A Tiger For Gaming Devs, Not Players
The Xe Max is not a step up from the Iris Xe when it comes to gaming, but when it comes to content creation this discrete card looks to take top spot. Interestingly it is only slated be available in Tiger Lake systems, which already feature the on-die Iris Xe which we have seen a bit of and which already outperforms NVIDIA’s MX350.
This enhancement to the on-die GPU runs at a peak of 1.65GHz with 96EUs and a 25W TDP which is completely separated from the CPUs power budget. They do share one thing however, the 4GB of RAM it uses comes from the system’s LPDDR4X-4266, which should reduce power consumption but lead to greater memory latency. This also means it is coming over the PCIe bus as opposed to being on-die.
Ars Technica reached out to Intel to get an idea as to how effectively this GPU will handle machine learning and similar workloads, to which they were told to expect the Xe Max to complete Gigapixel AI workloads seven times as fast as an MX 350, five times as fast as an RTX 1650, and 1.2 times as fast as a Tiger Lake laptop with Iris Xe graphics alone. This is a decent boost for a couple of hundred dollars for those that make use of it.
There is also good news for those that like to game on the same laptop as they work, the architecture is designed to recognize workloads and so will immediately switch to the onboard Iris Xe if it spots you taking a break to do a it of gaming, with no extra steps required from the user.