Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Microsoft says new Surface Pro is faster than 15" M3 MacBook Air (macrumors.com)
81 points by mgh2 on May 21, 2024 | hide | past | favorite | 111 comments


> As for battery life, Copilot+ PCs support up to 15 hours of web browsing or 22 hours of local video playback. The MacBook Air models offer the same 15 hours of wireless web browsing, but only 18 hours of local video playback.

This is so underrated. I got a new macbook air as a temp while my m1 max gets repaired and it blows my mind how I had it unplugged for 2 days over the weekend and the battery is still going.

It's like range anxiety in cars, not worrying about hunting down my usb-c charger I left in some random room is nice


I have serious trust issues when microsoft makes a claim about battery life.


Yeah, I bought past versions of the Surface and the Surface Book. They would often never wake up from sleep or the battery would be dead unexpectedly. Hard for me to trust them after feeling like I wasted money multiple times.


The Microsoft ecosystem seems to invite it.

Even on an old, well-understood, and revered[0] laptop (the ThinkPad T530), I was having an unexpectedly-dead battery when running Windows.

It's a work laptop, and it is pretty much only used for occasional actually-portable computing work like programming of other devices in the field. It spends most of its time in the work truck, hibernated and unplugged -- sometimes, for weeks at a stretch.

I began to accept its increasingly-poor apparent battery health and started to explain it to myself as "Well, it is pretty old."

Or so I thought, anyway: One day it was sitting there on a table, unplugged, and I noticed that it came to life by itself and then hibernated again a few minutes later.

WTF?

Waaaay too much investigation later, I found that a part of an HP printer driver was forcing it to wake every couple of hours...for reasons that I don't care to explore, since none of those reasons could possibly have any positive merit.

Waaaay too much poking-and-prodding after that, I was able to disable the offending thing using powercfg on the command line.

And now, it seems fine. The battery life is not particularly good and never will be, but at least it's not "Surprise! I'm completely dead!!!" anymore.

This kind of sloppiness in software seems to be considered normal in the Windows space, and an abusive HP printer driver doesn't care if it is running on a 9-year-old ThinkPad or a 9-day-old Surface: It will abuse all of them just the same.

[0]: I hate numeric keypads, and gamer laptops, and computers that are built down to a budget as a primary design criteria. The T530 is approximately the last 15" PC laptop that lacks a numeric keypad, and that does not have stupid gamer-glam functions or styling, and that isn't built down to a budget price. It is stoic and plain and black, and both the keyboad and the touchpad are centered on the screen.


Macs have always had great battery life and the Apple Silicon chips took it to another level. It remains to be seen if any Windows laptop can actually consistently achieve their marketed battery life given Windows’ terrible track record for power management and proper sleeping.


Macs were long hindered by Intel processors to have long battery lifes AND/OR good performance. Two examples of MacBooks I had:

The 12" MacBook. Great form factor, bad software (graphics driver code was especially poor), poor keyboard, lousy battery life, slow CPU & graphics. Similar size now with the 13" MacBook Air, but much better.

The i9 MacBook Pro. Too heavy, not enough cooling, extra GPU chip (in the laptop) with lots of power demand, Thunderbolt chips with lots of power demand, CPU with lots of power demand, fans were often on (for example with anything which uses the extra GPU). Now a 16" MacBook Pro with Apple Silicon is virtually silent AND fast.

I currently have a Mac mini with M2 Pro, a MacBook Pro with M2 Pro and the new iPad Pro M4. All are very snappy and virtually silent. They use very little idle power and don't get warm in normal use. The i9 MacBook Pro would produce a lot of heat when attached to an external screen and doing a "simple" video call, due to the power hungry additional GPU needed to drive the external screen.


Yeah, no. Some macs had good battery life. The last Intel MBP was really bad though. (Less than 2h of basic desktop use and lots of overheating after a year)


I don't understand why battery life in gaming laptops is so incredibly shit even when doing non-gaming tasks. I thought a while back Intel and AMD worked on hardware/software to automatically switch to integrated graphics when you don't need the discrete GPU? Did nothing come of that?

