I didn’t stream them, though--I just played the files directly from the SSD using Movies and TV. In this first screenshot you can see the Kaby Lake XPS 13 playing the 1080p version of the 10-bit HEVC file. Because the GPU is doing all of the work, the CPU is basically idling.
With hardware acceleration of 10-bit HEVC video, the 7th-gen Kaby Lake Core i5-7200U is buttery-smooth playing both 1080p and 4K video.
In the next screenshot, you can see the same file being played on the Skylake XPS 13 in Movies and TV. It’s pretty apparent what’s using all the power on the Skylake XPS 13: The CPU is working hard to decode that 10-bit content, and with both cores running at 2.7GHz during the playback. What’s worse: The Skylake XPS 13 constantly drops frames even with the CPU running at near top speed. To put a finer point on it: The Kaby Lake XPS 13 can play the file at 4K resolution with 10-bit color and HEVC and still cruise, while the Skylake XPS 13 can barely play the 1080p version. Ouch.
Intel says you can expect similar results for Google’s VP9 video, too, using the Chrome browser. VP9 is a video codec Google supports, and how it encodes all videos on Youtube. I tried replicating Intel’s tests using Chrome on Youtube but didn’t have the network bandwidth at home to reliably stream two 4K VP9 videos simultaneously (one to each laptop). I see no reason to doubt Intel’s claims, though, as the claim about 10-bit content in HEVC is pretty crystal-clear.
One thing that should be pointed out: 10-bit content is very rare today. One area I think it does matter though, is in HTPC use. If I were building or buying a machine to run my 4K-and-up television, I’d want Kaby Lake over Skylake.
This 1080p HEVC file encoded with 10-bit color depth will drag a Skylake Core i5-6200U to the ground.
In the end, Kaby Lake is a decent step forward. Is it as exciting as reading about a 10nm-based CPU? No. But it does deliver a margin of improvement that’s actually in line with the evolutionary steps that what we’ve seen from the last few generations of chips. Haswell to Broadwell gave us maybe 10 percent. Going from Broadwell to Skylake yielded similar results.
Does it mean dump your Skylake laptop and rush out to buy a Kaby Lake laptop? No, not at all. However, when you’re looking at a 20 to 25 percent difference between Kaby Lake and Broadwell, then you start to wonder. As you get to Haswell, where it probably opens up to 30 to 35 percent, or Ivy Bridge and even Sandy Bridge, then yes, an upgrade would be a game-changer.
Sign up for CIO Asia eNewsletters.