NVIDIA: mobile phones, tablets and HPC (cloud)

Posted by Vincent Hindriksen on
If you want to see what is coming up in the market of consumer-technology (PC, mobile and tablet), then NVIDIA can tell you the most. The company is very flexible, and shows time after time it really knows in which markets is currently operates and can enter. I sometimes strongly disagree with their marketing, but watch them closely as they are in the most important markets to define the near future in: PCs, Mobile/Tablet and HPC.
You might think I completely miss interconnects (buses between processors, devices and memory) and memory-technologies as clouds have a large need for high-speed data-transport, but the last 20 years have shown that this is a quite stable developing market based on IP-selling to the hardware-vendors. With the acquisition of Cray’s interconnect technology, we have seen this is serious business for Intel, so things might change indeed. For this article I want to focus on NVIDIA’s choices.

Profit markets

NVIDIA sees two growth-markets where there is growth in profit: HPC and mobile. PC-sales are declining.

HPC

AMD has not targeted HPC as a specific market with GPUs and Intel MIC has not arrived in the stores yet, so that market is an open road now. I first was a bit disappointed with the GPGPU-capabilities of the current GPUs in the 600-series, but it actually makes sense if NVIDIA wants to get more into HPC. Splitting up the market more clearly in GPGPU-cards and gaming-cards could support marketing to get a Tesla-card in each computer-centre around the world. If they would have invested on GPGPU-capabilities on their gaming-cards, then it could get disastrous if a research-centre published wrong results computed on their GPUs.

PCs

Windows-gaming has been a main drive behind GPUs, but with the release of DirectX 10 the comments in comparison articles got a lot more sarcastic. Also more gamers bought consoles for their games. Why do you need an even faster GPU? And what drove sales after 2006 besides new PCs? So hence the decline.
Why still make GPUs for PCs? NVidia could easily say: goodbye PC, we don’t believe in this sinking ship – go buy a mobile phone with Tegra. Sales are still high enough, but this moment will come. The split between HPC-cards and gaming-cards I see as the beginning.

Mobile

A possible future with mobile phones+tablets well-connected with HPC-clouds having completely replaced PCs is very probable in 2016/2017.Where in the market of PC-components it was very hard to come with better products actually needed by its buyers is very hard since years, hence the growing embedded GPU-market (competition on price, not performance). In such markets it is important to become unique. At the mobile platform there is lots of competition possible, as faster mobiles using less power will be requested for years to come. It was a wise choice to specialise on the mobile platform.

Where CUDA and where OpenCL?

NVIDIA serves quite some markets currently, and this is what I’ve found on what choice they made for OpenCL/CUDA:
  • HPC: definitely CUDA
  • Gaming at consoles: NVIDIA lost most market to AMD – but margins are very low anyway
  • Mobile: OpenCL (see what Neil Trevett of NVIDIA/Khronos says on mobiles and OpenCL)
  • PCs: DirectCompute, CUDA, OpenCL (undecided – see remark below)
  • Browser: WebCL (see also the mentioned link)

Most of all, NVIDIA has discovered LLVM to be able to handle all types of programming languages. While last December a presenter from the CUDA-team got updated on OpenCL-capabilities (last minute of this video), they’ve seemed to have figured out all the possibilities of using LLVM and are heavily betting on it. This will solve their multi-language problem, so even DirectCompute could get an LLVM-frontend. More about LLVM later, as this a very important subject.

We will certainly hear more about how NVIDIA sees the different market segments at their GTC conference.

Read more

Written by Vincent Hindriksen

Mr. Vincent Hindriksen M.Sc is the founder of Stream HPC, a collective of performance engineering experts.

Do you need expertise in performance engineering? We have several experts available (HPC, GPGPU, OpenCL, HSA, CUDA, MPI, OpenMP) and solve any kind of performance problem. Contact me directly to discuss further: +31 854865760, vincent@streamhpc.com or Skype