Your channel is one of the best things that happened to the internet recently. I've been here since you started talking about M1 machines, and I absolutely love that you have that scientific and mathematical approach to test everything. I trust a lot your information and I feel that this commitment to the truth is something hard to find
Which one is better in your opinion, the Apple M chip lineup - what's frequently reviewed and compared on the MaxTech channel, or the Snapdragon X Elite?
@@AZisk About the plugged/unplugged issue, you might want to check in the power options (8:07, controle panel) : change plan settings / advanced power settings / Processor power management, the max state on battery might be below 100%. Even worse, there is sometimes a third power settings : on my xps, I have My Dell and a thermal mode (cool, optimized, ultra perf, silence : processor and ventilator speed)
TLDR: This video is pure gold! Man... Everybody out there can buy a device, use it and tell "I (dis)like that thing for reason X, Y" and its ok because most of the other people don't care. They just need Laptop and they should just pick what people think is good. However, to make an actual good review for nerds who don't just want a device but want to nerd themself in the deepest rabbit hole ever, it needs so much more and there comes the problem: Besides all the private people who simply use the device and say "yes, that's what it looks like, I like it" there are also "professional" reviewers. Professional in the way that they earn money with it. But not so professional in a way that they understand what the actual heck is going on. They run some tests and benchmarks and put them in a list, sort by ascending and say "XYZ is the best product" and then other people come and say "benchmarks are useless I like another device more even though its lower on benchmarks" and that is true, if you use benchmarks like that. But, benchmarks are absolutely needed to measure and get objective results. And with benchmarks I don't just mean cincebench but aslo standardized procedures for special tasks from people who know what they are doing, who know how to use them and what is going on under the hood. And if these people do benchmarks and especially also interpret them you get something that is pure gold for a nerd like me. I don't have the money nor the time to do all this myself and I really appreciate your content because it is fun to watch, teaches you something and one day will help me deciding on the next product. Thanks Alex, I really appreciate it!
It feels like Samsung nerfed their machine to make it run longer on battery. Which is unfortunate, because a lot of people will buy them and decide they're rubbish and never go back to an Arm machine again. When my Lenovo arrives, first thing I'll be doing is checking performance and power management
Or maybe the opposite as they like the batter and don't notice the lack of performance. The Phawx video shows the performance and battery better then any other, you could Mach with an and chip and you could use the snapdragon battery in 1h. There are other tests that show that in a long stress test these chips go to 2.5ghz. Not samsung ,just any laptop with these chips would do it, that's how they get these batte live. These chips could destroy the battery in minutes if allowed to.
@@flyinginthedark6188 you might be right. Looks like they set it to run slower on battery and normal when plugged in. Seems a missed opportunity, though given other machines are able to get the performance both plugged and unplugged with reasonable battery life
Different approaches Like commercial HP elitebook they use low power limits and the fan never roar like hell All core benchmarks are crap compared to unlimited power limit model I prefer adjustable by user as sometimes so need more boost , sometimes I need it running silently ,however Samsung don't agree this 😂
The most important part of this video was the part where you said you have/will release a video on how to setup the qualcom version of Debian image on these machines. Subscribed.
@@richard.20000 Let's be honest, all Windows machines work better under Linux. I'm allergic to Linux, but it's reality. What I point out is that the better a machine performs on Windows, the better it is for Linux users. At least from the feedbacks I had from the few friends that do use Linux.
@@Belaziraf The only real exception I've found to this is brand new products being untested on Linux, but that's more a problem with the manufacturers themselves not implementing support before release.
@@ContraVsGigi yup all the open box deals u get are returns from TH-camrs who have probably ran all the tests and already sucked out the life of the hardware...
@@MindMorphMedia I mean if all it takes for a device to die is to run some benchmarks for a few hours, maybe a few days, then it isn't good to begin with.
@@TheZaman_ but obviously Windows (Executing the WSL) take also Resources. Who serves the CPU driver in this scenario I have no idea, that’s out of my expertise
@@ianneub9796 There are two WSL flavors: WSL1 is a linux-abstraciton-layer on top of win11, everything runs natively on the win11 scheduler; WSL2 is an actual VM where you have the entire linux virtualised. Basically, tests under WSL1 are as native as they come for compute tasks, but suffer in IO (disk/network/etc) due to win11 essentially emulating the linux IO (and win11 being objectively weaker than linux in some IO areas).
It is ridiculous that you have to tweak so many settings just to get the performance out of a Windows laptop. These laptops should have optimal performance and battery life out of the box. If Apple can do it, why can't Microsoft? It seems like Windows is put together by a bunch of teams who don't communicate with each other. Hell, Gnome and KDE desktops are much more coherent and polished these day.
Alex for the 8 BILLIONTH time - the i386 32-bit Chrome executable can only use 1GB memory, as an OS level limitation, and also can’t use PRISM SSE2/3/4 instruction set emulation. You NEED to get the amd64 release, otherwise the benchmarks get slaughtered.
I can’t believe as someone who says they are programmer doesn’t know about this. This is like the intro of computer architecture in a computer science program.
Subbed! Definitely didn’t see other celebrity YT-ers bother to dive in and show and appreciate actual performance and engineering of these new CPUs from Qualcomm. Every review these days feel sponsored or biased.
Thank you for the amazing explanation, and finally, straighten out some of the confusion. I most likely would by the machine for the nnpu so will be exciting to see how that plays out!
Hey Alex, In Vietnamese, teenagers have a word called "cuốn" or "dính," which describes something so captivating that we want more and more of it non-stop. This is exactly how I feel about your videos, and it's a term I often use when talking to friends about them. Regarding my preference at this time, your videos are amazing and the only best among thoudsands of vids out there explaining the performance differences between laptop/processor models. It truly stood out. I also watched your live stream unboxing the Snapdragon, and at that time, I was curious about how to change power plans, which is sooo well-explained in this vid. Thank you for your fantastic content, and I wish you all the best!
Finally, a good video about the topic. Now we know why the scores are different between devices. I was going to buy one (with the Snapdragon X Elite 80 or a MacBook with the M3), but I waited for the reviews and they are either very pleasing or quite disappointing. In my opinion, the raw power of the Snapdragon X Elite 80 can be impressive and it has potential, but you'll run out of battery faster on the high-performance plan compared to the MacBook.
