When Geekbench 5 runs the same CPU tests on my M1 Mac mini, the picture is quite different. My conclusion is that the resulting benchmark doesn’t fully assess the capacity of all eight cores, but it’s probably not far off. Load distribution is also fairly even and follows a similar pattern on each core shown.
The ‘multi-core’ tests only attain 100% briefly on all cores, but average well over 50% throughout, and were sufficient to bring the iMac’s fans up to speed. In fact the ‘single core’ tests are distributed across all eight cores, but look as if their total represents something approaching 100% load on a single core, confirmed by the figure given in Activity Monitor’s main window. Being an Intel CPU, the cores on the left with odd numbers are ‘real’, and those with even numbers on the right are virtual cores achieved in Hyper-Threading. In this case, Geekbench ‘single core’ tests were run for the period starting about a third of the way across each panel, then the ‘multi-core’ tests cut in just after half way, and are reflected on all the cores, until they complete and load drops to almost zero.
In each of these CPU History windows, time passes from left (oldest) to right (newest) for each of the panels, with red representing system load and green the app load. Here’s what I see in Activity Monitor’s CPU History window for a typical test run. My starting point is running widely used benchmarks in Geekbench 5 on the 8-Core Intel Xeon W processor in my iMac Pro.
#Geekbench mac m1 pro pro
Is it faster than equivalent processors made by Intel or AMD, and is an M1 Pro faster than the original M1? Over the last year, I’ve been looking at different ways of measuring this for Apple’s M1 chips, and this article and its sequels summarises some of the lessons so far. One of the first things you want to know about any new processor or chip with processor cores is its performance.
0 kommentar(er)
