Return to

Intel dGPUs are coming


I’d suspect it’s a lot about machine learning.


The branding of the GPUs will be XE, and by the looks of the roadmaps we’ll have gaming gpus for sure.


They really need something better on the table than what GEN is putting out.

Linus was playing with a couple Larraby cards… I think he mentioned this stuff.


Intel just need something that is “good enough” that people can do the things they want to do with an integrated GPU.

I don’t think intel are going to catch up to AMD or Nvidia in the discrete GPU market in any hurry, but they desperately need a competent integrated graphics solution to compete with Navi or Vega.

It doesn’t matter if they can say, beat Ryzen by 10-12% in performance in mobile or OEM systems if AMD can ship Ryzen with a GPU that eliminates the need for a discrete mobile GPU, or a low end desktop GPU for OEM systems - if intel don’t have something equivalent.

If they have nothing to compete with integrated Navi, they’re going to be at a massive disadvantage in OEM / mobile systems.


Intel put out a LOT of awesome SfFy :slight_smile:

End of next year they get 10nm working or more delays.

7nm is in the pipeline and could come into the mix.

When Intel ships this tech. It will be cool and I may use it. I can see myself either Threadripper next year and make my 1700 a nas + vm machine. Mayby be a poor person and try a 8 core APU and passthrough my GPU for a window VM.

Intels not ringing my 2019 bell. I look forward to 2020 and what comes next but. Intel / Nvidia / AMD all GPU focused. 2020’s will be good for gamers. If only there was not a console leash.


Dude they need to catch up in a LOT of stuff right now. Its almost hilarious how incompetent they’ve shown themselves to be.

Depends on the market.

I don’t think that is their worry. A couple BIG HPC machines just got put together, one of the larger ones using EPYC, or at least the planning changed to EPYC. At that, their Knights line is kinda getting stopmed by GPU’s. If you look at whats going on, they really need a GPU offering to keep up. CPU clusters just don’t have enough Floating Point Units to keep up, unlike a Nvidia GPU, where all cuda cores are are FPU’s. They just can’t keep up.