X Elite has similar IPC to Apple M1 (same chief architect, similar cores) however X Elite scales better with frequency (so it's enhanced M1). Almost 3000 pts in GB6 for laptop is very impressive (intel and AMD have no chance to match it without going over 5.0 GHz). X Elite is x86 killer. ( in Windows laptops ) Just some people didn't realised that yet. Maybe they don't remember P4 vs. K8 clash - low IPC speed demon vs. high IPC beast at moderate clocks. IPC always wins, especially in mobile/laptops.
The IPC of the X-Elite is far from that of the M1 and close to that of the M3. Especially considering that the X-Elite is on 4nm meanwhile the M3 is on 3nm. The scores of 2.9k Vs 3.1k ST @4.0 Vs 4.05 GHz from both chips should show you that the M3 is just slightly ahead and it's mostly because of the semiconductor architecture.
Great. So setting power settings on a Windows laptop requires sifting through layers and generations of UI like a digital archeologist, and unless you throw guids at the CLI you still dobt know what it’s going to do. You couldn’t have written a better advertisement for MacOS and Linux.
I've been told "you see, this is better though, because hard-to-find default settings will result in lower greenhouse gas emissions when the average user is stuck on "energy saver"!"
@0bsmith0 nope, Windows is the best and easiest to use OS. There is a reason why MacOS and Linux distros constantly have features as well as design mimicing the Windows. Windows just has everything anybody needs in an easy to use package.
Wow, this is really a deep dive into the power characteristics of these new machines! Thanks a lot for bringing this to the surface, so to speak! ;) What I"d really appreciate is a nice little graphical "utility" (what we called these things 20+ years ago) to quickly change between max. battery and high power modes! (Alex, a challenge for you?!)
Yes it dose, But there are a number of facts you have to take into consideration that usually nobody thinks actively. The Snapdragon X Elite 84100 (the highest SKU) variant can peak up to 100W of power during a burst load but does not sustain for long, whereas Apple M3 has a reported peak load of around 50W. This is due to the fact that M3 is a TSMC 3nm process node where all X Elite are TSMC 4nm process. Now to get a comparable perspective, the 3nm process node enables a 20% to 25% improvement in performance over the M2 chip, according to Apple’s claims. That still does not account for the 50% higher power demand for the X Elite. That is due to the fac that (my guess), the X Elite Orion CPU core is heavily based on the Phoenix CPU core architecture designed by Nuvia for the server space (before Qualcomm bought Nuvia). Though these Phoenix CPU cores were designed to be scalable, it still is not an entirely mobile/desktop focused CPU. Qualcomm heavily boosted the clock speed for adapting these CPU cores to be on par with the M3 (not the M3 pro or ultra). Given this is just the 1st Gen product, we will get much more efficient and comparable CPU s within a few generations. But these will still be parity between the M series and everything else as Apple usually book/block a full new gen TSMC process node for its own CPU, leaving the competitors to use older last gen process node.
Very great video, one of the best if not the best explaining the new X elite line. One thing I wish you mentioned is the battery life in these different power plans.
As a Mac using TH-camr, I’m now looking into trying out ARM surface windows machines.. this is doing my head in. Too finicky to get good performance with battery.
The level of detail and knowledge you have on display was so refreshing. So many "tech" channels don't actually focus on the behind the scenes, just raw numbers. Also, wonderful production value. You just earned a new subscriber.
They were not faking, not really telling the whole picture. Showing benchmark scores when only on high perf mode, and showing battery life only on battery saving modes, it's just deceptive.
It has to be said that a well tuned high performance mode can retain most of the performance without pushing too many watts. At least, this is how it works currently on intel machines from what i tested
If you guys are going to cheat cause you are sponsored by Qualcomm. Put the M3 in high-performance mode and compare apples with apples. Also, as he stated, making huge claims of battery life hours on battery saving and high performance on high performance is extremely misleading. The best way is to choose a balanced mode and run drain battery test doing a single loop task like Cinebench for heavy usage and light usage loop of videoplay back. Including the M3.
Amazing in-depth investigation! Never saw anyone talk about all those power plan/mode configurations and how they affect the performance! Nice job bro!!
But it's very liberating. In one click, you can make it shutdown the fan and run 100% silent mode, or push the chipset to the extremes and get some nice boost for multitasking and gaming. > For light tasks, choose efficiency. > If you don't want to think, set it to Balanced and it will do the job for you. > Want to get boring stuff done quickly (e.g., video rendering), then set it to Best Performance.
There's another power setting in the manufacturer apps also from what I've seen in other videos (asus has "myasus", samsung has their own also). I believe it controls the fan curves and tdp iinm.
Can you do full power normalized test at several power levels? It's good to know it can reach some good benchmarks scores, but without any power consumption data it doesn't give you the whole picture.
Yes, this is what is needed, otherwise it is highly misleading, the only times where this chip beats M2 series is where it consumes 2x the power which would kill the battery life and the other way round.
If you edit one of the plans, another configuration sets the CPU utilization. The default is around 40%. Did you run on something like this? It could be found in the advanced settings
Man, i've had enough headache with that power plan vs power modes when testing laptop CPUs. Good information btw, can't wait to get my hands on one of these X elite laptop :)
super useful video, many thanks, finally someone on YT clarified the naming schemes and power plans/modes, btw I noticed same power fckery with latest Xeons (2465X) which also heavily depend on what power plan+mode is set in Windows 11
I feel it so dumb that given all the fancy AI they've been talking about, the Windows users are now still left with a bunch of power plan, battery plan to bother with. Why can't I just turn on the machine and use it without worrying about what I should choose for my today work load (or even the next 30 minutes). Come on Microsoft.
@@plaintext7288 Did you even watch the video? It. Is. Not. That. Simple. Why are there hidden settings that are unavailable in the UI? Why are there two set of options? Why is there a difference between plugged in and not? (when in performance mode)
Congratulations! This is the best performance test of the Snapdragon X Elite I have seen so far! Please also try running Linux on these machines and then Windows in a hypervisor.
