The mobile PC market is about to get a whole lot competitive with the coming of the Ryzen 4000 “Renoir” APUs later this month. AMD is starting its journey in the mobile market with a bang. The Ryzen 7 models will be the first octa-core processor family in the low-power segment. On the other hand, Intel’s first 10nm lineup, Ice Lake, is limited to quad-core models. Quite the difference isn’t it?
In this post, we’ll be comparing the Geekbench scores of the Ryzen 7 48800U and the Intel Core i7-1065G7 and see how the two mobile processors stack up in single-threaded and multi-threaded tests:
Despite the 10nm Ice Lake part packing half the number of cores, its high IPC allows it to stay in the fight. In the single-threaded test, the former is nearly 10% faster than the Renoir APU. The tables turn in the multi-threaded benchmark and the 4800U leads by a lofty 50%, courtesy of its high thread count.
Overall, this is a clear win for AMD. The single-threaded performance may be slightly lower, but the multi-threaded score is vastly superior to the Intel competition. Considering that this is the company’s first major foray into the mobility market, that makes it even more impressive.
For the sake of relevance, let’s throw in the accompanying Comet Lake-U processors as well. Although based on the older 14nm Skylake core, these chips feature a higher core count (up to 6) as well as boost clocks in the vicinity of 5GHz!
Surprisingly, not only is the single-threaded performance lower than both the Renoir and Ice Lake parts, the multi-threaded score is barely a step up from the latter. There’s a delta of just 500 points despite the Comet Lake CPU featuring four additional cores. There’s only so much you can do by increasing the boost clock.
Now, let’s check the performance of the three chips across different types of workloads:
The Intel Ice Lake architecture has a major advantage in AES encryption while in the remaining integer and floating-point workloads, the two chips are mostly on par with each other.
The Comet Lake CPU is progressively slower across all workloads, with the main weak points being AES, ray-tracing, Gaussian Blur, and Face Detection. The Skylake core despite being nearly half a decade old bosts formidable machine learning and speech recognition capabilities. Looking at most tests, it’s clear that while the 14nm core is slower than the new Zen 2 core, the deltas aren’t as pronounced (in single-core tests) as you’d expect from a five-year-old design.