The big picture: In the desktop space, Intel has had a long-running battle with main competitor AMD, where the market has recently favored Team Red. The Ryzen series has pushed Intel out of the spotlight, but as the company now gets closer to launching its Alder Lake CPUs to level the playing field and regain some market share in a business where it was once dominant, it is also preparing to gain ground in another highly competitive industry.
In a recent interview with CRN, which covers multiple topics with Intel CEO Pat Gelsinger, the upcoming battle with Nvidia was highlighted, among other things. Going forward, Intel plans to differentiate itself from the GPU giant by being “much more ecosystem-friendly” in terms of software that supports its components.
“Nvidia has gotten too proprietary, and that’s rife in the industry, and that’s why we’re going to populate this stack with oneAPI, but do it in a way that is much cheaper and more open to the industry and its innovations,” he said.
oneAPI is an open programming model that covers various types of architecture, including Intel’s silicon products such as CPUs, GPUs and FPGAs.
Regarding Intel’s “under-focused” graphics business, Gelsinger stressed that the company’s efforts in the GPU market in terms of features, performance, and performance must go beyond what must all be offered at a price that is convincing to the consumer.
Gelsinger largely attributed Nvidia’s success to how his leadership has run the company over the past decade, but also pointed out that they “have been really lucky” in the artificial intelligence space.
“Nvidia had essentially a 10x or better performance lead for a decade. If you have that, a 10x leadership for 10 years, people are going to take advantage of that. And then they were really lucky: AI happened. 30 years of overnight success and they got it really well at this point. So you worked hard, earned it, and then you got lucky in that regard, “explained Gelsinger.
Gelsinger went on to explain how Intel specifically intends to bring attractive products to the GPU market and how they will move seamlessly from integrated to discrete on the Intel platform.
“So what do we have to do? Deliver great products in these segments (…) and the market demands of us to provide them with an alternative. We then have to deliver them with a unique, differentiated added value.
And in the GPU business, we go to the customer and say, “Well, guess what, we’re just the undisputed leader in integrated graphics. You are already qualifying all of our stuff all the time for every unit that you are going to be shipped, and we will Make the transition from integrated to discrete on the Intel platform seamless.
And even better, we will enable an integrated and discreet collaboration. So when you have three [execution units] worth in the built-in [GPU], then you have 10 EUs in discrete value, we will give you 13 EUs in value, and you will only buy 10 EUs in discrete GPU value, and you will qualify a product that [works] seamlessly between these two. “Well, that’s pretty nuanced. And that’s just an example.
To add weight to his words, Intel has hired a former top GPU architect at AMD, Vineet Goel, to oversee its Xe GPU architecture on which future GPU products will be based. He will lead a team of architects and design engineers working on the “Architecture, Design and Verification of Intel’s Xe IP Roadmap”.
Related reading: Last time Intel tried to make a graphics card
Intel certainly has a lot to do if it is to take a dent in the graphics market. Nvidia is firmly established in the industry, with AMD being a strong secondary player – not to mention its dominance among chipmakers – and it will become an even bigger force to be reckoned with should the acquisition of Arm happen.
Nevertheless, Intel got off to a good start, measured by the anticipation for the upcoming Xe-HPG graphics cards. As for the fight against AMD, Gelsinger believes his competitor’s success will not last in recent years with the upcoming Alder Lake and Sapphire Rapids launches.