Intel reveals their first discrete graphics card, already en route to developers

The graphics cards space has been immutable for decades, with Nvidia and AMD each taking turns at the top of the discrete market and Intel chipping away with their integrated offerings. Now, years after the first rumblings that things were about to change, Intel is sending out its first discrete GPU to developers. It's called the DG1, and it could have an impact far in excess of its modest frame. Two days after first revealing the card on stream, we have more details about how this card works - and what Intel is planning to do with it.

Let's start at the beginning. Intel's new graphics architecture is called Xe, and comprises of three microarchitectures that operate at different scales - Xe-LP for low power mobile devices, like phones, tablets and laptops, Xe-HP for workstations or gaming-class desktops and Xe-HPC for high performance computing - think server clusters and supercomputers. The DG1 is from this first class of devices, so don't expect desktop-class performance that rivals the best from AMD or Nvidia - that's coming later.

Rather than being a standalone release for gamers, right now the DG1 is intended mostly for developers to get to grips with the new Xe architecture. Intel's 11th-generation Tiger Lake mobile processors are driving the release, as these will include the new Xe-LP integrated graphics, which the company promises will offer double the performance of the Iris Plus graphics built into 10th-generation Ice Lake chips. (Intel also claim that the new CPUs will boast double digit upticks in CPU performance and a heavy boost to AI processing as well.) Of course, it can be hard to program applications to take full advantage of a chipset that hasn't been released yet, so Intel is sending out to the DG1 to independent software developers ahead of the launch of Tiger Lake. That should ensure everyone is up to speed with the Xe-LP microarchitecture, no matter whether it's built into a laptop or present as a discrete GPU.

Read more