Battery life is atrocious lets be honest. So far, with what they have proposed, I'm getting 8 hours of video playback, no other tabs open and its the 84-100 16 inch variant. Samsung: Let the battery callibrate itself for 2 weeks and see. Which means, you're outside of your return window her in the UK. Not getting anywhere near the battery life and the macbook, regrettably will always beat these. AND I HATE APPLE
I am getting similar battery life. 6-8 hours of light use with the 16 inch 80 elite edge. Nowhere near the 22 hours advertised. I have the surface laptop 7 arriving today and that has had much better feedback regarding battery life. Fingers crossed.
Andrew Marc David run a test with Surface Laptop, 14 inch, at 120 Hz, 40% brightness, 4k TH-cam video playback. The battery lasted 20 hours, 9 minutes and 19 seconds. Isn't that phenomenal?
@@arvandvarahram I still do not believe that as no one has replicated it and it seems impossible. Sadly, I have called him out too many times and he has blocked both my accounts! I would not take one review as gospel.
M3 has efficiency cores, X chips do not. M3 has 4 high power and 4 efficiency cores. X chips have at minimum 10 "powerful" cores. They claimed it is faster than an M3, and it isnt. Based off that alone, Qualcomm was at minimum deceptive about their claims. Disabling all power management features isn't delivering imo. Also this isn't even against M4 or the new Intel and AMD chips. Also I find it interesting that traditionally ARM processors have been heterogeneous processors, especially in mobile devices, with "big" and "little" cores, where as these chips are only 1 processor type. Could this design decision be as a result of the "technical" limitations of Windows and their scheduler not being up to snuff? Wouldn't surprise me. I am all for more than just x86 and having ARM/RISC-V CPU support, but Qualcomm is lying about so much stuff, especially the readiness of their graphical drivers.
@@jopansmark Yes they are in fact lying about many things. They stated "flawless" game compatibility, yet many games refuse to launch or have graphical artefacts, not to mention that it is slow. They said "faster than an M3", and in many cases it is slower, despite having more cores. Additionally they stated 40% better battery life in some workloads, but it is on par with many last gen AMD 8000 chips in those same workloads. Although improved, MSFTs Prism emulation doesn't work with apps that require AVX, which is common in many apps. These ARM chips are not what they promised to be, simple. Early adopter hardware. Is it promising? Yes. But it is incomplete. Drink the Kool-Aid if you want, I'm sure Microsoft would like that.
@@OnlyCitrus I saw people complaining about stuff not being supported. Can I use popular code related tool? Let's say Android studio flutter, web dev tools or python side data science and analysis or pytorch? I want to buy but I'm scared
microsoft doing microsoft stuff. they complicate stuff when it doesn't have it to be that way. Just putting it out as an option and having it do what it says would be so much better. thank you, you just got a sub
Great video. Most tech channels are ignoring the massive restructuring in tech driven by AI. While benchmarking Qualcomm chips is interesting, it's crucial for PC and Android users to understand how dependent their manufacturers are becoming on Nvidia. Reports suggest Nvidia is charging $30-40k per chip, costs that will inevitably be passed on to customers. Apple surprised the industry and now stands out among competitors by creating their own chips and developing a unique, privacy-centered LLM architecture. This could shield Apple from the financial impact of Nvidia's pricing. It's hard to see how Nvidia’s AI pricing won’t eventually disrupt the PC market.
"creating their own chips" That would be years away. Apple car would probably be a reality first. "unique, privacy-centered LLM architecture." LOL. I wonder how does Xi Jiping feel about that.
@@moozillamoo2109 I don’t know what you mean here. Apple Intelligence refers to Apple's in-house development of AI technologies, including their own machine learning models and architectures. Unlike many companies that rely on Nvidia's GPUs for AI processing, Apple designs its own chips, such as the M1, M2, and now M3 series, which integrate specialized AI and machine learning capabilities. This vertical integration allows Apple to optimize hardware and software together, ensuring performance, efficiency, and privacy without depending on third-party hardware like Nvidia's GPUs.
Typical iphone user, everything is streamlined and in a fixed set without any space for flexibility so your lazy ass can forget about it and know that it can still work
I tested a school laptop that has an i9-13900H, 8GB of RAM, 512GB of SSD, it scored 9 on speedometer. In another laptop it scored 4, but that laptop has a Pentium 4417U, 8GB of RAM and 1TB HDD/128GB SSD, I don't know what the problem is
And don't forget it half or even just third of the price like thise new laptops, even if it's a bit slower, still fine for daily usage (also with this inconsistent power profile, even Snaodragon users get the similar performance and they think woah.. because even Alex said for the 30 it's the first time he see that..)
@@jopansmarkwhat you yapping about? Shit OS cos its build upon the same thing over and over. I know its hard to do one from scratch but 6+ years is good enough for a new one
Great video! This is about the only one I've seen so far that's actually really informative about how these chips perform. I'm not really in the market for a new laptop right now but I'll definitely be keeping the 2nd or 3rd gen X Elite laptops in mind when that time comes based on the performance they delivered here. Hopefully the battery life is up to snuff though.
So from what I've understood with my experience on Windows, power mode is kind of a subset of power plan. Basically, power mode affects, well, the power given to the components from the battery. Best performance maximises the energy that the components are given, while best battery minimizes it. Power plan, on the other hand, affects not just this, but a lot more things (sleep/standby and hibernate behaviour, among others). The standard or balanced power plan is configured to work alongside the power mode settings in the Fluent settings app, where the performance settings vary based on your power mode, but all the other settings remain. However, when you force a plan that isn't available by default, that disables and overrides the power mode. From what I can tell, those power plans are meant for desktops, where power draw isn't as much of an issue as I have yet to see any laptop with the default windows high performance or maximum performance power plans, though I do see those on all non-OEM desktops I've seen. Moreover, for laptops, it's the manufacturer that creates the power plans and puts those on their copy of Windows, not Microsoft as the OEMs have better knowledge of the machine and what settings to use to get the best balance of performance and battery life out of the device. This explains the different names for power plans between the Surface and the Book4 Edge. I haven't seen this for power modes, as all the laptops I've seen have the same combination of Best Power Efficiency, Balanced and Best Performance. Perhaps it's something that Microsoft has changed with the arm64 version of Windows (allowing manufacturers more control over power mode names), or maybe Microsoft changed the names of the power plans for the new Surfaces. I genuinely have no idea about this, though I'd imagine the power modes are the same? Also yet again please don't take this as fact. I have no sources for this, just my hypothesis from what I've seen on different machines.
