AMD Ryzen 5 (2400G )/ Ryzen 3 2200G: the APUs back?

AMD Ryzen 5 (2400G )/ Ryzen 3 2200G: the APUs back?

After multiple variations of Zen as integrated on the Zeppelin chip, from laptop to server, AMD is launching a new variation of its new CPU architecture, this time adding an integrated GPU based on the Vega architecture. All codenamed Raven Ridge. For AMD, the situation is not particularly new, the manufacturer has been pushing the concept of APUs for a very long time, the integration in the same die of a CPU and a GPU. A vision inherited from AMD’s takeover of the graphics card brand ATI a dozen years ago.

The positioning of APUs on desktop PCs has often been complicated for the manufacturer, generally offering a rather interesting GPU accompanied by a rather light CPU compared to the competition. In the details, the question of placement and the interest of these platforms have long been questioned, limited functionality on the part of the processor is often a problem, but not the only one.

Added to this often was a limited interest in the more powerful GPU, because if AMD has always shone against the integrated graphics offer from Intel, and even against entry-level graphics cards, the use cases were quite complicated to define with a level of performance that remained, in absolute terms, light.

Beyond the endless HTPC, where graphics power needs have increased with the arrival of DirectX accelerated interfaces for example under Kodi, APUs have often presented themselves as an uncomfortable in between, a less powerful GPU sufficient for office tasks, and a more powerful GPU amply recommended for playing in good conditions.

As a result, the chosen design point and set permission (GPU embedding is not free) has long been a problem, exacerbated by CPU design problems and weaknesses in the production process.

A turnaround for Raven Ridge?

With Raven Ridge, you would think the situation has turned a bit around. We will not come back once again to the massive leap forward achieved by AMD’s CPU architecture, which is largely necessary, nor to Global Foundries’ 14nm, which is particularly effective if it is limited to a reasonable voltage level. It is on the creative side, whatever its use in Vega 56 and 64 where the maker was disappointed last year.