A little over a week ago, the client computing space changed with the shocking announcement of an Intel CPU with integrated AMD graphics. This is not an error. In a bid to allegedly better compete with graphical market leader nVidia, the primary x86 CPU competitors teamed up to offer Intel processors cores and an appropriately-sized Radeon RX Vega GPU in a “package” unlike any prior processor launched by either manufacturer. Less than two days after this landmark development, Intel obtained a new employee which happened to be the former CTO of the Radeon Technology Group. The “sabbatical” that was announced right before the launch of the Radeon RX Vega GPU escalated quickly. Our initial concern with these strange bedfellows was twofold based on the fallout of the dissolved partnership between Apple and Imagination Technologies.
Apple relied on Imagination’s IP and solutions for multiple generations of the processors that power their tablets and phones. Eventually, Apple was able to build their own solution and no longer required Imagination’s services. Our initial perception of the Intel CPU/AMD GPU tandem in a singular package evoked thoughts of a comparable end game. Intel would use AMD’s offerings based on performance and cost until such time that an in-house GPU that offers a dramatic increase in performance would come to light. Who better to develop this than someone with extensive experience? Although the overlap between an AMD APU using the Vega graphics architecture and an Intel CPU paired with the same design may seem like a win for AMD, the reality for workloads that require high performance and minimal cores will result in the cannibalization of market share that could have been taken by the newer AMD APUs. The initial benchmarks on the only publicly available Ryzen 5 2500U-based system are interesting to say the very least. While it’s impressive to see what’s been achieved in a 15W TDP design, the lack of power optimization in comparison to Intel-based systems doesn’t necessarily close the gap in the areas that count most. Performance is certainly “good enough”, but the poor design decisions by HP around screen quality and, optimistically speaking, potential firmware that may lack the necessary optimizations for acceptable battery life don’t provide the slam dunk needed to take market share.