You summarised everything pretty well. I wanted to say this in every review of youtubers' comments. And you did it Alex..Hoo.. And I hope now I can use Windows ecosystem for development and not be ashamed about it. Graphics and NPU usage to be covered later?
This video made me a subscriber! Love the deep dive into these machines versus other reviewers who just did surface level coverage just for views. Keep it up!
I put my cpu in battery mode on 55 percent and in cord mode on 75 percent since then it stays very cool and under 5 watt since it was going nuts before and burning hot. It’s an almost 14 inch Acer tablet and has a removable keyboard with its own drive. without heat the excessive heat it’s much better now
Finally someone with sense has addressed the issue. I've seen reviews bashing the Samsung when doing performance tests without plugging it in or tweaking to override Samsung power saving settings
Thank you, good video, yes power plans are always different by OEM. I remember getting a configuration of x86 HP and Dell for the kids' college, each had about the same hardware, but the Dell blew away the HP when it came to battery life. What I found was the so-called HP "Optimized" power plan was the reason. I custom built a new plan like the Dell and both laptops had approximately the same battery endurance after that. That was about 5 years ago and I am sure I am going to be playing around with the config on my Surface Laptop 7 15" for a while to better optimize it.
Your channel is one of the best things that happened to the internet recently. I've been here since you started talking about M1 machines, and I absolutely love that you have that scientific and mathematical approach to test everything. I trust a lot your information and I feel that this commitment to the truth is something hard to find
Wow, thank you!
Very true
@@AZisk All cool but why Mac, man?
Seconding the statement
All he's doing just web coding. Calm down kiddo 🤣
Amazing video! Great work explaining the issues with the Book4 Edge and it's power modes, as well as the coil whine issue.
Thanks! Yep, the Surface is definitely a bit more stable and consistent than the Book4.
Which one is better in your opinion, the Apple M chip lineup - what's frequently reviewed and compared on the MaxTech channel, or the Snapdragon X Elite?
🙏
@@RealEverythingComputersapple M series is much more mature
@@AZisk About the plugged/unplugged issue, you might want to check in the power options (8:07, controle panel) : change plan settings / advanced power settings / Processor power management, the max state on battery might be below 100%.
Even worse, there is sometimes a third power settings : on my xps, I have My Dell and a thermal mode (cool, optimized, ultra perf, silence : processor and ventilator speed)
This is the ONLY video on which I trust... great, you are the man!
Wow, thanks!
Right, that dave2d dude praises stuff that isn't even out yet. What a shill
TLDR: This video is pure gold!
Man... Everybody out there can buy a device, use it and tell "I (dis)like that thing for reason X, Y" and its ok because most of the other people don't care. They just need Laptop and they should just pick what people think is good. However, to make an actual good review for nerds who don't just want a device but want to nerd themself in the deepest rabbit hole ever, it needs so much more and there comes the problem: Besides all the private people who simply use the device and say "yes, that's what it looks like, I like it" there are also "professional" reviewers. Professional in the way that they earn money with it. But not so professional in a way that they understand what the actual heck is going on. They run some tests and benchmarks and put them in a list, sort by ascending and say "XYZ is the best product" and then other people come and say "benchmarks are useless I like another device more even though its lower on benchmarks" and that is true, if you use benchmarks like that. But, benchmarks are absolutely needed to measure and get objective results. And with benchmarks I don't just mean cincebench but aslo standardized procedures for special tasks from people who know what they are doing, who know how to use them and what is going on under the hood. And if these people do benchmarks and especially also interpret them you get something that is pure gold for a nerd like me. I don't have the money nor the time to do all this myself and I really appreciate your content because it is fun to watch, teaches you something and one day will help me deciding on the next product. Thanks Alex, I really appreciate it!
🤩 amazing! i really appreciate it!
@@AZisk You are truly awesome!
@@AZisk I have a small request, if possible can you please try and see if you can run Linux on this laptop?
It feels like Samsung nerfed their machine to make it run longer on battery. Which is unfortunate, because a lot of people will buy them and decide they're rubbish and never go back to an Arm machine again. When my Lenovo arrives, first thing I'll be doing is checking performance and power management
They set it up as if it was an Intel CPU. Let's see how much battery life they got in exchange.
Or maybe the opposite as they like the batter and don't notice the lack of performance. The Phawx video shows the performance and battery better then any other, you could Mach with an and chip and you could use the snapdragon battery in 1h. There are other tests that show that in a long stress test these chips go to 2.5ghz. Not samsung ,just any laptop with these chips would do it, that's how they get these batte live. These chips could destroy the battery in minutes if allowed to.
@@flyinginthedark6188 you might be right. Looks like they set it to run slower on battery and normal when plugged in. Seems a missed opportunity, though given other machines are able to get the performance both plugged and unplugged with reasonable battery life
Different approaches
Like commercial HP elitebook they use low power limits and the fan never roar like hell
All core benchmarks are crap compared to unlimited power limit model
I prefer adjustable by user as sometimes so need more boost , sometimes I need it running silently ,however Samsung don't agree this 😂
I don’t think his Geekbench is right. My 14 inch edge 80SKU scores higher
The most important part of this video was the part where you said you have/will release a video on how to setup the qualcom version of Debian image on these machines. Subscribed.
Arm support shouldn’t be a problem now considering it’s been 8 months
Actually very impressive results. Thank you man, you're a legend 👏
Thank you for demonstrating Windows ease-of-use.
Please run Linux Geekbench 6 too. (X Elite should reach 3200 pts in Linux)
@@richard.20000Geekbench sucks. It always has been a mess.
@@richard.20000 Let's be honest, all Windows machines work better under Linux. I'm allergic to Linux, but it's reality.
What I point out is that the better a machine performs on Windows, the better it is for Linux users. At least from the feedbacks I had from the few friends that do use Linux.
@@Belaziraf The only real exception I've found to this is brand new products being untested on Linux, but that's more a problem with the manufacturers themselves not implementing support before release.
@@richard.20000 I have an x86 laptop that lasts above 10hrs on linux. I have no idea why i get 4 hr max sot on windows.
Different approach, different results
Nice effort for new POV.
