At this year’s CES, Intel demonstrated for the first time the Xe_LP architecture unique card for the development code DG1, and said it had shipped it to core developers. The DG1 features a short, compact styling with no external power supply, a single fan, and is said to consume less than 25W in the GPU’s design.
Pictured is the DG1 development card
The latest reports from the Indian media said that the second “DG2” unique show is also under development, the use of TSMC 7nm process, positioning a lot of high-end.
From the paper analysis, DG2 foreign commission TSMC contract workers, must want to launch earlier, can be reported that the launch of DG2 is 2022. You know, Intel’s own 7nm is ready in the fourth quarter of next year, and such a product strategy is a little weird.
Of course, sources say Intel’s 7nm will be used mainly for the CPU, and the GPU currently only knows the “Ponte Vecchio” of the Xe_HPC architecture. If so, it means that Intel has to turn to TSMC for the sake of capacity balance. In fact, tSMC is not the first time it has received a single from Intel, having previously worked on a baseband, Atom, and chipset.
Currently, the DG1 has published a minimum performance level of 30FPS under Star Wars, which appears to be the 2012 GTX 650. I don’t know if the so-called DG2 can have the power of the World War I GTX 1060/RX 480.