Hi Everyone, my now second favuorite yearly tech conference (after VMworld!) NVIDIA GTC has wrapped for 2021 and as it was the first one I attended, I’d like to share with you some of my personal highlights!
For anyone not aware of GTC, it’s NVIDIA’s yearly conference, they describe it as the #1 AI conference for innovators, technologists, and creatives. The description fits it perfectly, it’s not just for techies, but creatives too. The conference runs every year, this time online on April 12-16.
According to wikipedia (https://en.wikipedia.org/wiki/Nvidia) “It originated in 2009 in San Jose, California, with an initial focus on the potential for solving computing challenges through GPUs. In recent years, the conference focus has shifted to various applications of artificial intelligence and deep learning, including: self-driving cars, healthcare, high performance computing, and Nvidia Deep Learning Institute (DLI) training. GTC 2018 attracted over 8400 attendees. GTC 2020 was converted to a digital event and drew roughly 59,000 registrants.”
So then, the work that NVIDIA is doing, really is shaping and driving forward the world of AI and Machine Learning. If you’re someone with an interest in AI/ML, or even just interested in the future of technology in general, this is THE conference to go back and watch!
At VMware, we work closely with NVIDIA in a number of areas, with an overall goal of delivering AI to every Enterprise. No small feat!
VMware has been an enabler for virtualized GPUs for years now. With vSphere, we were able to leverage NVIDIA Grid and carve it up in many ways to share the GPU workload for many different applications. There’s even more to come moving forward, in particular, two main areas we work together are on End User Computing and the AI Ready Enterprise Platform:
End User Computing
Virtualized GPUs for VDI dramatically increase the performance. For VDI, the major benefit of GPU virtualization is that you can run very performance-hungry and graphics rich applications in the datacenter, letting the physical GPUs take the strain, whilst also being able to access these applications from an iPad on the beach (or wherever you like to work!).
AI Ready Enterprise Platform
VMware provides an NVIDIA certified platform to run NVIDIA AI Enterprise. AI Enterprise is a suite of AI software for IT Administrators, Data Scientists, and AI Researchers to quickly run NVIDIA AI applications and libraries optimized for GPU acceleration by reducing deployment time and ensuring reliable performance.
Two areas of interest across the partnership at the moment are RDMA and GPU in general.
RDMA is a feature in new Smart NICs which allows you to offload traffic from the CPU by creating a direct connection from the NIC to the GPU memory. This is something which will be configurable inside of vCenter, making the route to adoption much more straightforward. RDMA will allow you to scale out your deep learning training workloads whilst preserving the performance required.
The benefits of virtualizing GPUs are similar to the benefits of virtualizing compute in general. You get to share resources across workloads, reduce the overall amount you will need because you’re being much more efficient, management of the GPU becomes more straightforward, the list goes on. vSphere supports virtualization of the latest GPU architectures from NVIDIA, such as the A100. This new architecture can leverage multi-instance GPU, which further increases the efficiency of GPU usage across workloads.
Check out the alliances page to see more about VMware’s partnership with NVIDIA: https://www.vmware.com/uk/partners/global-alliances/nvidia.html
Jensen Huang, CEO of NVIDIA takes you through a whole list of incredible achievements and highlights the approach of the company as a place where art and technology is blended to immerse our senses. Any CEO who wears a leather jacket for the keynote already has my attention, but the content to follow really lived up to that promise, right down to the background music, which was created by artificial intelligence!
I encourage everyone to check out the keynote on the GTC website here: https://www.nvidia.com/en-us/gtc/keynote/ but my personal highlight (other than the cloud and DC announcements) was Omniverse….
Omniverse is a way in which NVIDIA is starting to build standards for virtual worlds. If you think of a video game with a large world to walk around in, this is all created by the games developer, pretty much custom every time. Omniverse is a collaborative cloud service which allows you to build virtual worlds quickly and leverage content from other creatives. Jensen talked about how you can teleport into Omniverse with VR, but also that AIs can come from Omniverse into the real world, through the power of AR (augmented reality). This is really exciting stuff to see. Beyond this, omniverse obeys the laws of physics, right down to individual particles. This has major implications for any kind of robotics or manufacturing, where you can teach your AIs in a world which is as real to them as the real world itself. For anyone who subscribes to simulation theory (https://en.wikipedia.org/wiki/Simulation_hypothesis) this is where things start to get creepy in the best way.
Finally to wrap up, there were many VMware specific sessions at GTC, run by some brilliant VMware minds, from the likes of Kit Colbert, our CTO of Cloud Platform, to staff engineers across different business units and end user computing specialists, talking about the latest and greatest our partnership has to offer. Register for access to the recordings here: https://gtc21.event.nvidia.com/
Then some sessions to look for below:
|Session ID||Session Title||Speakers & Titles||Topic|
|S31895||NVIDIA CEO Keynote||Jensen Huang – Founder and CEO, NVIDIA||Keynote|
|SS31788||How to Best Deploy AI in the Enterprise today, and Where Things are Going Next||Kit Colbert – CTO, Cloud Platform, VMware||AI|
|SS31757||Best Practices for Virtualizing NVIDIA AMPERE GPU in VMware vSphere||Hari Sivaraman – Staff Engineer, VMware
Uday Kurkure – Staff Engineer, VMware
|SS31812||Virtual Creative Workstations for Media and Entertainment||Tom Breakiron – Staff Solutions Engineer, EUC Specialist, VMware||VDI|
|S31631||Enabling and Optimizing Network Function Virtualization with NVIDIA GRID vGPU in VMware vSphere||Lan Vu – Senior Member of the Technical Staff, Vmware
Avinash Chaurasia – VMware Intern
|GPU Virtualization & Networking|
|S31786||Making AI Mainstream in the Enterprise through Compute Virtualization and Container Orchestration||Manuvir Das – Head of Enterprise Computing, NVIDIA
Krish Prasad – SVP and GM Cloud Platform Business Unit, Vmware
|E32023||Harnessing the Power of Python to Control NVIDIA vGPUManagement in VMware vSphere||Tony Foster – Principal Technical Marketing Engineer, Dell Technologies
Johan van Amersfoort -Technologist EUC and AI, ITQ
|SS31474||NVIDIA and VMware Collaborate on Enterprise Deployment of GPUs with Kubernetes: The Technical Details||Justin Murray – Technical Marketing Architect, VMware||AI & Tanzu|
|S31811||Accelerate Performance for Health Care with VMware Horizon and NVIDIA||Tom Campbell – Staff Solutions Engineer, EUC Specialist, VMware||VDI|
|S31710||Apache Spark Acceleration over VMware’s Tanzu with NVIDIA’s GPU and Networking Solutions||Mohan Potheri – Staff Solutions Architect, VMware||Tanzu|
|S32636||How to Optimize Modern Workloads over Next-Generation Hybrid Cloud Architecture||Sudhanshu (Suds) Jain – Director of Product Management, Vmware
Motti Beck – Senior Director – Enterprise Market Development, NVIDIA
|Project Monterrey & Networking|
That’s it for today, thanks for reading!