Great Thxs Alex 👍👍
“I’ve bought them myself” see ya returning them back after review
Here in Europe you have 2-3 weeks to return a product you bought online. I think most youtubers return them.
@@ContraVsGigiExcept when you buy from Amazon. Then it’s 6 months
@@ContraVsGigi yup all the open box deals u get are returns from TH-camrs who have probably ran all the tests and already sucked out the life of the hardware...
@@MindMorphMedia not at all lol
@@MindMorphMedia I mean if all it takes for a device to die is to run some benchmarks for a few hours, maybe a few days, then it isn't good to begin with.
Would be interested to see a Linux Native (No VM) Test of these to see where the Linux Drivers are for this Chip.
i thought that Linux run natively in Windows through Windows Subsystem for Linux
@@TheZaman_ but obviously Windows (Executing the WSL) take also Resources. Who serves the CPU driver in this scenario I have no idea, that’s out of my expertise
@@ianneub9796 There are two WSL flavors: WSL1 is a linux-abstraciton-layer on top of win11, everything runs natively on the win11 scheduler; WSL2 is an actual VM where you have the entire linux virtualised. Basically, tests under WSL1 are as native as they come for compute tasks, but suffer in IO (disk/network/etc) due to win11 essentially emulating the linux IO (and win11 being objectively weaker than linux in some IO areas).
afaik, the bootloader is locked so you can’t install linuux as dualboot
They're not ready yet. Qualcomm is estimating they will have the key pieces up streamed by kernel 6.11. That should be a few months from now.
That is impressive results. Thanks man!
You bet!
To be honest, this is the first time I have been amazed by a TH-cam reviewer. You are just amazing.
Pretty neat you could use a PowerShell script to manage the power plan. A little burdensome but at least an alternative
i’m sure there are scripts out there to automate this
Yes ai is there
Here we go, Alex explaines goodness :):)
Basically if you have a mac, you wasted your money : throw it away and come to the good side. 🤣🤣🤣
This is the best video explaining the power options for the Samsung machine. Other reviewers had no clue why Samsung laptop had a low score.
That was the best benchmark review on these machines
It is ridiculous that you have to tweak so many settings just to get the performance out of a Windows laptop. These laptops should have optimal performance and battery life out of the box. If Apple can do it, why can't Microsoft? It seems like Windows is put together by a bunch of teams who don't communicate with each other. Hell, Gnome and KDE desktops are much more coherent and polished these day.
Alex for the 8 BILLIONTH time - the i386 32-bit Chrome executable can only use 1GB memory, as an OS level limitation, and also can’t use PRISM SSE2/3/4 instruction set emulation.
You NEED to get the amd64 release, otherwise the benchmarks get slaughtered.
this is why i am glad i am not a programmer
I can’t believe as someone who says they are programmer doesn’t know about this. This is like the intro of computer architecture in a computer science program.
@@peanutcelery based
crap i've been running the 32 bit variant, why is that being set as default 😢
@@peanutcelery nah bro not that many programmers learn about all the limitations of a legacy architectures, chill
Subbed! Definitely didn’t see other celebrity YT-ers bother to dive in and show and appreciate actual performance and engineering of these new CPUs from Qualcomm. Every review these days feel sponsored or biased.
Thank you for the amazing explanation, and finally, straighten out some of the confusion. I most likely would by the machine for the nnpu so will be exciting to see how that plays out!
Hey Alex,
In Vietnamese, teenagers have a word called "cuốn" or "dính," which describes something so captivating that we want more and more of it non-stop. This is exactly how I feel about your videos, and it's a term I often use when talking to friends about them.
Regarding my preference at this time, your videos are amazing and the only best among thoudsands of vids out there explaining the performance differences between laptop/processor models. It truly stood out.
I also watched your live stream unboxing the Snapdragon, and at that time, I was curious about how to change power plans, which is sooo well-explained in this vid.
Thank you for your fantastic content, and I wish you all the best!
Nice to see a deeper dive into the power thing. Would be interesting to see what happens to battery life with the "Ultimate Performance" settings.
Excellent job. Thanks for the effort
thanks! 🙏
Thank you for the amazing work ! looking forward to your battery test video
This is the first time I've come across your channel. It's refreshing to see a real analysis being conducted on these devices.
such a detailed video, much appreciated!
Thankyou sir, you're the only one on TH-cam who is showing benchmarks in detail while others are still talking about camera and display quality
Finally, a good video about the topic. Now we know why the scores are different between devices. I was going to buy one (with the Snapdragon X Elite 80 or a MacBook with the M3), but I waited for the reviews and they are either very pleasing or quite disappointing. In my opinion, the raw power of the Snapdragon X Elite 80 can be impressive and it has potential, but you'll run out of battery faster on the high-performance plan compared to the MacBook.
X Elite has similar IPC to Apple M1 (same chief architect, similar cores) however X Elite scales better with frequency (so it's enhanced M1). Almost 3000 pts in GB6 for laptop is very impressive (intel and AMD have no chance to match it without going over 5.0 GHz).
X Elite is x86 killer.
( in Windows laptops )
Just some people didn't realised that yet.
Maybe they don't remember P4 vs. K8 clash - low IPC speed demon vs. high IPC beast at moderate clocks. IPC always wins, especially in mobile/laptops.
wait till see what amd has to offer in a few weeks
The IPC of the X-Elite is far from that of the M1 and close to that of the M3. Especially considering that the X-Elite is on 4nm meanwhile the M3 is on 3nm. The scores of 2.9k Vs 3.1k ST @4.0 Vs 4.05 GHz from both chips should show you that the M3 is just slightly ahead and it's mostly because of the semiconductor architecture.
Buying mac is really bad. Only get it when you can install Linux or Windows on it.
@@marceelino you can install linux on the metal on apple silicon and windows on a virtual machine... can you install macos on a snapdragon?
Definitely the most thorough explanation of the power plans I've seen so far, thank you! I will definitely be using the powercfg trick.
Great. So setting power settings on a Windows laptop requires sifting through layers and generations of UI like a digital archeologist, and unless you throw guids at the CLI you still dobt know what it’s going to do. You couldn’t have written a better advertisement for MacOS and Linux.
I've been told "you see, this is better though, because hard-to-find default settings will result in lower greenhouse gas emissions when the average user is stuck on "energy saver"!"
