Yesterday evening AMD surprisingly introduced a new Radeon VII ("seven"). The Roman Seven can also be interpreted as V(ega) II, because it is a Vega architecture (GCN5) shrunk to 7nm, which has been significantly extended in crucial places. Not only the registers were partly widened and the clock rate increased, but above all the memory throughput was raised to almost unbelievable 1 TB/s and thus more than doubled compared to the predecessor.
Such a high storage throughput is currently not to be found anywhere, not even in higher price classes. Current Nvidia cards up to 3000 Euro currently achieve a maximum of 672 MB/s with DDR6 RAM. Far more expensive models with HBM2 memory so far reach 900MB/s in the best case (with almost five-digit price tags).
The new AMD Radeon VII, on the other hand, will be available worldwide from February 7 with a generous 16GB of HBM2 memory for an incredible . AMD personally sees the computing power in the gaming sector at the level of an RTX 2080. At the same time, the Radeon VII should calculate 29 and 27 percent faster than a Vega 64 under Premiere and Resolve.
Since, according to our experience, memory throughput has a significant effect on video editing performance and a Vega56 expects an RTX 2080 at eye level is already available, the new Radeon VII under Resolve should probably play at least at eye level with an RTX 2080 Ti. If not even above...