Hacker News Reader

Comment by robin_reala | Hacker News Reader

The problem is that you can’t use both at the same time:[1]

To cut right to the chase on an important question for our more technical readers, Intel has not developed any kind of multi-GPU rendering technology that allows for multiple GPUs to be used together for a single graphics task (ala NVIDIA’s SLI or AMD’s CrossFire). So there is no way to combine a Tiger Lake-U iGPU with Xe MAX and double your DOTA framerate, for example. Functionally, Xe MAX is closer to a graphics co-processor – literally a second GPU in the system.

This basically seems to make it only useful as some sort of encoding/decoding co-processor, with potential but currently undescribed / undefined ML benefits.

[1] https://www.anandtech.com/show/16210/intels-discrete-gpu-era...

View on HN

SLI/CrossFire are dead technologies, the latest graphics APIs offer ways to use independent GPUs simultaneously without these.

The problem is neither the old or new method of multi-gpu has ever been really successful at actually working in the app you want it to.

> the latest graphics APIs offer ways to use independent GPUs simultaneously without these.

The problem is that DX12 or Vulkan require the application developers to implement multi-GPU capability.

SLI and Crossfire made multi-GPU support the responsibility of the drivers, and had at least the theoretical possibility of working for unmodified applications and games. That didn't work out too well in practice, but it was pretty obviously the only approach that had much chance of delivering widespread support.

DX12 and Vulkan are realistically, too low-level for ordinary applications to use - but engines like Unity and UE4 get to implement native multigpu without having to adversarially hack around driver multigpu behavior. Sure it has to be done in the application, but overall it will result in better outcomes for AAA titles.

For other apps using higher-level OpenGL/DX11 the existing multigpu drivers are still available - for now - but NVIDIA just want you to buy a single faster 3xxx-series card, that will almost certainly outperform implicit multigpu on an older card. They aren't updating SLI profiles after January, and aren't supporting them at all on the 3xxx series cards.

You can use dGPU for rendering and mirror frame buffer content over to iGPU mostly fine these days, as NVIDIA Optimus laptops do.

That you can’t do multi-GPU means you can’t combine GPUs to make them into a NUMA GPU cluster that is better performing than any of its nodes.

They expect users to pay twice for their GPUs - once for the integrated one and once for the dedicated one.

They are probably attempting this because for a long time OEMs would include both Intel's iGPU (not like they had a choice) but also an Nvidia GPU, even if it had the same or lower performance than the iGPU, just so Intel doesn't fully monopolize their devices.

Then again, by this logic not sure why any OEM would pay twice for Intel GPUs, either.

You are shadowbanned.

What makes you think that I can see them?

The reason you can see his comment is that I brought it back from dead status to make it possible to reply.

I don't think there was ever a case where the iGPU outperformed the discrete NVIDIA GPU for games. Maybe for encoding or something.

Reply on news.ycombinator.com