I don't expect the battery life to be as good as non-gaming laptops, of course, but I feel like it should probably last longer than like an hour and a half.


Been like that forever.

Usually gaming laptops prioritize their cooling systems over battery size. And don’t underestimate the high refresh rate panel likely optimized for at least 500 nits.


> And don’t underestimate the high refresh rate panel

Literally one of the reasons Apple added variable refresh rates on their pro phones: if you can vrr down, you save a ton of battery when you run the panel at 1Hz.


I’m pretty sure any MacBook Pro with ProMotion can only drop to 24hz

“the MacBook Pro’s ProMotion display can range from 24Hz to 120Hz.”

https://9to5mac.com/2021/10/19/macbook-pro-promotion-tidbits....


Ah cool, didn't know that, I thought it was only in the phones and tablets.


It works like that on most laptops. The GPU selection should be transparent on windows. (And can be done explicitly in Linux)

I know Linux has powertop which can measure energy usage per subsystem. Not sure if windows has something like that to answer your question?


I don't know how it compares, but my recent AMD ThinkPad uses about 1% of the battery per hour while in "modern standby." I figure I have at least three and up to 4 days of being unplugged before it's forced to shut down.


Pretty sure GP was using their MacBook over the weekend. The MBA is supposed to last something like 30 days in sleep.

Yours is the sort of things I got with my Linux work laptop, and it’s miserable: a few years down the line battery degradation means after a 3 days weekend you have even odds the laptop will have slept itself to shutdown.


I assume they meant unplugged and in regular use, not unplugged and hibernating the whole time.


The system wasn't hibernating, it was just sleeping.


That doesn't change much, I guess. Still very little power usage compared to an actual turned on machine.


My work MBP has been neglected in a bag next to my desk for 2 weeks now. I just opened OmniFocus on my phone and it shows that my laptop last synced this afternoon.

That’s about normal for it.


> Copilot+ PCs support up to

Support? Big "hmmm" energy.


i hate ios and apple software quality is a total joke.

ill give it up for any actually good things tho, i did try a macbook air during a long trip and the battery life is truly excellent and worth dealing with the downsides


Very excited to see this. Microsoft Windows was always a bottleneck for us when it came to ad content delivery to Windows users. With this incredible performance of Surface Pro, we will deliver ads to those users at a speed they have never see before. By the time you close one ad, 3 others are already open. Thank you Satya for making this happen.


Hehe spot on. 58% faster ad delivery, in 58% more random places than you would have expected!


> As for battery life, Copilot+ PCs support up to 15 hours of web browsing or 22 hours of local video playback. The MacBook Air models offer the same 15 hours of wireless web browsing, but only 18 hours of local video playback.

I don't know how good the Linux support for this hardware is, but, if it works well, the reason of buying Macs for their battery life gets blown out of the water.


The track record is important: PCs have a track record of reducing the battery life to 15 minutes within 2 years through simple Windows updates; It’s probable that it will take 15 years to overcome people’s assumptions.


Windows battery benchmarks have always been wildly overoptimistic on Surface products. Factor of 2 to 4 in my experience. Various bugs and bloat cause them to run hot when idling, or in the case of my most recent purchase, when sleeping! In both cases the device had these problems from day one.

As for performance benchmarks, I don't care what numbers they get if it takes a full 8 seconds to open a context menu when I right click something. I'm even less enthusiastic if I'm forced to use Windows 11 and put up with ads.

Surface hardware is amazing, but if Windows doesn't lose all of this bloat then I am done with it.


> battery benchmarks have always been wildly overoptimistic

Assuming that’s accurate, isn’t that just straight consumer fraud?


I've always wondered this. For example I can't imagine how they ever achieved their quoted battery benchmarks with system processes running in the background, like indexing. These regularly consumed >50% CPU resources on my older Surface pro even on battery and were not easy to disable (I never did because doing so would cause other problems). But I can't see how you would achieve the quoted battery life without doing so...

