Over the past month I have been running a lot of Linux benchmarks on Intel’s new Panther Lake using the Core Ultra X7 358H and its Xe3-based Arc B390 Graphics. The Arc B390 on Linux has been quite interesting with its OpenGL and Vulkan graphics performance compared to prior generations of Intel graphics plus the Intel Compute Runtime / OpenCL performance too. In today’s article are more benchmarks of the latter in looking at the Intel Rendering Toolkit and OpenVINO AI performance on the Xe3 B390 Panther Lake graphics compared to prior Lunar Lake and Meteor Lake.
Complementing the cross-vendor OpenCL benchmarks and the like from February, up today are benchmarks looking at how the Intel graphics performance have evolved from Meteor Lake to Lunar Lake and now Panther Lake when looking at Intel’s optimized software for their GPU compute environment using software from the Intel Rendering Toolkit (formerly known as the oneAPI Rendering Toolkit) including OSPRay Studio, OpenVKL, and Open Image Denoise. Plus benchmarks of Intel’s AI stack with OpenVINO and OpenVINO GenAI.
The recent generation Intel laptops I have on hand for this comparison included:
– Core Ultra 7 155H Meteor Lake within the Acer Swift SFG14-72T.
– Core Ultra 7 258V Lunar Lake within the Lenovo ThinkPad X1 Carbon Gen 13.
– Core Ultra X7 358H Panther Lake within the MSI Prestige 14 Flip AI+.
All three laptops were running a recent Ubuntu 26.04 development snapshot and using the Linux 6.19 kernel. Intel Compute Runtime 26.05.37020.3 and Intel Graphics Compiler 2.28.4 were installed on these laptops for evaluating the Intel graphics performance from the Intel Arc Graphics on Meteor Lake to now with the Arc B390 Graphics on Panther Lake.
The SoC power consumption was monitored in addition to the raw performance metrics for seeing how the performance-per-Watt has evolved as well.
