AMD is relaunching Radeon VII, with the designation “Pro”, as a graphics card intended for the corporate market. The GPU was launched in 2019 as the first 7 nm video card dedicated to the gamer segment, but has been discontinued because it consumes and heats up much more than its direct competitor, the RTX 2080.
Radeon Pro VII 16 GB HBM2 and 250 W TDP
The Pro variant of Radeon VII maintains some of the specifications of the model launched for games, but brings improvements from the maturation of the manufacturing process at 7 nm. The new GPU has the same 16 GB of HBM2 memory as the previous model, but has a TDP of 250 W, against the 295 W of its “predecessor”.
Regarding the frequencies, there are no major changes between the two cards, except for the peak clock, which was 1802 MHz on Radeon VII, and is 1750 MHz on Pro. Both GPUs have 3840 Stream Processors, with Pro has computational power of 13.06 TFLOPS in FP32 and 6.528 TFLOPS in FP64 (1: 2). In Radeon VII, these values were 13.44 TFLOPS in FP32 and 3,360 TFLOPS in FP64 (1: 4).
Up to 68% faster than Quadro RTX 5000
According to Videocardz, Radeon Pro VII was 68% faster, in a given benchmark, than its direct competitor, Quadro RTX 5000, even with 10 W less than TDP. While the Nvidia card still uses the old 3rd generation PCIe interface, AMD’s GPU already incorporates the 4th generation PCIe.
Another change between these two corporate GPUs is due to the maximum bidirectional bandwidth in multi-GPU: in RTX 5000, Nvidia NVLINK offers only 50 GB / s, while in Radeon Pro VII, the new Infinity Fabric Link provides a bandwidth up to 168 GB / s.
The Radeon Pro VII will sell for US $ 1,899 in the USA, approximately R $ 11,253 at the current price, and without taxes.