The R9 390 and RX 6800 both support DirectX 12 (feature set 12_0 and 12_2, respectively).
The RX 6800 has just over 3 times the theoretical performance, in areas such as TFlops, Pixel Rate and Texture Rate, and has a tdp of 250w, 25w less than the R9 390.
Then, there is the RX 6800 XT, which has just over 4 times the theoretical performance, in areas such as TFlops, Pixel Rate and Texture Rate, with the same memory config and a TDP of 300w. But, it only costs 12.2% more than the RX 6800, based on the US reference model prices...
One weak spot of the RX 6000 series could be the Memory bandwidth, which is only 33.3% higher than the R9 390. However, the infinity cache could potentially double the memory bandwidth performance, if AMD's statement on this is correct.
Finally - does anyone think there will be a memory bandwidth bottleneck on the RX 6000 series GPUs?
The RX 6800 has just over 3 times the theoretical performance, in areas such as TFlops, Pixel Rate and Texture Rate, and has a tdp of 250w, 25w less than the R9 390.
Then, there is the RX 6800 XT, which has just over 4 times the theoretical performance, in areas such as TFlops, Pixel Rate and Texture Rate, with the same memory config and a TDP of 300w. But, it only costs 12.2% more than the RX 6800, based on the US reference model prices...
One weak spot of the RX 6000 series could be the Memory bandwidth, which is only 33.3% higher than the R9 390. However, the infinity cache could potentially double the memory bandwidth performance, if AMD's statement on this is correct.
Finally - does anyone think there will be a memory bandwidth bottleneck on the RX 6000 series GPUs?
Last edited: