Will Wade Director Grid, NVIDIA
Traditionally, virtualised desktops and applications have been severely limited in both graphics compatibility and performance. This has made virtual desktops largely unusable for design and engineering purposes.
With recent advancements in GPU virtualisation and remoting, these challenges are being overcome allowing design and engineering professionals to utilise the power of the GPU wherever and whenever required.
Virtual desktop infrastructures such as Nvidia GRID enables the delivery of high-performance graphics to virtual desktop users creating graphics-intensive content.
This provides the same no-compromise, interactive experience they typically have at their desk when using a physical workstation.
With major virtualisation companies including VMware, Microsoft and Citrix utilising Nvidia GRID technology they are able to provide graphics acceleration to guarantee full scale design application compatibility performance.
In addition, IT managers are also able to provide virtual desktops from a range of virtual GPU profiles, which can deliver everything from an entry level virtualisation to a workstation-class graphics experience.
With GPU powered virtual desktops and applications, enterprises will benefit from higher productivity by providing their employees with secure mobile access from any device.
Additionally, if users lose or damage their laptop or tablet they can be assured that no intellectual property is lost along with the device.
Virtualised desktop solutions also enable contractors to have short-term access to sensitive material without having to transfer any of the project files to their personal machines, thus streamlining the process of working with freelancers.
VGPUs also offer design companies with a limited IT infrastructure the ability to implement virtual workstations for professional, graphics-intensive applications on an ad-hoc basis.
Nvidia GRID VCA offers a complete virtual workstation solution allowing designers to create complex models without the need to purchase dedicated, local workstations.
Even select parts of projects can be outsourced, giving freelancers access to in-house data and graphics performance while keeping data safe.
Advancements in VDI mean firms, large or small, are no longer shackled by purchasing expensive hardware or only employing freelancers who have the right kit.
Utilising the power of the cloud to solve hardware challenges allows businesses to easily expand or enhance their existing virtual desktop infrastructures and further expand the potential of their end users.
Bruno Stefanizzi, Sr. Manager Software Development, AMD
VDI is a hot topic. The ability to work on any device, but still have the performance of a traditional local 3D CAD workstation, is desirable to all designers.
IT managers also have their eye on VDI as they offer manageable solutions with the ability to scale as required. VDI also offers the benefit of centralising IT resources and solves some of the cross platform issues caused by different operating systems and form factors (tablets) that just weren’t there five years ago.
Even CAD software vendors are happy to promote VDI but usually using their servers as it gives them a degree of visibility and control. Software can’t be pirated if the code isn’t available.
Of course to make this work you need good infrastructure with the local or wide area network (WAN). Application performance is also dependent on effective encoding and decoding of the stream (the screen’s content).
There are two parts to this: first accelerating the CAD tool’s 3D graphics engine and then sending pixel data over the wires. For compressing and sending data AMD has RapidFire technology that brings the 3D content from the server with maximum image quality and low latency (time to transmit).
Some CAD software vendors have used earlier APIs but as RapidFire is an open platform working on multiple hardware (not just AMD) adoption is likely to be high like OpenGL and OpenCL.
RapidFire also has high adoption with game developers as the next batch of consoles have AMD APUs (Xbox One and PS4) so decoding streams on this hardware is going to be commonplace.
For professional 3D applications AMD offers solutions where GPUs do the 3D grunt work and compress the streams. Support for virtualised GPUs is also available so more than one user can access a high-performance GPU.
This offers maximum flexibility in resources making it easy to support all different levels of CAD users as well as staff who may only need occasional access to 3D models in PLM software.
GPUs can also be used for compute including FEA and CFD. The parallel architecture of the GPU offers good performance returns.
To make this all work performance is key with servers offering high densities of GPUs from the AMD FirePro S10000 dual GPU card or several single width AMD FirePro S7000 cards (pack more in a server).
AMD will also provide disruptive technology with High Density servers based on APU (CPU and GPU together in one die) for extremely high GPU density and power efficiency and low cost. The point is you need an open solution that works on the greatest array of hardware giving the best performance.