With Deep Learning experiencing unprecedented momentum in the HPC market, I find it exciting that Bright’s development team is focusing in on this area, and is involved in cutting edge Deep Learning innovation.
Processing large amounts of data for Deep Learning requires enormous quantities of computational power. To overcome this, organizations are embracing parallel compute architectures that harness thousands of small cores to accelerate these computationally intensive tasks.
It’s therefore no surprise that hardware vendors and partners of Bright are busy in the Deep learning market as well; in fact, Dell and NVIDIA are collaborating on Deep Learning configurations to support customers as they enter into this brave new world.
What role does Bright play?
One of the biggest challenges we see companies facing, is getting their Machine and Deep Learning environments up and running quickly and managing them effectively.
Bright’s solution makes Deep Learning more accessible by combining high performance hardware from Dell and NVIDIA with frameworks, libraries and drivers to create an integrated, highly effective Deep Learning platform.
Our solution provides a choice of GPU-accelerated versions of common Machine and Deep Learning frameworks, such as Caffe, Torch, Tensorflow, and Theano. Bright also takes away the pain and complexity of configuring and maintaining all the dependent components, allowing researchers to focus on their science and investigations, instead of managing the tools.
We also package and include a comprehensive selection of the most popular libraries, including MLPython, NVIDIA CUDA Deep Neural Network library (cuDNN), Deep Learning GPU Training System (DIGITS), and CaffeOnSpark (a Spark package for deep learning).
Finally, Bright’s Deep Learning solution includes over 400MB of Python modules that support the machine learning packages, the NVIDIA hardware drivers, CUDA (parallel computing platform API) drivers, CUB (CUDA building blocks), and NCCL (library of standard collective communication routines).
Bringing it all together
Dell EMC has worked with Bright Computing to offer Bright’s Deep Learning software stack on a number of Dell EMC Deep Learning hardware configurations.
These example configurations are available in several sizes, with all but the small (S) hardware-only configuration including Bright’s management and Deep Learning solution. The medium (M) configuration, actually a “single node”, comprises one Dell C4130 with two Tesla P100 (PCIe) GPUs. The Large (L) comes with one Dell C4130 with four Tesla P100 GPUs (PCIe or SXM2), and the Extra-Large (XL) gets you Four Dell C4130s, each with four Tesla P100 GPUs (PCIe or SXM2) plus an EDR IB switch – providing an enterprise-class, highly capable platform for Machine and Deep Learning.
So, whether you’re just getting started in Machine Learning or Deep Learning, or are ready to scale up, the Dell/NVIDIA/Bright solution has what you need.
Last month, at the International Supercomputing Conference (ISC), we were delighted to host both Dell and NVIDIA in our booth theater. Axel Koehler, Principal Solution Architect at NVIDIA, talked about “GPU-Accelerated Deep Learning”, and Onur Celebioglu, Engineering Director in charge of HPC Solutions at Dell EMC, presented on “HPC and Machine Learning Simplified”. So popular was Dell EMC’s commitment to this emerging new area of research, that they won the ISC Vendor Showdown. You can read all about this in an article posted on Forbes.com.
Needless to say, Machine and Deep Learning are one of the hottest technologies going. To find out more about the Dell EMC and NVIDIA presentations, or the Bright for Deep Learning solution, please click here.