Gaming performance and details about the Radeon VII

During yesterday’s announcement of the Radeon VII graphics card, AMD was skeptical about details about it, thus raising a lot of questions. Today, however, another dose of information about it has hit the network, focusing on the improvements that the Vega 20 core present on it offers compared to its predecessor (Vega 10).

So it is an improvement of the Vega architecture coupled with a descent to a lower technological process. The initial difference in the number of stream processors (4096 in the RX Vega 64 vs 3840 in the Radeon VII) suggested that something has been changed in terms of architecture, but in fact, the new Radeon was simply cut from CU blocks.

But let’s go back to the size of the GPU, because it measures only 331 mm², i.e. by 164 mm² smaller than that on board the Vega 64 model. By the way, its competitor from the Nvidia camp, the RTX 2080, already has a 12nm core with a size of 545 mm². However, we are concerned that the card offers 1.8 times more gaming performance, twice as much memory and 2.1 times more bandwidth in this comparison. The combination of 60 CU blocks with a maximum core clock of 1.8 GHz provides 13.8 TFLOPs in the FP32 calculations. A real bourgeoisie, we can, in turn, determine the presence of as much as 16 GB of HBM2 memory on a 4096-bit interface. However, it is really a very, very future-oriented move, considering that the titles in 2019 are to use these 11 GB of VRAM.

So let’s move on to a lot of tests in games in which the Radeon VII was compared to the RX 64 in 4K resolution. On average, we are talking about a 25% increase in frames per second, although in the most drastic cases (Fallout 76 here) the difference is as high as 68%. AMD even tried to test in synthetics, where you can see up to 65% increase between the two models. So we are waiting for independent tests, although AMD has already tried to adequately encourage players:

Also read: Nvidia’s CEO says the Radeon VII card is disappointing

Source: Wccftech