benchmarks on the NVIDIA DGX-2 server


12 Feb 2019 London - The worlds of AI and HPC have an insatiable appetite for more and more performance. With the rise of GPUs being used to run a lot of these AI frameworks, it is only fitting that vScaler put the fastest system in the world to the test. One of NVIDIA's DGX-2 servers arrived onsite recently, and vScaler's engineers have integrated this with their internal vScaler lab facility.

The DGX-2 server builds on the success of the DGX-1 server and increases and improves pretty much everything to create a 2Petaflop (tensor ops) monster of a system. Some of the hardware highlights include:

  • 16x V100 32GB GPUs (That’s half a TB of GPU HBM2 memory space when used with the CUDA unified memory and cudaMallocManaged()
  • 12x NVSwitch switches providing a non-blocking GPU fabric with 2.4TB/s bisection bandwidth.
  • 800GB of a network trunk to get data in and out
  • 30TB of local NVMe SSD to keep those GPUs busy

Another tip of the hat needs to go to the NVIDIA GPU Cloud as the number of containers/applications/frameworks that are available on this platform is growing daily. Optimised containers across Deep Learning, AI and HPC are readily available and vScaler used the Tensorflow container from this platform for the benchmarking exercise.

vScaler integration was seamless - the engineers had a preconfigured image that they have been using for their DeepOps integration and they flashed the system with that - bare metal provision, not virtualised. This provided them with all the tools needed to access the NVIDIA GPU Cloud container repository along with Kubernetes and other optimisation options, all based on Ubuntu Bionic 18.04 LTS.

All benchmarks were run using nvidia-docker, making use of the latest TensorFlow container provided by NVIDIA GPU Cloud, with the imagenet synthetic dataset, provided as part of the tf_cnn_benchmarks.

The benchmark script used was obtained from Github and they performed a sweep of batch sizes across the tests. All tests were run a number of times and the numbers reported were averaged.

To assess the performance of the system the vScaler engineers employed the commonly used ResNet Model which is used as a baseline for assessing training and inference performance. ResNet is shorthand for Residual Network and as the name suggests, it relies on Residual Learning - which tries to solve the challenges with training Deep Neural Networks. Such challenges include increased difficulty to train as they go deeper, as well as accuracy saturation and degradation. They selected two common models: ResNet-50 ResNet-152.

ResNet was introduced in 2015 and was the winner of ILSVRC - Large Scale Visual Recognition Challenge 2015 in image classification, detection, and localisation. There are of course many other Convolutional Neural Network (CNN) architecture models vScaler could have chosen from and in time it hopes to evaluate these also.

Each model was run using various batch sizes to ensure that each GPU was fully utilised, demanding the highest level of performance from the system. Each combination of batch size and GPU count was tested 3 times over 20 epochs and the average result recorded.

During the tests vScaler monitored the system power draw through the onboard sensors and captured data points using ipmitool.

More information on these benchmark runs is available at the vScaler website.
Source: vScaler

Back to Table of contents

Primeur weekly 2019-02-25

Special

Astronomers publish new sky map detecting hundreds of thousands of previously unknown galaxies ...

LOFAR and APERTIF telescope systems tooled with machine learning and FPGA/GPU technology make cosmic explosions more visible to astronomers ...

Focus

Over 700 million euro of budget allocated to e-Infrastructure related projects in H2020 - close to 600 million euro to HPC related projects ...

German Government reconfirms its support for supercomputing development on a national and European scale ...

Quantum computing

Dutch King opens Microsoft Quantum Lab on TU Delft campus ...

ISARA Corp. unveils tools to simplify and accelerate quantum-safe cryptography roll-out ...

Exploring the global landscape of quantum technology research ...

Focus on Europe

International research collaboration computes climate past, present, and future ...

Supercomputing Frontiers Europe again in Poland ...

Middleware

Open Compute Project announces updated market forecast ...

Excelero appoints HPC industry visionary Sven Breuner as Field CTO ...

Hardware

vScaler to perform ResNet benchmarks on the NVIDIA DGX-2 server ...

Cray elects Stephen Gold to Board of Directors ...

Atos helps CALMIP achieve leading calculation results with BullSequana supercomputer ...

Indian Prime Minister Shri Narendra Modi launches PARAM Shivay at the Indian Institute of Technology ...

HDR 200G InfiniBand sets new performance records, accelerating multiple high-performance computing and Artificial Intelligence platforms around the world ...

Supermicro expands Intelligent Edge product portfolio to address emerging AI and 5G technologies ...

Applications

HLRS Visualization supports Stuttgart 21 Construction ...

Lorentz-eScience competition 2020 to issue Call for Applications ...

Supercomputers target bovine tuberculosis ...

NESAP hosts first NERSC-9 GPU hackathon for application developers ...

Artificial Intelligence allows for identification of new cancer genes ...

Novel material properties of hybrid perovskite nanostructures for next-generation non-linear electronic devices ...

Clarkson University awarded National Science Foundation Research Experience for Undergraduate Site on High Performance Computing ...

The Cloud

Hyperion Research releases Cloud Application Assessment Tool ...