At this stage, patches for laptops are much more widely available than on desktops, especially from big name manufacturers. This makes it easier to test older hardware platforms, so today we’ll be looking at the impact on both the latest 8th gen Intel U-series parts, along with a 3 year old 5th gen Broadwell-based laptop. Both of the laptops I’m using for this test are Dell XPS 13, which should be representative of most premium ultraportable systems with 15W CPUs inside. The newer XPS 13 9360 model is powered by a Core i7-8550U with 8GB of RAM and a 256GB Samsung PM961 PCIe SSD. The older Broadwell model (XPS 13 9343) comes with a Core i5-5200U, 8GB of RAM, and a 256GB Samsung PM851 SATA SSD. As with desktops, patching these laptops requires two separate updates: a BIOS update specific to the device that tackles Spectre vulnerabilities, along with a recent Windows Update that kills Meltdown and supports the Spectre BIOS patch. We’ve tested the laptops in two configurations: before either update was applied and after both updates were applied. This will give us a good idea of how the performance differs between an unpatched system and a fully protected system (with the current set of patches). We should note that the performance differences you see here may not apply to all laptops with this sort of hardware inside, but it should give a good indication of how these patches affect a more performance-constrained system.
Let’s kick things off by looking at PCMark, which is a set of workloads designed to simulate real world tasks. Across the main PCMark 8 tests, there isn’t a significant impact from the Meltdown and Spectre patches, with performance declining by just a few percent in most cases. This is margin of error type stuff, and for most cases won’t be a noticeable difference in performance.
Cinebench R15 is an interesting one as we do start to see some performance degradation. The 8th gen platform seems to be more heavily affected here, dropping by seven percent in the multi-thread workload and three percent in the single-thread workload. Broadwell was still affected, but the difference was negligible in the single-thread test and just four percent in multi-threaded.
Cinebench isn’t the only rendering test that’s been affected. While rendering x264 videos in a two-pass encode, pass one performance dropped on both Kaby Lake Refresh and Broadwell, to the tune of 8 and 4 percent respectively. The more intensive pass 2, where most of the actual encoding occurs, isn’t significantly slowed on the i7-8550U but does suffer a marginal decrease on the i5-5200U.
Interestingly, rendering an x265 video in Handbrake with a single pass actually improves marginally in performance after the update, particularly on Broadwell where the render time is cut by 5 percent. So it’s not necessarily every rendering workload that is affected by the patches.
The final video rendering test I have is Premiere, which is affected by the Spectre and Meltdown patches. On the i7-8550U, both the Lumetri-effect enhanced test and the non-Lumetri test declined by around five percent, although the performance decline is slightly higher on the i5-5200U. On a performance constrained device like an ultraportable, seeing any performance slowdowns in Premiere is a big deal as most of these laptops run a fine line between being capable of editing videos, and delivering a choppy mess.
