AMD And Samsung Announce Partnership – Radeon Coming To Your Smartphone?
AMD has been riding a wave of optimism since it released details on its upcoming 7nm products including the firm’s 3rd-generation Ryzen family of retail CPUs as well as next-generation EPYC server parts.
Samsung LSI to receive graphics IP and patent license based on the RDNA graphics architecture
The deal we are learning about today does not concern CPUs, however – it involves graphics processors. AMD made a splash at Computex with a somewhat unexpected announcement about its upcoming Navi processors which will be using a new architecture under the hood named “RDNA”. That brings us to today’s news.
Samsung and AMD are entering into a multiyear strategic partnership to enable Samsung to utilize AMD’s host of patents and IP relating to video and graphics accelerators. Samsung LSI, the business unit of the Korean electronics giant is the chip design and fabrication arm of the company and its SLSI that is contracting with AMD to utilize RDNA tech to power its future mobile GPUs.
No one is quite sure yet what the details of the arrangement will be and just how Samsung will integrate AMD’s tech into their in-house GPU design. Samsung has been cited as working on a clean-sheet integrated GPU design for use with its Exynos SoC products that power many of its smartphones worldwide.
If Samsung has been indeed working on their own solution then that leaves two major ways we can interpret the partnership.
The first reality might be that Samsung is nearing completion of its design and is gridlocked because of industry-standard patents that it needs to commercialize its design. The graphics business doesn’t see very many new designs as NVIDIA, AMD, Qualcomm, and ARM have the very best patents and IP locked down at this point. The firms above field consumer graphic processors in the forms of Geforce, Radeon, Adreno (mobile), and Mali (mobile), respectively. These incumbents own large amounts of related patents and navigating this legal nightmare would be quite the challenge for any GPU design team.
A soon to be finalized Samsung GPU would need some access to patent licenses and AMD would be an excellent source given the company’s available technology in addition to its openness to working with outsiders for licenses. For instance, AMD entered into a joint venture with a Chinese firm to license out its “Zen” core, and graphics tech going to a Korean firm would definitely be seen as less critical/risky by AMD leadership. In this first instance, Samsung might already have its hardware ready to deploy in upcoming Exynos processors – it just needs AMD’s licenses to legally be able to do so.
AMD re-entering mobile graphics market after a 10-year hiatus
The second and in our opinion more likely reality is that Samsung needs AMD’s actual technology to complete its “S-GPU”. GPU development is exceedingly difficult in today’s world of advanced rendering, extreme compression, and complex shading pipelines. Samsung might be throwing in the towel and looking to AMD who supplies chips for not only itself but for the majority of home video game consoles in the form of Xbox and PlayStation APUs.
We might have some clues enclosed in the official announcement. To start here are CEO Dr. Lisa Su’s comments on the deal:
Adoption of our Radeon graphics technologies across the PC, game console, cloud and HPC markets has grown significantly and we are thrilled to now partner with industry leader Samsung to accelerate graphics innovation in the mobile market, this strategic partnership will extend the reach of our high-performance Radeon graphics into the mobile market, significantly expanding the Radeon user base and development ecosystem.
From the language here it sounds as though Dr. Su is pointing to actual Radeon technology getting used in Samsung’s eventual hardware application. She remains fairly vague in terms of the type or scope of graphics that Radeon will be used for, but the phrase “Radeon user base and development ecosystem” really seems to steer the conversation away from an all-Samsung chip using AMD’s patents and closer to an actual Radeon-based chip scaled down for mobile.
Beyond that, the company’s press release mentions specifically “custom graphics IP based on the recently announced, highly-scalable RDNA graphics architecture”. Remember, AMD also calls the graphics chips provided to Sony and Microsoft “custom graphics IP.”
AMD exited the burgeoning market in 2008 when it sold its mobile-focused graphics IP, back in those days called AMD(ATi) Imageon to Qualcomm, who integrated the IP into their Snapdragon SoCs as the “Adreno” graphics processor. The striking fact there is AMD sold the unit for a measly $65 million in cash! Of course, this was in late 2008 and the smartphone industry was basically non-existent compared to today, and at the time AMD management was focused on raising as much cash as possible. Today that deal could easily be valued in the billions!
We can’t say what the financial impact will be for AMD, but it seems like its a sound business move as AMD does not compete currently in the mobile graphics space and this can be a way the firm can increase its revenue – all based on existing tech that has largely already been paid for. Samsung sells millions of Exynos chips by way of its Galaxy smartphones, and it currently pays ARM for a license to use its Mali line of GPUs. Now all of that revenue can be diverted to AMD for “licenses and royalty payments”.
Shares of AMD spiked on the news in A.M. trading, however, they relaxed just about all the way down over the course of the day before settling to a muted gain of less than a percent. Its worth noting that AMD survived an overall red day for tech, as large companies like Apple (NASDAQ:AAPL), NVIDIA (NASDAQ:NVDA), Qualcomm (NASDAQ:QCOM), and Intel (NASDAQ:INTC) all seeing losses.
Today’s announcement is a truly disruptive move, as this solidifies Samsung’s plans to roll out its own graphics silicon which is a rare play in this industry. Is it possible AMD and Samsung will launch something that looks very similar to a Navi GPU, just heavily scaled down to fit inside a phone’s’ thermal envelope? Very interesting times ahead!