Well, I’m not really a “LisaLeaks” with my website, but the slides from AMD’s “Power and Thermal Data Sheet” for Renoir, which were kindly sent to me from a nice source today, do not lack a certain amount of explosiveness. To be fair, I have to preface this with the fact that it is a pure feature list of engineering samples, but their revision has progressed far enough to already draw direct conclusions on the final specs. Because really much will not change, if at all.
Nevertheless, the new Ryzen 4000 APUs, which are nothing more than a combination of a Zen2 core and a Vega GPU, are highly interesting as a current desktop counterpart and as an almost seamless transition to “Van Gogh” with Zen2 and Navi GPU and later “Cezanne” with Zen3. Whereby, it is propagated, one will already use LPDDR5 from “Van Gogh” on. But this is another topic again. I still have the wrong entries in the userbenchmark in mind, but an AMD datasheet should be more reliable.
As for the graphic units, according to the lists, there will be the 4-cores with 6 CUs (384 Shaders), the 6-cores with 7 CUs (448 Shaders) and the 8-cores with 8 CUs (512 Shaders), which in itself is not a real surprise. More interesting, however, is the clock rate, which could well promise decent increases in performance. The first table also contains some mobile APUs, but I didn’t want to tear this overview apart, but post everything in the original. The OPN can be found later in the three slides of the desktop versions:
In the following I have now the list of all six different 65-Watt and 35 models for you, whereby it is not quite clear which of the total of 12 versions will finally make it into the shops with which designation. But the clock rates are quite a hammer, especially with the 100-000000145 as 65-watt model.
Yes, all this makes me want more and I can’t wait to take an APU like this to its limits in real tests. Because for many applications, all this will already be quite sufficient. AMD has really put it to the test – all the way to Tiger Lake. What Xe and DG1 will be able to do, will of course have to be seen in the end. But at least the gene donor was the same with Raja-Koduri. It won’t take long, then we will know more at least with “Renoir”. And as long as AMD doesn’t end at “Colani” and tries to press round APUs into triangular sockets, the chosen path is surely the right one.