by Tobias Kreidl, CTP, Steering Committee
One of the reasons to be part of CUGC is end user advocacy. For me, that advocacy sometimes centers on my work at a university. If you work at a university or a nonprofit, you know that our budgets are often smaller and our procurement procedures more complex than many in the private sector. As a result, we often cannot add technologies we may want or need to our environment. Because of that, I have participated in discussions with NVIDIA about how they can help reduce costs for our unique circumstances in higher education. Recently, we took another step forward when NVIDIA announced a new licensing option for educational institutions and nonprofits. I share the details below. I also want to disclose that my only relationship with NVIDIA is as an end user customer.
Following the price reductions and the introduction of 30% educational discounts on the NVIDIA GRID 2.0 licensing for Virtual PC and Virtual Workstation together with dropping the price for Virtual Applications connections last year, NVIDIA has now announced an even more aggressively priced vGPU licensing option available to not only educational institutions, but also to other non-profit organizations.
In this new GRID 4.0 model, the Virtual Applications licensing is left intact whereas the Virtual PC (vPC) and Virtual Workstation (vWS) offerings are now both covered by a single educational vGPU license available as a perpetual as well as annual license. The base license runs for $99 with a $25 annual SUMS (Support, Update, and Maintenance Subscription) fee. In place of a discount, this offering now reflects a completely different and separate license option.
For a three-year contact, this saves around 67% compared to what the previous three-year vWS, educational license cost, hence $174 vs. $525. The educational vWS license can also be used to obtain a token for a vPC instance, if desired. The chart below shows the pricing comparison between the old and new products.
To provide some additional detail, I created the spreadsheet below to show a more detailed cost breakdown that incorporates the specific differences between annual licensing and subscription support components.
Above all, the cost savings will clearly become even greater beyond the three-year mark.
Note that the Virtual Applications licensing pertains to any running virtual machine (VM) that has the capability of connecting to one of the GRID servers. The hosting server instance can make use of an “A” profile (for example, an 8A would incorporate one of the full GRID engines on an M60). That can be run for free on the server end. For each and every instance, a separate license is needed to leverage Quadro capabilities (including on the server, itself) and this requires specifically a “Q” configuration all around. At present, NVENC support requires a “Q” or “B” profile under GRID 4.0 or above, both on the server and client sides. In addition, the client currently requires a minimum 1 GB frame buffer size.
This step accentuates NVIDIA’s interest in service to the educational and not-for-profit communities and with this new licensing, provides a more economical way to leverage vGPU technology than ever before. The licensing pertains to all Tesla GRID models. The introduction of this specialized pricing is a welcome enhancement to NVIDIA’s GRID program and should encourage even more participation by educational and non-profit entities as a consequence of this more affordable licensing option.