If they do disable these things in their test, then that's super dishonest.


That's why battery claims are always UP TO not AT LEAST.


It still fraud (in my opinion) if you can never reach those numbers under any circumstances.

It's not "up to a million hours usage", because it's impossible to use it for a million hours.


>It still fraud (in my opinion) if you can never reach those numbers under any circumstances.

Who said they never reach those numbers? All manufacturers have disclaimers that the "UP TO numbers" have been achieved in their labs in their own controlled environment, and some even specify some the conditions: Pass Mark, wifi, power-mode screen brightness, etc.

Your car also specifies some very optimistic MPG/range specs that most owners will never reach but they were reached in a lab somewhere running on a dyno.


The root comment we're replying to here is saying that the previous model Surface was out by 2-4x.

I'm of the opinion that's getting into the realm of impossible numbers, absent evidence otherwise.

If you get it out of the box and it doesn't get what they specify - that's misleading at best, fraud at worst.

The car tests do specify the testing conditions, so I would expect that my car would perform the same under the specified conditions.


>The root comment we're replying to here is saying that the previous model Surface was out by 2-4x.

And how can we verify that he didn't have some power virus running in the background to generalize that 2-4x claim as being a universal fault with all devices?

The manufacturer makes a claim based on a (obviously biased) controlled lab condition. When you use the device outside that lab condition then it obviously won't apply anymore (duh).

The manufacturer can't promise an exact battery life on YOUR use case since everyone's use case will be completely different, and once you introduce external apps beyond the manufacturer's own control, then everything goes out the window.


Of course.

But if I get a new device out of the box, make no changes, and it doesn't get those numbers, then it's a lie.

That's been my experience before (not with a Surface), and it was OP's experience with a Surface.


Sure, the question is, have you tested the device fresh out the box and found a 3-4x less battery life than advertised in the same test scenario?

I guess that would be fraud, but that depends on what all the fine print on the battery claim states.


Personally, not 3-4x but I'd estimate 1.5x. But I never did a proper test.


I think hours of video call with five other people need to be a metric for devices now.


Does battery drain depend on how many people are on a call?


Yes, every person in the call has a video feed that must be decoded and rendered.


So does it take more battery to render several small video feeds versus one large one? If I'm on a call with a bunch of people, does "speaker mode" take less battery than "gallery mode"? Somehow I doubt it would be noticeable, if the effect exists at all.


Possibly, but it would have to be way better specified. You have dimensions like the OS, conference service, camera quality, browser, etc. that all affect it.


Hopefully it forces apple to up the base ram or storage. Competition is good!


There is no need to. 95% of users are never going to need to more than the base.

Desktop apps are increasingly moving to the web not the other way around.


But the browser wants all your RAM.


If the browser needs more than 8GB it will swap. Which will be completely unnoticeable with today's SSDs.


Software is a gas.


Yet it runs Windows which automatically negates any hardware advantage. Surfaces and other windows laptops have always had "good" hardware, but the software is just garbage.


Very interesting developments. I wonder how good X Elite's performance per watt is compared to the M3. Then again, Apple will soon restock all Macs with M4-level processors (reported, take it with a grain of salt). M4 beats the X Elite, but its not really a fair comparison since the X Elite is meant to compete with the M3. It seems like the X Elite is close to M3, but uses much more power [0].

[0]: https://nanoreview.net/en/cpu-compare/qualcomm-snapdragon-x-...


> Then again, Apple will soon restock all Macs with M4-level processors

Probably not soon. They just refreshed the macs with M3s, and put the M4s in the ipads pro, I assume due to TSMC yield issues with a bleeding edge process (yes, ipads outsell macs, but not the ipads pro, which is a low volume "halo" product).


If these numbers are representative, they seem to show apple silicon's efficiency advantage is now pretty minor.


The power efficiency numbers in that link do not appear to be derived from anything; definitely not actual measurements. The score out of 100 is almost certainly complete bullshit, and the site as a whole probably is worthless.


