ikjadoon
Senior member
- Sep 4, 2006
- 241
- 519
- 146
I can't read German so I didnt read the article, just quickly glanced theough it. But those numbers don't make sense at all. Are the AMD/Intel parts using CPU decoding? What codec is being used? VP9/AV1? CPU or GPU decoding? Browser used? Chrome/FF/Edge/Safari? The non x86 numbers align and are where they should be; really hardware decoding of 4k video should really only use around ~1WHr of power with the latest processors, however those x86 numbers don't look right at all. Makes it seems like the CPU is doing all the decoding at that point. Anyone who can read the article can offer more insight?
I also can't read German, haha. Luckily, the official English translation has now been posted.
But, without additional data, I'm also surprised at the Redmi & Schenker power draw. I don't have an x86 laptop left in the house these days to run a sanity check. Could something have gone awry with the iGPU → external monitor waking up too many parts of the SoC that would otherwise be asleep or, as you posit, was it forcing CPU decode somehow?
Are these laptops really burning 16-18W on YouTube videos? You've asked great questions and NBC did not provide much methodology for this new test they've introduced.
The only other questions I'd ask: is this over Wi-Fi? These don't use the same Wi-Fi chips. They also use different SSDs (and, IIRC, some browsers needlessly cache video playback, causing excessive writes).
These are more laptop comparisons than SoC comparisons, then, in an article mostly about an SoC, but perhaps they thought it was interesting to share.
I hope they keep their promise and add more laptops, however. I'd love to see a more traditional OEM's laptop shown on AMD / Intel, e.g., HP / Lenovo / Dell, as I'm less familiar with Schenker's and Redmi's designs.
I've tested with this video. On a 2019 Intel 16" i7 MBP. What NBC got is weird. I get a 12watts total core on a 14nm i7. Those x86 figures should be much lower.
View attachment 101503
Thank you for running this sanity check. I believe NBC is showing total system power draw with an external display. Would you know if you're using hw decode or sw decode here? Chrome has a way to check in DevTools once the video playback starts:
DevTools → Media → select the playing video and it'll show up on the right side. I can't remember which VP9 profiles are accelerated on Macs, as I remember Apple had that long spat:
NBC uses a flawed methodology. While understandable, measuring AC power means most laptops do not have all the power management turned on, thus you often see a difference in figure between actual battery life and power in W.
You'd see "Idle minimum" at 7W, but the laptop is getting 10 hours on a 70WHr battery or the same 7W under WiFi browsing, which isn't "Idle minimum".
That is interesting, yes. NBC's power figures do not seem reliable sometimes. This is where a clear methodology is quite important so at least it's reproducible but they also don't have that.
HWInfo64, I believe, has the battery drain power draw and I'd think reducing it to 150 nits or even 50 nits would've been a more interesting comparison than just one power figure with an unknown methodology.
Can't even tell if that is a peak and for how long they tested.
Those idle numbers show that there is certainly overlap between Intel & AMD designs and Qualcomm designs. The Redmi laptop (with MTL) has a lower idle than the ASUS laptop (Qualcomm 78).
Notebookcheck tables for ST and MT look great actually. So what is the catch?
The 1T performance is quite good & the power looks nice compared to Intel & AMD. To get the 84 1T perf, AMD & Intel virtually demand a gaming laptop. But, again, very very few laptops ship with the 84. At least one major laptop only ships with the 78 (HP OmniBook X).
However, the picture is a little more murky depending on what device and may only matter depending on your expectations: the ASUS Vivobook cuts 1T by over half automatically on battery by default; some laptops only come with the slower 78 SKU (e.g., HP Omnibook X); virtually no laptops come with the 00 tier that Qualcomm used in its launch presentation; zero laptops nor tablets (!) are fanless; compatibility is not as seamless as first claimed (actually in the reddit thread linked earlier); lots of little bugs still in shipping systems (e.g., Just Josh livestream showed one laptop suddenly falling asleep → the webcam couldn't wake up properly); the GPU drivers are rough and, as of this week, can't even be re-installed as Qualcomm's installer errors out.
As @coercitiv mentioned, Qualcomm's CEO & technical marketing went overboard, showing off results that are not generally applicable: 3.2K 1T GB6 scores that won't ever ship in a Windows laptop; "fastest laptop! Period!" when Cristiano well knew his devices won't ship for 7+ months; multiple scores revised down as the launch drew closer. But while I don't like this side of it, it really only affects us, the people that bother to listen to CEOs and technical marketing, lol.
Actual consumers just want to see how devices perform & how simple / easy it is to move to WoA device. Yet even that is still up in the air as reviewers only just got devices.
Last edited: