GPUs

Will Intel’s Xe Graphics Cards Beat NVIDIA and AMD? It’s Complicated

Intel is one of the largest tech companies, and most certainly the biggest chipmaker. However, Big Blue is going through a rough patch. AMD’s Zen 2 based Ryzen 3000 and Epyc Rome CPUs have been annihilating the Intel competitors on all fronts. To make matters worse, it seems like Intel’s next node shrink (10nm) has been further delayed until mid-2020 for the server market. And god only knows when it’ll come to the mainstream desktop space.

Intel, however, has got a ton of projects in the pipeline. The Xe discrete GPU lineup is one of the most interesting ones. Things might not be going as well as they want you to believe though. The first and foremost problem is most likely with respect to the 10nm node. Given that volume production of Ice Lake chips has still not started, it’s highly likely that production capacity for the 1st Gen Xe chips is limited. This might push the release by a few quarters or so.

The second problem Intel has to face is pertaining to the driver software. Although they do have their iGPU driver, it will require significant work to become compatible with discrete GPUs. The Xe team isn’t starting from square one from it’s close enough.

The third problem is a more serious one: efficiency. AMD’s older Polaris and Vega cards based on the GCN architecture although they have excellent compute capabilities, they’re still quite power-hungry and run hot (despite being mature). Intel’s Xe GPUs will be anything but mature, so there’s a good chance that they won’t be power efficient. Imagine a GTX 1660-class GPU that draws as much power as the RTX 2080.

This will not only limit Intel Xe to the lower-end GPU space but also cause thermal problems. To top things off, unlike NVIDIA and AMD, Intel isn’t working with third-party board partners like MSI, ASUS, EVGA, and Gigabyte to sell their graphics cards. They have to do all of it themselves.

This means that we’ll be getting only the crappy reference design which will run hot and limit clock speeds and overclocking. It will take Intel at least 2-3 generations to catch up with AMD and NVIDIA (if the Graphics Odyssey doesn’t fall flat on its head before it actually takes off).

The recently spotted DG1 GPU is most likely a mobile GPU based on the Gen12 Xe design to go alongside Tiger Lake (11th Gen Mobile lineup). We are looking at 96EUs (Execution Units) or 768 shaders/cores (basically ALUs) running in tandem.

I’d reckon this GPU will most likely perform similar to a GTX 1050/1050 Ti. It’s still not entirely known how well Intel’s architecture scales with core counts and what the IPC is like. However, my best guess is that it’s lower than NVIDIA and AMD’s present architectures. Even if this GPU isn’t that good, it might just sell well if priced low. After all, most gamers look towards the budget graphics cards market when buying. The masses still game at 1080p.

Areej Syed

Processors, PC gaming, and the past. I have been writing about computer hardware for over seven years with more than 5000 published articles. Started off during engineering college and haven't stopped since. Mass Effect, Dragon Age, Divinity, Torment, Baldur's Gate and so much more... Contact: areejs12@hardwaretimes.com.
Back to top button