Do they say that when they developed it or after when they find out they cannot beat the M4 that they are not competing with something they cannot beat?


Those chips are in the pipeline for years. There's a chance M3 wasn't even out by the time they had the mostly finalised design for this processor. They definitely couldn't aim for any comparison with M4 at any point of that process.


The hardware looks and sounds great. Too bad it's running Windows 11.


The hardware is a Trojan horse to get the software in your house.


What concerns me the most is the repairability. We make extensive use of Macbook Air M1 and M2 units, and we have a support contract for repairing the devices onsite.

We also have a support contract for repair of Surface Pro units, and when these die they get sent away to Microsoft and a replacement eventually arrives. From what I can tell, they have dreadful repairability.


Surface for Business service and repair...

>Built with integrity by design, many Microsoft Surface devices are now easier to repair1 and maintain, providing commercial customers with greater flexibility while extending the use of devices.

>Commercial customers can obtain service for Surface for Business devices in several ways. You can get service directly from Microsoft or use a third party Authorized Service Provider. Or with skilled technicians, you can repair2 devices yourself following the applicable Surface Service Guide or article.

https://learn.microsoft.com/en-us/surface/surface-service-an...


Who offers on-site repair for your Macs? All my employers have sent us to the Apple store.


CompNow. We have thousands of Macs (we are a school). Many private schools in Sydney have paid for on-site repairs.


MS is trying to improve Surface repairability, no idea how it is going...

https://techcommunity.microsoft.com/t5/surface-it-pro-blog/s...


I wonder how much performance will be used to run all the advanced Microsoft telemetry software (which collects & reports user data for Microsoft and which can't be disabled) and all that ad software they force on users.

I still wonder how it is legal to use Microsoft software in the EU in government surroundings. The German authorities publish tons of papers on how to use Windows&related ( https://www.bsi.bund.de/SiteGlobals/Forms/Suche/EN/Servicesu... ) , instead of simply disallowing to use Microsoft software.

With the data collection for "AI" it only gets worse.


How does the memory bandwidth compare? Based on my usage, that's key in responsiveness between Mac and PC now. Also for LLMs in memory. I compare the current top Surface Pro with the current Macbook Airs.

51-83 GB/s for current Surface Pro 9 with Intel Core i7-1265U (3200-5200 MT/s). [0]

100 GB/s for current Macbook Air 13/15" M2/M3. [1]

0: https://www.intel.com/content/www/us/en/products/sku/226258/...

1: https://www.apple.com/macbook-air/specs/


Qualcomm is a little ahead of the M4 (8400 vs. 7500) but they're behind the M3/M4 Pro/Max/Ultra.


Do they sleep/wake properly though? If not, it's going to be absolutely useless for me.


Can’t believe it’s 2024 and Microsoft has not sorted this out yet. I switched to MacBooks 15 years ago because I liked being able to close the lid at any moment and stuff it into my laptop and go.

Got burned one too many times trying to sleep/hibernate HP business laptops by closing the lid and they stayed on and got hot. Not once has that happened with a MacBook.


It's not that they haven't sorted this out, it's that they intentionally broke it. S3 sleep was a solved problem all the way back in the Windows XP days.

The problem is that S3 sleep means the machine is actually sleeping, and short of timers being set ahead of time or explicit wake events by PCI devices, the machine is unable to think which means it can't do important things like download ads^W suggestions or similar. It's technically possible to do Wake-on-WLAN/WWAN but would require moving a ton of logic to the WAN card (it would need an entire network and app stack in order to maintain high-level TLS/etc connections with the various sync servers autonomously and wake the host when something new comes in) and in practice nobody bothered to do it because it's just not actually solving any problem the users care about - even when useful data is being fetched, it turns out that in practice nobody minds waiting 5 seconds for their Outlook to refresh.

