After the results of the Apple M1 CPU benchmark were outstanding, now it is the turn of the Apple M1 GPU benchmark to spread. How are the synthetic test results?
The appearance of the Apple M1 GPU benchmark results appeared on the GFXBench 5.0 page, where the Apple M1 GPU successfully ‘destroyed’ two discrete graphics cards from Nvidia and AMD, namely the GeForce GTX 1050 Ti and Radeon RX 560.
Previously, the performance of the Apple M1 GPU was still questionable. Although the CPU benchmark score is amazing, the GPU capability is unknown, as quoted from Toms Hardware, Monday (11/15/2020).
The M1 is built using a 5nm process, based on ARM, and on a system on a chip (SoC) is stored an eight-core CPU and an eight-core GPU (seven cores for the MacBook Air).
Not many specifications were revealed by Apple at its launch event. Unless the GPU has 128 executions units. Apple also said that the GPU has 2.6 TFLOPS capability, which is probably single-precision performance (FP32).
With this reference, the GPU M1 has the same power as the Radeon RX 560 (2.6 TFLOPS), and is only slightly less than the GTX 1650 (2.9 TFLOPS). Yes, an integrated GPU chip has the same capabilities as a discrete graphics card. If so, that would be pretty awesome.
This computational ability was then proven through the results of a benchmark test at GFXBench 5.0, which is actually not the best software for testing GPUs. However, currently only this application is available.
The M1 was tested through Apple’s Metal API, and its score far exceeds the scores obtained by the GTX 1050 Ti and also the RX 560. Keep in mind, both the GTX 1050 Ti and the RX 560 are not new GPUs.
But still, both are desktop class GPUs whose power consumption reaches 75 watts. While the Apple M1, although it is not disclosed how much power it consumes, of course it will be much more economical than the two GPUs.