Intel hd 530 is combined with a rx 480 with multi function

In an online presentation during GDC 2020, Intel discussed the possibility of offloading the work of a discrete graphics card to the integrated graphics of a CPU with Direct3D 12 (D3D12). Using a multiple adapter involving asynchronous workloads, Intel simulated particles using Intel HD 530 integrated graphics with AMD's Radeon RX 480 graphics card.

Intel combines HD 530 graphics with a Radeon RX 480

Intel hosted a virtual event and created a repository of its GDC 2020 content. Several presentations are already online.

Among them is a function called "Multi-Adapter with Integrated and Discrete GPUs". Intel noted that discrete graphics cards often pair with CPUs with integrated graphics, and that it explored such "cases (asynchronous computing and post-processing) where Intel's latest integrated GPUs add significant performance" compared to usage. Exclusive to discrete graphics card.

This is accomplished by offloading some workloads to the integrated GPU of the Intel Core processors. Intel's methodology is to run the simulation (compute shader) on the integrated GPU, so graphics cards have more space to work in other areas.

Visit our guide on the best graphics cards on the market

Intel's proof of concept involved Microsoft's n-body D3D12 particle simulator, and simulated four million particles using Intel HD 530 graphics in combination with AMD's discrete Radeon RX 480 GPU. Intel did not say how it worked with just the discrete GPU. Intel did comment on the PCIe bandwidth using PCIe 3.0 x16: 4 million particles occupy 64MB, which means that the PCIe bus will be saturated at 256 Hz.

How does it work?

Intel said there are two ways to achieve a Multi-Adapter on D3D12. The first is the Linked Display Adapter (LDA). Here, the configuration appears as an adapter (D3D device) with multiple nodes, and the resources are copied between the nodes. Intel said this is typically symmetric, which means that identical GPUs are used.

The second approach is explicit multi-adapter with shared resources, which is what Intel did.

Intel also listed three possible uses for the Multi-Adapter. One is to share rendering, like toggling frames, but Intel said this is not suitable for asymmetric GPUs.

Another is to do post-processing on the integrated graphics, but this requires crossing the PCIe bus twice.

Finally, the third approach is to do "asynchronous computing" workloads , such as AI, physics, mesh deformation, particle simulation, and shadows, in embedded graphics. Intel considered this approach to be the best, where the PCI bus only intersects once. Also, it is advantageous if the rendering does not have to wait and if the computer is allowed to take more than one frame.

It would be very interesting to see this implemented to get more graphics performance with Intel Core processors that have an integrated GPU. We will keep you informed.

Tomshardwaremydrivers font