GPUDatacenterReleased Jan 10, 2020

AMD Radeon Pro V520

8 GB HBM2 · 2048-bit · 2,304 cores

A passively cooled, full-height, double-slot data center card built for SR-IOV hardware virtualisation, splitting one GPU across several cloud desktops, remote workstations and game-streaming sessions. It is the accelerator behind Amazon EC2 G4ad instances, introduced in December 2020. The 36 compute units are paired with 8 GB of full-chip ECC-protected HBM2 on a wide 2048-bit interface, which gives it 512 GB/s of bandwidth despite the modest capacity. Board power is 225 W and the card draws from a single 8-pin connector.

General

Manufacturer: AMD
Model: Radeon Pro V520
Segment: Datacenter
GPU chip: Navi 12
Architecture: RDNA
Release date: Jan 10, 2020
Price: Search on Amazon

Memory

Memory size: 8 GB
Memory type: HBM2
Bus width: 2048-bit
Bandwidth: 512 GB/s
Memory clock: 1,000 MHz

Clocks & silicon

GPU clock: —
Boost clock: 1,600 MHz
Shader units: 2,304 Stream Processors (36 CU)
Tensor / AI cores: —
RT cores: —
Transistors: —
Technology (nm): 7 nm
TSMC N7
TDP / TBP: 225 W

Interface & APIs

Bus interface: PCIe 4.0 x16
SLI / multi-GPU: —
DirectX: 12.0 (12_1)
OpenGL: 4.6
Vulkan: 1.1
Other APIs: OpenCL 2.2

AI support

Precisions: FP8FP16BF16INT8INT4TF32FP64
AI accelerator: None (no matrix units)
Peak AI TOPS: —
Notes: RDNA has no dedicated matrix hardware, so machine learning runs on the vector units. AMD rates the card at 7.4 TFLOPS peak FP32, 14.75 TFLOPS peak FP16 through packed math and 460 GFLOPS FP64.

All AMD GPUs →

Datacenter gpus from AMD announced around the same time as the AMD Radeon Pro V520. Add any of them to your compare list to see the specs side by side.

gpu