Tested: Ryzen AI 300 vs Snapdragon X Elite & Meteor Lake Efficiency

In this video Gordon tests AMD's Ryzen AI 300 series power efficiency versus comparable Intel, Qualcomm, and Apple laptops.

Subscribe to our PC hardware podcast The Full Nerd: @thefullnerdpodcast

=============
Follow PCWorld!
=============

This video is NOT sponsored. Some links may contain affiliate links, which means if you buy something PCWorld may receive a small commission.

Timecodes:
00:00 Intro
00:46 Test machine specs
03:40 Microsoft Office results
05:27 Cinebench SoC battery test results
08:32 WebXprt system power results
11:12 Chrome browsing results
12:06 Cinebench system power results
15:27 Mixed browsing/YouTubing results
25:04 Laptop screen / powered panel test results
32:30 MS Office perfromance battery test results
33:45 Windows 11 power mode AC/DC multicore test results
36:23 Windows 11 power mode AC/DC single core test results
38:10 Final thoughts

#amd #ryzen #laptop

[greasebob]

Hey Gordon, just wanted to say thanks for the testing and I hope you are feeling well!

[LeesChannel]

28+ hours of browsing on the Ryzen 300 (with an external monitor) is insane! You could fly to Japan AND BACK with the same charge!

[lubossoltes]

And I was thinking at the beginning: "Why not just plug in an external monitor to eliminate the internal screen?" And he did it. Ok one graph that I was missing was normalize the external monitor numbers to battery capacity.

[samsonadeboga]

So good to see and hear Gordon ❤❤❤❤❤❤❤

[johnh]

Great to see you Gordon! Best wishes on a speedy recovery! And as a software engineer, i'l like to see more Ryzen 9 16 core laptops ... right now the market is bare

[MrMartinSchou]

With the OLED vs External Panel, we can estimate the power draw of the panel, if we assume we used up the entire stated battery capacity in the runs.

Zenbook 14:
With the internal OLED panel, the device drew an average of 7.087 for 10.6 hours. 10. 6 hours * 7.087W = 74.98046Wh. That's 0.03% below 75Wh.
With the external panel, the device drew an average of 3.679W for 20.38 hours. 3.679W * 20.38h = 74.97802Wh. That's 0.03% below 75Wh.
That means the OLED drew 3.407W of power.

The OLED's power draw is 92% of that of the rest of the system.

Zenbook S16:
With the internal OLED panel, the device drew an average of 7.009W for 10.7 hours. 10.7 hours * 7.009W = 74.9963Wh. That's 0.005% below 75Wh.
With the external panel, the device drew an average of 2.655W for 28.25 hours. 2.655W * 28.25h = 75.00375Wh. That's 0.005% above 75Wh.
That means the OLED drew 4.354W of power.

The OLED's power draw is 164% of the rest of the system.

I couldn't find a reputable source, but chatGPT suggested somewhere between 20 to 50% more power draw for an OLED panel compared to an IPS panel. If we go with a 20% reduction in power draw, an IPS Zenbook S16 could be drawing 3.4832W for its display, for a combined total power draw of 6.1382W. With a 75Wh battery that would get us to 12.21 hours or 732 minutes - 7 minutes shy of the Surface Laptop 7.

[andyH_England]

I am glad you did a more realistic battery life test as I have tried a few X Elite laptops and they do not come close to the early sponsored reviews. I settled on a Surface Laptop 7 that gets between 8-10h SOT which is not game-changing despite the hype. But the main reason was standby, thermals, noise and its fast responsiveness. I certainly would not try and convince anyone to buy it on battery life alone, as M3 Macs, the new AMD and Intel are all competitively matched taking into account the variables.

Interestingly, I am seeing many returns of the Snapdragon, also Amazon has removed some of them maybe because of that, so I was probably not the only one misled by the barrage of sponsored reviews that sadly reflect unreal use cases. Samsung Book4Edge already have £200 off and a free £240 display to try and drum up business as demand dries up.

I would say that AMD and Intel are still likely to be better choices by the end of the year due to compatibility and pricing as their availability increases and deals become more widespread. The M4 MacBooks are also a choice for non-Windows diehards.

[HypnoticSuggestion]

The performance of these machines versus my MSI gaming laptop is insane, four to five times better battery life and CPU performance is better across the board. Wild. Thanks for the thorough comparison.

[jd]

Glad to see Gordon back making videos!!

[daghtus]

It's a bummer the Qualcomm based laptop featured an IPS screen. That most likely influenced the battery life quite a bit unless the OLED laptops used black color scheme across the board.

[mrgee]

You are totally right about the external panel scenario. USB-C vs HDMI uses a totally different PHY (physical layer) for the signal and the usb-c standard requires alot more power, ESD, and signal integrity protection circuitry, especially if the port allows power delivery above 10 Watts.

You probably needed a trio of comparisons with the external panel based on connection type, with snapdragon as a DNF for missing the HDMI port. Would be interesting to see how the other two fair with display over usb4/TB4 connection! Cheers!

[Mecrom]

2, 65W office 365 for ryzen ai is crazy. 1700 Minutes 27:00

[hyperhawk]

I hope you’re doing well Gordon, great video

[stevenliu]

Gordon, there is no HDMI alt mode over USB-C. An active DisplayPort-to-HDMI converter is required in those USB-C to HDMI adaptors, so your result makes perfect sense.

Edit: More precisely, there _was_ HDMI alt mode as the _standard_ did exist, but HDMI dropped support for it due to non-existent adaptation from manufacturers. At the time HDMI dropped support, they admitted that, as far as they knew, no adaptor implementing the standard was ever made (USB-IF later said the same).

Passive DP-to-HDMI does exist for DP++, but DP alt mode does not support DP++.

[lorenzo.c]

Just a note on the graph showing the WebXprt 4 run: instead of using phrases as "more spiky", I reckon a distribution would be more meaningful.
You set a reasonable bin width (2W? 5W?) and then, for each power segment, you show the sums of how many units of time each laptop spends at a certain power range.

[nexusyang]

Gordon: maybe one way to test these laptops that have different screens would just be to turn the screen off and test their power efficiency through an external display. Granted, driving an external display could generate a bit of power loss but it might better than trying to discern the power usage between laptops of varying types of monitors. Since win10/11 have the option to turn off the native display this might be a viable test.

Just a thought. :)

[tmutube]

That real world browsing battery test gives excellent information, thanks!

[DJ_CRIZP]

Can't wait for the lunar lake comparisons thrown in here. Hopefully team blue actually regains some ground here. It whould be nice to have a 'real' PC that can go toe-toe with SD for battery life and thermals. I would not mind a world where we have AMD as the less efficient but more powerful choice, and intel as the slower multi-core, but snappier single core and better battery life option. And hopefully Snapdragon grows up in the next few years to the point of be real PCs instead of the compatibility mess that they are now. Great content and information as always!

[ole]

Gordon, great discussion on OLED and the difference between the Office and Browsing benchmarks. Dark mode seems really important for the AI 9 to show its massive advantage in efficiency as compared to the 7940HS.

Unfortunate that an external display won't factor into those situations where battery life is important.

[nviorres]

Great test, I hope we see Lunar Lake laptops soon :)
BTW, I would take battery life over OLED any day.