Intel said recently that its new Knights Landing Xeon Phi processors are faster and can be scaled better than graphics processing units (GPUs) that NVIDIA makes.
Unsurprisingly, NVIDIA didn't take too kindly to the comparison. The company wrote a blog post in response, saying that it needed to correct a few of Intel's "mistakes," and took a jab at the company, saying, "It's understandable that newcomers to the field may not be aware of all the developments that have been taking place in both hardware and software." Them's fightin' words in the semiconductor business!
Essentially, NVIDIA said that Intel's data was 18 months old, and pointed out that the new data (available to anyone) shows that NVIDIA's Maxwell GPUs are actually 30% faster than Intel's Xeon Phi.
The company also highlighted that its new Titan X server with its advanced Pascal architecture is actually about two times faster than Intel's new processors, and its DGX-1 supercomputer is is five times faster.
There's really not much up for debate here, considering that NVIDIA is using the same data source and standards that Intel is using, but with just the most recent numbers.
Why this actually matters
Here's why Intel and NVIDIA are taking off the gloves: The entire deep-learning market (including software, services, and hardware) is expected to surpass $100 billion in annual revenue by 2024. And right now NVIDIA is becoming increasingly important in the space.
NVIDIA's CEO, Jen-Hsun Huang, said back in April that "deep learning is no longer just an app, not just a field. We think this is so important, we think this is going to utterly change computing. We think it's a big deal."
Many tech companies are already using NVIDIA's GPUs to build their own machine-learning (part of artificial intelligence) computers. For example, Facebook uses NVIDIA's Tesla M40 GPUs for its Big Sur machine-learning computers.
NVIDIA has started marketing some of its GPUs, like the Tesla P100, more to the enterprise market for corporate data centers, and that's likely to increase as server and cloud-computing tasks become more complex. And the company's DGX-1 deep-learning supercomputer is marketed to enterprise customers looking for serious data processing.
NVIDIA saw 112% year-over-year growth from its data center revenue in Q2 2017, driven in part by deep learning. Meanwhile, Intel's own data center growth was just 5% over the same period.
Intel's fighting to keep itself relevant in an industry that's increasingly pivoting toward GPUs for deep learning. That's likely the reason it used old NVIDIA specs to compare its new processors to.
Intel's latest Xeon Phi processors are certainly serious competition for NVIDIA, and will likely create very good deep-learning machines. But I still think investors would be remiss to look past NVIDIA's growing position in the space.