AMD Radeon Pro WX 8200 Review: Powerful, Affordable Workstation Graphics
AMD Radeon Pro WX 8200 Vega-Powered Pro Graphics
Take a gander at the Radeon Pro WX 8200’s main features and specifications below and then we’ll dig in a little deeper to see how the card compares to its little brother and a couple of NVIDIA Quadros...
|
GPU Architecture | Vega |
Lithography | 14nm FinFET |
Stream Processors | 3584 |
Compute Units | 56 |
Peak Half Precision (FP16) | 21.5 TFLOPs |
Peak Single Precision (FP32) | 10.75 TFLOPs |
Peak Double Precision (FP64) | 672 GFLOPs |
Memory Size | 8 GB |
Memory Type (GPU) | HBM2 |
Memory Bandwidth | 512 GB/s |
Memory ECC Support | Yes |
TDP | 230W |
Cooling | Active |
Board Width | Double Slot |
Board Length | 10.5" |
Board Height | Full Height |
External Power Connectors | 1x PCIe 6-pin, 1x PCIe 8-pin |
Display Outputs | 4x Mini-DisplayPort 1.4 |
Supported Technologies | Radeon VR Ready Creator, AMD Eyefinity Technology (Professionals), Radeon ProRender, Radeon Rays, Unified Video Decoder (UVD), Video Code Engine (VCE), AMD DirectGMA Technology, S400 Synchronization Module Support |
10-bit Display Color Output | 10-bit Display Color Output |
3D Stereo Output | 3D Stereo Output |
As we mentioned, the Radeon Pro WX 8200 is based on AMD’s Vega GPU architecture. As it is configured in the Radeon Pro WX 8200, the GPU sports 3,584 active stream processors (56 compute units) capable of delivering up to 10.8 teraflops of single-precision (FP32) performance, and 21.5 TFLOPs of half-precision (FP16) performance. That level of performance is just a tick behind the pricier Radeon Pro WX 9100, a higher-end workstation card that delivers 12.3 TFLOPs of FP32 and 24.6 TFLOPs of FP16 performance, but is priced significantly higher. The Radeon Pro WX 9100’s GPU is outfitted with a full complement of 4,096 stream processors, which accounts for its higher compute performance.
The Radeon Pro WX 8200 also wields 8GB of HBM2 memory, linked to the GPU over a 2,048-bit interface, with ECC support to ensure accuracy in mission critical situations. The memory bandwidth on the card is a beefy 512GB/s, which is actually higher than the pricier WX 9100’s 484GB/s. Part of the reason for the Radeon Pro WX 9100’s higher cost is its 16GB of HBM2 memory. The 8GB on the WX 8200 obviously brings costs down by halving capacity, but it is clocked slightly higher, resulting in the increased peak memory bandwidth.
Like the other members of the Radeon Pro WX family, the 10.5” long WX 8200 is outfitted with a blower-style cooler, with an understated, but beautiful blue fan shroud that’s adorned with nothing but the card’s model number. Though some of the lower end-cards in the family are single slot, the WX 8200 is two-slots wide and it requires a pair of PCI Express power feeds – one 6-pin and one 8-pin. The card has a 230W TDP (Thermal Design Power), so those two supplemental power feeds are more than enough to keep the card humming.
The Radeon Pro WX 8200’s display output configuration consists of a quartet of Mini-DisplayPorts (v1.4). It’s somewhat quizzical to not see any full-sized ports on a two-slot desktop GPU, but there’s no real downside, other than potentially having to use an adapter to connect up a display. Of course, the card supports AMD Eyefinity technology for multi-display setups, and can utilize all four of those outputs simultaneously. The card also offers 10-bit display color and 3D Stereo output, and it supports the S400 Synchronization Module as well. We should note that this is another area where the Radeon Pro WX 8200 differs from the WX 9100; the higher-end card has a total of five display outputs, a few of which are full sized.