For VMworld, the AMD is launching a new virtual desktop infrastructure (VDI) solutions. The AMD Radeon Pro V340 uses two Vega architecture GPUs with a total of 32GB of onboard memory to support a multitude of users. We have known this is coming for some time, and have a few “not in the press release” details to share.
AMD Radeon Pro V340 32GB Dual Vega VDI Solution
The AMD Radeon Pro V340 is intended for enterprise VDI, desktop as a service (DaaS) and cloud gaming workloads.
The card itself is comprised of two Vega 56 style GPUs, each with 16GB of HMB2 memory giving each card a total of 32GB of HBM2 memory, with ECC, onboard. This is higher-end GPU memory than NVIDIA’s GDDR5(X) memory. Using SR-IOV based virtualization, these cards support up to 32x 1GB virtual desktops.
Software support is focused on VMware ESXi and vSphere. We would have liked to have seen Citrix Xen, Windows Server, and Linux (KVM) supported as well. The guest OS is focused on Windows but we hope this includes Linux soon.
Using AMD’s MxGPU technology and SR-IOV based hardware virtualization, AMD is able to virtualize graphics without having to utilize extra virtualization drivers in the hypervisor like NVIDIA uses.
That approach also means that AMD is able to utilize the standard Radeon Pro driver and does not have the driver checks like NVIDIA GRID. This has major impacts on cost as AMD currently does not have an NVIDIA GRID license model for GPU virtualization.
AMD Radeon Pro V340 Hardware Details
The AMD Radeon Pro V340 utilizes two 56 compute unit Vega GPUs for a total of 112 on the card. Each GPU has 16GB of HBM2 memory for a total of 32GB on the card.
Here is an expansion of the hardware and the cooling needed for this 300W add-in card:
AMD seems to be using a Microsemi SmartFusion2 ARM/ FPGA SoC as the security processor + a 48 port PCIe switch to help manage the complex infrastructure. We cannot share the photo, but the Microsemi SmartFusion2 is on the back middle of the PCB.
The AMD Radeon Pro V340 graphics card is expected to be available in Q4 2018 from through the company’s system partners.