Ten years ago, AMD bought Canadian GPU manufacturer ATI Technologies, in an attempt to expand their portfolio, and to work towards integrated systems like we are seeing with their Fusion APUs (not to be confused with the other just announced Fusion chip), and with mobile SoCs today. Shortly after the acquisition, they sold off some of their non-core business units in order to focus on CPUs and GPUs for desktops and laptops.
In 2008 and 2009, AMD sold off their fabrication division (GlobalFoundries), their set-top box SoC division (Xilleon), and their mobile GPU division (then known as Imageon, and now known as Qualcomm Adreno). While they sold Imageon to Qualcomm for relatively little, just $65 million as it was underperforming in the mobile market that was dominated by feature phones and PDAs at the time, they did see the potential to re-enter the market in the future, and decided to keep the Imageon brand name (something that they didn't do for Xilleon). That short period between AMD purchasing ATI and selling off Imageon resulted in some interesting combinations like the HTC Advantage, a 5" touchscreen phone in early 2007 with an ARM Intel CPU and AMD graphics.
AMD kept the Imageon brand name because even at that point in time, the convergence of phones, laptops, and desktops was already starting to become apparent, although few predicted just how quickly it is happening. As processors get faster, we're hitting a point where we no longer need more processing power in laptops to complete the tasks that most people use their computers for, so instead we're seeing a focus on power efficiency and battery life.
"With AMD's Zen processor coming up, the time may be ripe for AMD to re-enter the tablet market"
As CPU and GPU power usage in laptops drops, we're beginning to see smaller and smaller gaps, with some laptops even running on what are essentially tablet chips. Small little 4.5 W TDP SoCs that don't need a cooling fan, with an integrated GPU and CPU, just like what you're seeing in phones.
With AMD's Zen processor coming up, the time may be ripe for AMD to re-enter the tablet market. Unfortunately, their lack of LTE radios would prevent them from directly entering the phone market (and is the same reason why Nvidia and Intel ran into trouble), however it doesn't prevent them from licensing their technology out to other manufacturers that are looking for something beyond ARM's Mali GPUs.
AMD is no stranger to licensing out their GPU technology, having licensed it to STMicroelectronics back in 2007, and reportedly being in talks to license it to MediaTek last year. To see rumors that Samsung is attempting to license either Nvidia or AMD's GPU designs really shouldn't come as a surprise. Nvidia seems a bit unlikely, especially seeing how they directly compete with Samsung in the tablet market and are investing substantial amounts of money into their dynamic binary translation based Denver CPU cores that are only used in their Tegra chips, but AMD could be an interesting fit.
AMD is looking for additional revenue streams to help fund their two-pronged fight against Nvidia in the GPU market (where Graphics Core Next is showing strong Vulkan performance), and Intel in the CPU market (where their Zen CPUs appear to be a return to competitiveness, after years of struggling with the failure of Bulldozer). The extra bit of revenue from licensing their GPU technology for phone and tablet chips could help them fund the substantial R&D spending required to keep up with Nvidia and Intel, but more importantly, it would help pave the route for future attempts at entering the mobile sector. It would get a lot of the smaller issues of Android support for AMD GPUs out of the way.
And that is the key. Support. Software support. If AMD wants to enter the tablet SoC and fanless laptop market, they need to have software support for the OSes that run on those platforms. AMD has been making a massive push to improve their Linux support, open sourcing a large part of their GPU stack over the past two years (and peripheral tools like GPUOpen), and a substantial portion of that carries over to Android (and that open source code is especially helpful with facilitating aftermarket dev support). Getting their GPUs into a couple third party SoCs would give them time to really iron out the bugs and improve their support for Android (and Linux) before launching a tablet SoC of their own.
Can AMD leverage their Heterogeneous Systems Architecture to successfully enter the tablet market? Should AMD try to get their GPUs into third party SoCs like Imagination Technologies (PowerVR), ARM, and Vivante do? Let us know in the comments below!
from xda-developers http://ift.tt/2clK1u2
via
IFTTT
