Intel DG2 (Xe-HPG) flagship GPU tested in Geekbench along with 16-core Alder Lake-S CPU
The first Geekbench results for Intel’s DG2 gaming GPUs have recently been spotted by TUM_APISAK. It looks like Raja Koduri’s team is hard at work testing the upcoming models, and hopefully the launch date is not that far away anymore. Just a few days ago, we were speculating that Koduri teased the DG2 connected to an Alder Lake platform and the Geekbench confirms this, as the tested GPU is coupled with an Alder Lake-S 16-core / 24-thread CPU.
From the specs presented in the Geekbench database entry, we can immediately identify this as the flagship Gen 12 DG2 model with 512 execution units. We even see that the GPU clock is set to 1.8 GHz and the test identified only 12.6 GB of VRAM, but previous leaks suggested maximum VRAM capacity should be 16 GB. We also see some specs for the Alder Lake CPU, like the base frequency set to 2.2 GHz and the maximum boost set to 4.6 GHz plus 30 MB L3 cache. Presumably, the boost clocks are reported for the big cores, as the small cores are supposed to be energy efficient and might not be able to hit high frequencies.
As for the Open CL results, these seem very low for a top-of-the-line GPU that is supposed to match the RTX 3070. For reference, the mobile RTX 3050 Ti with 4 GB of VRAM has recently been tested in Geekbench as well, and it scored around 60K points in the OpenCL tests, while the DG2 only scored a little under 8K. As we have seen with previously unreleased processors tested on Geekbench, we suspect these were stability tests rather than performance ones, as Intel is probably still working on driver support.