Intel Xe integrated graphics are seen contesting Nvidia's discrete GeForce MX350
Rumor mill: Intel’s upcoming Tiger Lake mobile processors will be the first to showcase the Xe GPU platform. A leaked OpenCL Geekbench result shows the mid-range Core i7-1165G7 achieving an impressive score of 59,845, which puts it in direct competition with AMD’s and Nvidia’s finest in this segment.
In terms of raw horsepower, this Core i7-1165G7 result beats the average Nvidia MX350 result by ~17%. That’s a pretty nifty lead, given that the MX350 has 2GB of VRAM, and the G7 has no dedicated memory. The kicker is that they have the same TDP of about 25W (depending on the configuration) – but the G7 comes attached to a whole quad-core, octa-thread Willow Cove CPU.
Admittedly, though, the GeForce MX350 is a somewhat lame comparison. AMD killed the MX350 with the release of the Ryzen 4000 APUs, which offer better GPU performance and fantastic CPU performance at the same price. So how does the G7 compare to AMD’s stuff?
Intel should theoretically win, on paper. AMD’s GPUs have between 384 and 512 shaders and clock speeds around the 1600 MHz mark. Intel’s Xe GPU has up to 768 shaders and a clock speed of 1300 MHz.
In the OpenCL benchmark, the Ryzen 4600U reaches an average score of 38,000, the 4700U gets 40,100, and the 4800U gets 44,000. Even if the leaked ~60,000 result is an overperforming result, Intel beats AMD here, too.
However, both theoretical performance, and the OpenCL benchmark, are very general results. Intel has often maintained a lead in these categories in the past even when AMD’s hardware has been outperforming them in real-world tests. For now, the evidence suggests that Intel will at least be competitive, but there’s no winner at this point.
The elephant in the room is the CPU side of these processors. AMD seems to have a clear advantage in CPUs. Ryzen 4000 APUs have up to eight cores, while Tiger Lake maxes out at four. Can Willow Cove compete with Zen 2 with half the number of cores?