It's literally one click on the battery icon to change the mode so seems pretty obvious to me.
@broimnotyourbro too bad MacOS is more limited OS.
Anything is better than Windows.
@0bsmith0 nope, Windows is the best and easiest to use OS. There is a reason why MacOS and Linux distros constantly have features as well as design mimicing the Windows. Windows just has everything anybody needs in an easy to use package.
I rarely comment on videos, but well made video. This is the first video about Windows on ARM that was actually useful.
thanks for doing these tests - am quite excited by Windows on ARM vs M-series
More to come!
Great explanation on the power plans!
wow thanks so much!
Excellent explanation about the power plan and power mode
Wow, this is really a deep dive into the power characteristics of these new machines! Thanks a lot for bringing this to the surface, so to speak! ;) What I"d really appreciate is a nice little graphical "utility" (what we called these things 20+ years ago) to quickly change between max. battery and high power modes! (Alex, a challenge for you?!)
Thx for this honest Review Alex
Quality content 😊😊
The best comparison I have seen so far
Does the Qualcomm chip consumes more power than the M2 & M3 to deliver the same performance?
Lol I think so
Yes it dose, But there are a number of facts you have to take into consideration that usually nobody thinks actively.
The Snapdragon X Elite 84100 (the highest SKU) variant can peak up to 100W of power during a burst load but does not sustain for long, whereas Apple M3 has a reported peak load of around 50W.
This is due to the fact that M3 is a TSMC 3nm process node where all X Elite are TSMC 4nm process. Now to get a comparable perspective, the 3nm process node enables a 20% to 25% improvement in performance over the M2 chip, according to Apple’s claims. That still does not account for the 50% higher power demand for the X Elite.
That is due to the fac that (my guess), the X Elite Orion CPU core is heavily based on the Phoenix CPU core architecture designed by Nuvia for the server space (before Qualcomm bought Nuvia). Though these Phoenix CPU cores were designed to be scalable, it still is not an entirely mobile/desktop focused CPU. Qualcomm heavily boosted the clock speed for adapting these CPU cores to be on par with the M3 (not the M3 pro or ultra).
Given this is just the 1st Gen product, we will get much more efficient and comparable CPU s within a few generations. But these will still be parity between the M series and everything else as Apple usually book/block a full new gen TSMC process node for its own CPU, leaving the competitors to use older last gen process node.
@@rezaulkarim7703 M4 will smoke Qualcomm chips easy. When M4Pro/Max comes out, it wouldn't even be competition.
@@RunForPeace-hk1cu You talk like that's something new. lol. year by year even on mobile qualcomm gets smoked. I'm an Android user.
@@RunForPeace-hk1cu You think apple engineers in Qualcomm would be sitting quietly ?
Finally! The variations in the scores are now demystified! Thanks a lot Alex!
I like your mic sound, may i know which mic are you using?
Sennheiser MKH50
Very great video, one of the best if not the best explaining the new X elite line. One thing I wish you mentioned is the battery life in these different power plans.
How much does the battery life suffer in normal use with the high performance power plan?
tbd - more tests coming soon
Very impressive dive and analysis with experimentation on the power modes and battery connected vs not in regards to the benchmarks.
Lovely, a proper review.
As a Mac using TH-camr, I’m now looking into trying out ARM surface windows machines.. this is doing my head in. Too finicky to get good performance with battery.
I love how you're able to flow between MacOS, Windows, and Linux, like butter. It must be amazing to have such options in front of you.
The level of detail and knowledge you have on display was so refreshing. So many "tech" channels don't actually focus on the behind the scenes, just raw numbers. Also, wonderful production value. You just earned a new subscriber.
This makes other reviews useless. No one ever spoke about the power plan they jus giving us numbers
amazing work!!! ur channel is VERY underrated
They were not faking, not really telling the whole picture. Showing benchmark scores when only on high perf mode, and showing battery life only on battery saving modes, it's just deceptive.
It has to be said that a well tuned high performance mode can retain most of the performance without pushing too many watts. At least, this is how it works currently on intel machines from what i tested
If you guys are going to cheat cause you are sponsored by Qualcomm. Put the M3 in high-performance mode and compare apples with apples.
Also, as he stated, making huge claims of battery life hours on battery saving and high performance on high performance is extremely misleading.
The best way is to choose a balanced mode and run drain battery test doing a single loop task like Cinebench for heavy usage and light usage loop of videoplay back. Including the M3.
That's what AMD and Intel do too btw
@@DerpSenpai that doesn't really matter cause x86 high performance laptops usually have like 4-8 hours of battery life. We are know they terrible
Great review as always! 🎉 Looking forward to dev review, around wsl, nginx, ollama, rustc, ruby and php benchmarks 🙏🏻
alex pls install a Linux distro as dual boot and test these systems
Their bootloader is locked :(
@@laloreta798 WHAT???
fuck them then i am not going back to windows
to get linux on there might take a little more work, still researching an approach
@@AZisk what kind of work?
Amazing in-depth investigation! Never saw anyone talk about all those power plan/mode configurations and how they affect the performance! Nice job bro!!
Power mode in windows very confusing 😅
But it's very liberating. In one click, you can make it shutdown the fan and run 100% silent mode, or push the chipset to the extremes and get some nice boost for multitasking and gaming.
> For light tasks, choose efficiency.
> If you don't want to think, set it to Balanced and it will do the job for you.
> Want to get boring stuff done quickly (e.g., video rendering), then set it to Best Performance.
@@1centimetre The option is cool, the issue is with usability. Microsoft just needs to make it simple and obvious like the Lenovo Vantage app.
And Microsoft doesn't event update the powerplan settings UI since like 2005
There's another power setting in the manufacturer apps also from what I've seen in other videos (asus has "myasus", samsung has their own also). I believe it controls the fan curves and tdp iinm.
Can you do full power normalized test at several power levels? It's good to know it can reach some good benchmarks scores, but without any power consumption data it doesn't give you the whole picture.
Yes, this is what is needed, otherwise it is highly misleading, the only times where this chip beats M2 series is where it consumes 2x the power which would kill the battery life and the other way round.
Oh my, you’re killing them all 😄🔥
If you edit one of the plans, another configuration sets the CPU utilization. The default is around 40%. Did you run on something like this? It could be found in the advanced settings
The High Perf mode sets MAX CPU Util to 100
Man, i've had enough headache with that power plan vs power modes when testing laptop CPUs.
Good information btw, can't wait to get my hands on one of these X elite laptop :)
Put linux on it.
get a life
I kind of like that you need basically run a virtual Linux machine to get stuff done.
Put Arch on it
That's how I'm going to be using my laptop
@@jonragnarssonno, native Linux, we don't want a reduced virtual machine running in the confines of Windows that's the whole point!
Awasome stuff I actually enjoy your reviews from a technical support point of view.
Goat 🙌🏼
super useful video, many thanks,
finally someone on YT clarified the naming schemes and power plans/modes,
btw I noticed same power fckery with latest Xeons (2465X) which also heavily depend on what power plan+mode is set in Windows 11
I feel it so dumb that given all the fancy AI they've been talking about, the Windows users are now still left with a bunch of power plan, battery plan to bother with. Why can't I just turn on the machine and use it without worrying about what I should choose for my today work load (or even the next 30 minutes). Come on Microsoft.
Just use battery saving until it becomes a limiter
👆🏾👆🏾👆🏾👆🏾👆🏾
Exactly!!!!!
@@plaintext7288 Did you even watch the video? It. Is. Not. That. Simple.
Why are there hidden settings that are unavailable in the UI? Why are there two set of options?
Why is there a difference between plugged in and not? (when in performance mode)
@@hundvd_7 bro isheeps want everything from default , so usually apple doesn't provide customisation choice 😂😂
How stupid and lazy are you? 2 clicks on the task bar are hard to do?
Congratulations! This is the best performance test of the Snapdragon X Elite I have seen so far! Please also try running Linux on these machines and then Windows in a hypervisor.
je parle français mais n'empêche pas de profiter de tes excellentes videos en anglais
W8ing for some dual boot and Linux tests, and ofc to see if basic stuff works😉
Poor Qualcomm, their chips seem good, but are dragged down by Windows; its power settings are a mess
how is something dragged down by something you can change yourself???? make that make sense?
X elite still has mediocre gpu though.
And yet the Air gets high scores *AND* good battery life from default, without any of that hassle
@@swapneelbehera260 still has about 10000 times more games to play though
@@kaltimoktober talk english please
Been waiting for more alex's content :)
I only exited to run linux on that laptop :/
Battery life is atrocious lets be honest. So far, with what they have proposed, I'm getting 8 hours of video playback, no other tabs open and its the 84-100 16 inch variant. Samsung: Let the battery callibrate itself for 2 weeks and see. Which means, you're outside of your return window her in the UK. Not getting anywhere near the battery life and the macbook, regrettably will always beat these. AND I HATE APPLE
I am getting similar battery life. 6-8 hours of light use with the 16 inch 80 elite edge. Nowhere near the 22 hours advertised. I have the surface laptop 7 arriving today and that has had much better feedback regarding battery life. Fingers crossed.
@NMCloud355 be good to know what there battery is like. Also can get a 32gb variant I believe
@@NMCloud355 legion slim 7 yoga is probably the best one
Andrew Marc David run a test with Surface Laptop, 14 inch, at 120 Hz, 40% brightness, 4k TH-cam video playback. The battery lasted 20 hours, 9 minutes and 19 seconds. Isn't that phenomenal?
@@arvandvarahram I still do not believe that as no one has replicated it and it seems impossible. Sadly, I have called him out too many times and he has blocked both my accounts! I would not take one review as gospel.
Thanks for covering this, I wish more reviewers would cover this and show us the battery in hours for power saver, balanced and performance modes.
M3 has efficiency cores, X chips do not. M3 has 4 high power and 4 efficiency cores. X chips have at minimum 10 "powerful" cores. They claimed it is faster than an M3, and it isnt. Based off that alone, Qualcomm was at minimum deceptive about their claims. Disabling all power management features isn't delivering imo. Also this isn't even against M4 or the new Intel and AMD chips.
Also I find it interesting that traditionally ARM processors have been heterogeneous processors, especially in mobile devices, with "big" and "little" cores, where as these chips are only 1 processor type. Could this design decision be as a result of the "technical" limitations of Windows and their scheduler not being up to snuff? Wouldn't surprise me. I am all for more than just x86 and having ARM/RISC-V CPU support, but Qualcomm is lying about so much stuff, especially the readiness of their graphical drivers.
10 powerfull cores need more power but still is efficient.
@@jopansmark Yes they are in fact lying about many things. They stated "flawless" game compatibility, yet many games refuse to launch or have graphical artefacts, not to mention that it is slow.
They said "faster than an M3", and in many cases it is slower, despite having more cores.
Additionally they stated 40% better battery life in some workloads, but it is on par with many last gen AMD 8000 chips in those same workloads.
Although improved, MSFTs Prism emulation doesn't work with apps that require AVX, which is common in many apps.
These ARM chips are not what they promised to be, simple. Early adopter hardware. Is it promising? Yes. But it is incomplete. Drink the Kool-Aid if you want, I'm sure Microsoft would like that.
1:50 thanks for recommending/mentioning these useful tools!!!
I want to know if I can code with it. Looks very promising.
It has really good performance for code compilation. Only GPU is lacking.
why would you not be able to code with it?
@@OnlyCitrus I saw people complaining about stuff not being supported. Can I use popular code related tool? Let's say Android studio flutter, web dev tools or python side data science and analysis or pytorch? I want to buy but I'm scared
@@ActionReaction.. toolchain compatibility, IDEs, cross compiling, etc
microsoft doing microsoft stuff. they complicate stuff when it doesn't have it to be that way. Just putting it out as an option and having it do what it says would be so much better. thank you, you just got a sub
Great video. Most tech channels are ignoring the massive restructuring in tech driven by AI. While benchmarking Qualcomm chips is interesting, it's crucial for PC and Android users to understand how dependent their manufacturers are becoming on Nvidia. Reports suggest Nvidia is charging $30-40k per chip, costs that will inevitably be passed on to customers.
Apple surprised the industry and now stands out among competitors by creating their own chips and developing a unique, privacy-centered LLM architecture. This could shield Apple from the financial impact of Nvidia's pricing. It's hard to see how Nvidia’s AI pricing won’t eventually disrupt the PC market.
"creating their own chips" That would be years away. Apple car would probably be a reality first.
"unique, privacy-centered LLM architecture." LOL. I wonder how does Xi Jiping feel about that.
@@moozillamoo2109 I don’t know what you mean here. Apple Intelligence refers to Apple's in-house development of AI technologies, including their own machine learning models and architectures. Unlike many companies that rely on Nvidia's GPUs for AI processing, Apple designs its own chips, such as the M1, M2, and now M3 series, which integrate specialized AI and machine learning capabilities. This vertical integration allows Apple to optimize hardware and software together, ensuring performance, efficiency, and privacy without depending on third-party hardware like Nvidia's GPUs.
Brilliant work done by explaining the discrepancy between benchmarks. Seriously impressive!
This is why people hate windows….9 different power settings…holy crap !!!!
i appreciate the flexibility to dial in what my needs are over the one size fits all approach of apple. the technical part appeals to me
eh they're in a transition period, but I tend to agree. They're slowly getting rid of all the control panel stuff tho...
As a Linux user, I find the number 9 being on the low end...
@@Prakyya transition period that started with windows 8? How many years has it been?
Typical iphone user, everything is streamlined and in a fixed set without any space for flexibility so your lazy ass can forget about it and know that it can still work
...who writes these jokes?". You gained a subscriber with that one. ❤️
Test Android studio please
not supported at moment sadly
@@BrandonLackey even with prism? I guess it will be available in arm pretty soon, as it already exists for Mac
@@Giggs995 Android Studio ran but not the android emulator... got an error.
Aw yeah! You finally made it on Techlinked brother. Hope it adds more growth!
really? haven’t seen it yet. will check it out!
My m1 MacBook Air gets 27 in speedometer 😂😂…it came out 4 years ago
And the galaxy book gets 30.... While m3 macbook air gets 29... What is your point?
Best laptop ever made.
I tested a school laptop that has an i9-13900H, 8GB of RAM, 512GB of SSD, it scored 9 on speedometer. In another laptop it scored 4, but that laptop has a Pentium 4417U, 8GB of RAM and 1TB HDD/128GB SSD, I don't know what the problem is
And don't forget it half or even just third of the price like thise new laptops, even if it's a bit slower, still fine for daily usage (also with this inconsistent power profile, even Snaodragon users get the similar performance and they think woah.. because even Alex said for the 30 it's the first time he see that..)
@@TamasKiss-yk4st I did 4k video editing and other work on WIFI for a full day easily.
The BEST review!!! Amazing job!
The kryptonite of these machines will be windows itself…
I'll just install linux.
was waiting for this, thaqnks
Great chip. Shit OS.
@@jopansmarkwhat you yapping about? Shit OS cos its build upon the same thing over and over. I know its hard to do one from scratch but 6+ years is good enough for a new one
@@jopansmark built upon windows 10 but then again i am a "special" person so what do i know
@@jopansmark i don't use MacOS lol. I'm ranting about the thing that i use
Great review! Waiting for more detailed analysis on especially Book4 Edge.
I knew waiting for your video on these would be worth it. My favourite laptop reviewer.
finally an engineering perspective on X Elite, very helpful!
Great video! This is about the only one I've seen so far that's actually really informative about how these chips perform. I'm not really in the market for a new laptop right now but I'll definitely be keeping the 2nd or 3rd gen X Elite laptops in mind when that time comes based on the performance they delivered here. Hopefully the battery life is up to snuff though.
So from what I've understood with my experience on Windows, power mode is kind of a subset of power plan. Basically, power mode affects, well, the power given to the components from the battery. Best performance maximises the energy that the components are given, while best battery minimizes it. Power plan, on the other hand, affects not just this, but a lot more things (sleep/standby and hibernate behaviour, among others). The standard or balanced power plan is configured to work alongside the power mode settings in the Fluent settings app, where the performance settings vary based on your power mode, but all the other settings remain. However, when you force a plan that isn't available by default, that disables and overrides the power mode. From what I can tell, those power plans are meant for desktops, where power draw isn't as much of an issue as I have yet to see any laptop with the default windows high performance or maximum performance power plans, though I do see those on all non-OEM desktops I've seen.
Moreover, for laptops, it's the manufacturer that creates the power plans and puts those on their copy of Windows, not Microsoft as the OEMs have better knowledge of the machine and what settings to use to get the best balance of performance and battery life out of the device. This explains the different names for power plans between the Surface and the Book4 Edge. I haven't seen this for power modes, as all the laptops I've seen have the same combination of Best Power Efficiency, Balanced and Best Performance. Perhaps it's something that Microsoft has changed with the arm64 version of Windows (allowing manufacturers more control over power mode names), or maybe Microsoft changed the names of the power plans for the new Surfaces. I genuinely have no idea about this, though I'd imagine the power modes are the same?
Also yet again please don't take this as fact. I have no sources for this, just my hypothesis from what I've seen on different machines.
Can't wait for complete comparison
You summarised everything pretty well. I wanted to say this in every review of youtubers' comments. And you did it Alex..Hoo.. And I hope now I can use Windows ecosystem for development and not be ashamed about it. Graphics and NPU usage to be covered later?
This video made me a subscriber! Love the deep dive into these machines versus other reviewers who just did surface level coverage just for views. Keep it up!
Probably one of the most important discussions I've seen regarding the Snapdragon laptop performance concerns. Nicely done!
I put my cpu in battery mode on 55 percent and in cord mode on 75 percent since then it stays very cool and under 5 watt since it was going nuts before and burning hot. It’s an almost 14 inch Acer tablet and has a removable keyboard with its own drive. without heat the excessive heat it’s much better now
Finally someone with sense has addressed the issue. I've seen reviews bashing the Samsung when doing performance tests without plugging it in or tweaking to override Samsung power saving settings
Thank you, good video, yes power plans are always different by OEM. I remember getting a configuration of x86 HP and Dell for the kids' college, each had about the same hardware, but the Dell blew away the HP when it came to battery life. What I found was the so-called HP "Optimized" power plan was the reason. I custom built a new plan like the Dell and both laptops had approximately the same battery endurance after that. That was about 5 years ago and I am sure I am going to be playing around with the config on my Surface Laptop 7 15" for a while to better optimize it.