Intel details Iris Xe Max dGPU for Tiger Lake laptops — no SLI-esque features for gaming but Deep Link for content creation and AI

[Redaktion]

Intel has detailed some of the features of its new Iris Xe Max DG1 discrete GPU solution for Tiger Lake laptops. Xe Max works together with the Xe iGPU in Tiger Lake processors to offer accelerated AI and media encode performance for creative workflows. While it does not work in an SLI-esque mode for gaming, Intel says Xe Max does offer great 1080p gaming performance that rivals an NVIDIA GeForce MX350 for thin and light gaming notebooks.

https://www.notebookcheck.net/Intel-details-Iris-Xe-Max-dGPU-for-Tiger-Lake-laptops-no-SLI-esque-features-for-gaming-but-Deep-Link-for-content-creation-and-AI.501307.0.html

[Tov]

They should compare it with MX450.

[béla]

I agree. They should've tested against the MX450 but right now there none laptop with such a dGPU.

We have to wait and see what they can do.

[John Meier]

Not surprised tbh, we've seen Radeon Dual Graphics in many laptops from 2013-2016

Doesn't work very well

[JayN]

another article pointed out that the chip also includes a full display controller, and so could drive another 4 displays. 

Maybe this will happen on the add-in card version.

It could also mean that the entire gpu and display controller could be shut down on the core chiplet while the exernal gpu chiplet is operating.  This would give the CPU more thermal headroom while processing.  Perhaps this is already part of the Dynamic Power Share decisions.

[JayN]

"making it easy for synergized workload sharing "

what does that mean?  The gpu chip has its own memory controller and apparently its own block of memory.  I believe the integrated GPU is on the ring bus and shares the CPU's memory controller.

[Vaidyanathan]

"making it easy for synergized workload sharing "

what does that mean?  The gpu chip has its own memory controller and apparently its own block of memory.  I believe the integrated GPU is on the ring bus and shares the CPU's memory controller.

What I meant was, since both iGPU and dGPU have the same kind of LPDDR4x memory, both the GPUs can work together to accelerate a particular workload that is distributed between them however the driver decides.