I left two comments today that I hope they address:
My older, slower RTX 3060 laptop got an Open CL score of 81694 in Geekbench 6. So, I am not sure why this one is so much slower. You are showing in the capture I am looking at OpenCL vs Metal. So, that is wrong. You showed the slide of the 4050 using the OpenCL score (a mighty low one) vs Metal on the 16″ Macbook Pro. In Vlulkan, my old slow 3060 got 73636. Somewhere, your testing has gone wrong and in favor of Apple. Very biased.
In 3D Mark Wildlife Extreme, my RTX 3060 to 82.73FPS with a score of 13815. Still way better than the 4050, which I find hard to believe. Also, “Unlimited” was misspelled in your chart.
[I didn’t add this part: the Samsung with the RTX 4050 scored a 11244 (67.3 FPS) and the Macbook Pro a 12985 (77.8 FPS).]
Wildlife Extreme
Description
3DMark Wild Life Extreme is a cross-platform benchmark for comparing the graphics performance of the latest Windows notebooks, Always Connected PCs powered by Windows 10 on Arm, Apple Mac computers powered by the M1 chip, and the next generation of smartphones and tablets.
Wild Life Extreme uses the Vulkan API on Android and Windows. On iOS and Mac, it uses Metal. On Windows 10 on Arm, it uses DirectX 12.
Graphics test
Wild Life Extreme measures GPU performance. With new effects, enhanced geometry and more particles, it is three times more demanding than Wild Life. The test uses a 3840 × 2160 (4K UHD) rendering resolution before scaling the content to the display.
Please note that Wild Life Extreme does not have multi-GPU support. It always runs on a single GPU.
My take
It should test equally regardless of the output resolution since it downscales it if it is less than 4k. Again, why is a RTX 4050 scoring less than a RTX 3060, especially one in a $900 (at the time I bought it) laptop? This is fishy to me, especially the comparing a OpenCL to Metal score when they clearly say in the slide “Metal/Vulkan”, which also are not comparable. Very different API’s. That’s what Wildlife Extreme will use because Apple will not embrace Vulkan and obviously since Metal is written with Apple macOS and the Apple Ecosystem in mind, it can’t (and doesn’t need to be) ported to Windows. There will never be parity here. OpenCL is the only cross-platform API and Apple and Microsoft may implement differently, so still not 100% parity here, either.
This is just wrong and very biased. I have an M2 Mac mini, a MacBook Pro (2018), a MSi GF65 Windows Laptop, and a custom built PC. I know how these things perform. It upsets me when a “trusted” source lies.