Microsoft's "solution" however was to simply remove S3 sleep, so now the machine is incapable of it and is constantly in S0 (powered on state) and it's up to the OS to ensure nothing resource-hungry is running so that the CPU can actually enter its low-power states. This might be doable on an OS built from scratch (it's after all what all smartphones do successfully), but not with Windows if they want to preserve backwards compatibility with the PC we know and love, so in the end you get machines that have the risk of overheating & draining their whole battery while "sleeping"... but hey at least when it works your ads are up to date, progress I guess.


While it's deeply frustrating that S3 standby is increasingly unsupported on laptops with new firmwares, I find that hibernate works perfectly instead. It's been my preferred option whenever I stop using my PC for a short while and blazing fast modern SSDs have made it more useful than ever.

It's bewildering that Microsoft have made hibernate hidden by default. Yet at the same time, "fast startup" is enabled by default! Something as simple as power states has deliberately been completely broken.


It's Windows Modern Standby, one of the stupidest choice Microsoft ever made. Basically the laptop stays on to connect to the Internet to still receive mail and notifications.


MacBooks also have this or something like it. You can ping the MacBook with find my even if it’s been asleep for ages.


MacOS does that too, it's called Power Nap. You can turn it off, though.


Same thing happened with my Dell Precision laptop. Fixed it by wiping the hard drive and installing Linux.


Notably on recent CPU generations the CPU vendors yanked S3 support completely, so you have to rely on your Linux installation being able to adequately manage power in S0ix. While it won’t be hostile to you like a Windows one, I’m yet to see this work as well as S3 with regard to battery life.


Within the last month, a BIOS update was released for my ThinkPad that finally allows this to work reliably under Linux.


Exactly. I wonder if they actually use their own laptops for work. I had one HP laptop literally melt its own plastic after being left inside my bag for a night. Made it a point to always keep it on a desk overnight.


We're talking about the Surface line here, so yes, in my experience power management is flawless.

It gets everything we'd expect from a vertically integrated product. You still need to deal with windows' paradigms, but also get a real linux as a bonus.


So snapdragon believes they’ve built a faster and more energy efficient chip than Apples M-series.

I thought the general consensus was that Apple was years ahead of the competition. What breakthrough did Snapdragon come up with?


The same as Apple. Qualcomm bought Nuvia, which was started by engineers working at Apple on their CPUs.


Buying TSMC silicon just like Apple and AMD.


So the new pro-level laptop from MS+QCOM (they are bragging of the specs of the one with Snapdragon X Elite chip which starts at $1,500) and ships in a month beats last year's entry-level mac, which starts at $999 and has been shipping for a year?


Much as I love to bash MS, in this instance the M3 MacBook was announced March 4, so it’s not last year’s product.


What does it matter what it starts at? The 16GB RAM, 512GB SSD model of the Macbook Air is $1400 with an M2 and $1500 with an M3 (on the Apple website). The preorder price for the Surface Laptop with the same configuration is $1200 with a Snapdragon X Plus and $1400 with a Snapdragon X Elite (on the Microsoft website).


The one who starts at 999 is MacBooks M2. In this case,

Surface Laptop with X Plus also beats M2 and starts at the same price.


If they ship on time they might get four months before Apple has a faster machine. And the last six weeks to two months of that, people will likely wait and see (if they consider anything from MS at all that is).


Is anybody swapping ecosystems based on one-off specifications? That seems incredibly foolhardy. Platforms are sticky.


First rule of marketing is people buy a story, not a product.

If people are bailing it could be the new competition is that compelling, or it could be that your customers have been looking to jump ship and this product was compelling enough to tip them over.


Apple already has much faster machines - the pro line...


Seriously. I'm already a little annoyed that Apple has moved away from "and you can buy it today", but at least they're way ahead still.


while using more power ;)


Competition is good but somehow I just know this won’t actually come close to competing. Between Microsoft and Qualcomm I don’t trust a single metric quoted until it’s actually in consumers hands.


if they have better trackpad, I will buy one.


Who cares about performance minutia? It ships with windows. That information should be enough for any sensible buyer.


