AMD’s Radeon RX 590 GPU is a refined version of the Polaris chip used for the RX 480 and RX 580. It’s manufactured at GlobalFoundries using a 12 nm production process, compared to the predecessor’s 14 nm. This has allowed AMD and its partner manufacturers to increase the clock rates while staying the same power envelope.
In the reference design, the maximum GPU (boost) clock rate is 1545 MHz, while the memory clock is 8000 MHz (effective). The reference card requires one 8-pin PCIe power connector from the PSU. However, many third-party RX 590 graphics cards need additional connectors. AMD recommends a 500 W power supply.
See also: our roundup of the best RX 590 cards.
The Polaris 30 chip has the same feature set and shader count as its predecessor. It supports Direct X 12 and Vulcan 1.1.101. Like in the older versions of the chip, there are 2304 shading units, 144 TMUs (texture mapping units), 32 ROPs (render output units) and 36 Compute Units.
Although it’s essentially the same GPU, the Radeon RX 590 performs noticeably better than its predecessor. This is a result of the higher clock rates allowed by the smaller production process.
On average, graphics cards with the Polaris 30/RX 590 offers about 10% better performance than the Polaris 20/RX 580. It’s also faster, on average, than the Nvidia GeForce GTX 1060. However, it’s noticeably slower compared to the GeForce GTX 1660 Ti in most games and benchmarks.
Power-efficiency is much improved from earlier versions of the chip, but still much worse compared to similarly-performing Nvidia graphics cards. The TDP (Thermal Design Power) for the Radeon RX 590 is 175 W.