You'd think they'd compare Pro models v. Pro models.


Can you run Windows 10 on these? Not interested in 11


Probably not considering Qualcomm ARM CPUs and GPUs will require new drivers Windows 10 never had.

Even the Intel and AMD machines coming later down the line will require Windows 11 to be utilized to their full potential. Intel Thread Director being an easy one to point out; Alder Lake onwards all run suboptimally on Windows 10 due to lack of support and awareness. Also new drivers for the NPUs, for whatever good they will serve.


Do I have to pay extra for no ads?


Is intel dead at this point?


Partially copying my comment from the Copilot+ thread

They do some sleight of hand when doing the comparisons during the release today.

The Surface Laptop is physically a competitor to the actively cooled MacBook Pro (in fact it’s thicker).

Their performance and battery metrics are against the slimmer and passively cooled MacBook Air.

Their performance comparison is between the M3 and the Snapdragon X Elite when the M3 has throttled (their wording is sustained performance). The Elite is a significantly higher TDP product line than the M3 and its own sibling.

Their battery comparison is between the M3 and the Snapdragon X Plus. This is the lower wattage SoC, paired with a thicker body that likely has a larger capacity battery as a result.

It’s very unlikely that the device has both stronger performance AND longer battery life, because those are handled by different SKUs. Likely you will choose whichever metric matters more to you.

Still a strong showing, but take the “benchmarks” with a heavy grain of salt.


It's clear that elite x is not better than M3 in everything. But considering how far behind ARM on windows was, even having 5-10% slower chip is also a great improvement.

Hope Qualcomm manage to make 15-20 jump with each generation like Apple.


Sure, that’s why I end my comment with saying that it’s a strong showing.

I’m looking forward to more aarch64 support everywhere. I just wish companies were more straightforward about what’s being presented.

Though I believe that perf per watt will end up being closer to an M2 than an M3. If you take the battery size of an M2 MacBook Pro it also claims similar battery life as the Snapdragon X Plus here.


Until I can see some real benchmarks it's all bullshit. I watched their presentation and the lack of specificity is a glaring red flag. I'll happily eat my hat if the claimed performance is replicated across more than some rather targeted benchmarks.


Hashtag doubt. They don’t mention how many hours of ads in your start menu it supports, nor how much telemetry it can send


Faster doesn't tell the whole story. For example, I'll take 80% as fast with 8 hours of battery life over 100% speed at 2 hours.


If you read more than just the headline you'd find that the third sentence of the article talks about battery life.


Yes but you'll note MS has made no specific claims about battery life and performance.


You can't have both. With the current M3 you also won't get 20h on battery while running an hadoop cluster on it.


Performance-per-watt leagues ahead of the competition is the key metric Apple is selling their silicon on. The whole reason these chips are getting so much acclaim is they do have both.



That story is not about speed, it's about Copilot. To be fair, everything is called Copilot now, so maybe fast Surface Pros will be called Copilots too.


The discussion about the speed and surface specs are all in the announcement and that's where the threads were merged to. That's where the discussion is.


People that want gaming choose Windows, people that want productivity choose Mac. I don't see where this new product fits.

You can't game with the paltry "arm companion" GPU and you can't be productive with tons of ads (product placement, "notifications" and tips, as well as the literal "hot singles near you" variety included).

Maybe this is some kind of corporate devices for the old-school enterprises that rock Teams and Office and all that jazz.

It's a shame because if the hardware is as good as they make it sound, it might've been a great productivity device, but they need serious slimming down work in Windows, or an unlocked bootloader for Linux.


It’s a fairly competent and competitive GPU for a low end SoC. Allegedly roughly comparable to an AMD 780M.

https://videocardz.com/newz/qualcomm-snapdragon-x-elite-adre...

That said, Microsoft’s comparisons should tell you who this is for: people who might otherwise buy a MacBook Air. That’s the segment that’s largely not doing creative work or gaming, but needs something for business use, or productivity on the go.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: