Lunar Lake +$100 to add 16GB LPDDR5X-8533MHz vs M3 +$400 to add 16GB slower LPDDR5-6400MHz

[leman]

Ars had a pretty good one a couple of weeks ago:

Review: Intel Lunar Lake CPUs combine good battery life and x86 compatibility

But it’s too bad that Intel had to turn to TSMC to make its chips competitive.

arstechnica.com

It's not fast (slower than M3), but it's a good improvement in efficiency. I'm no fan of Apple's upgrade pricing, but if they get me to the point where I rage quit and buy a PC laptop, it'll be powered by AMD and running Linux.

Unfortunately there is no way to compare their battery test results to the wider set of results. Notebookcheck's battery tests, although not unproblematic, at least try to follow standard methodology.

There is also a large degree of discrepancy between these tests. In the Arstechnica review, the X1 Carbon Gen9 has very high battery runtime, but in notebookcheck reviews its battery is much lower than Gen11 for example. Really makes it different to compare. If we assume that the 155H UX3405M tests are consistent between the two reviews, than the new Lunar Lake model should score around 950 minutes on the WiFi test, which is again pretty much equivalent to the M3 Air. But I don't think that the results can be compared in this way. It is really unfortunate how poor the state of computer reviews is. Results are all over the place and there is no way for the consumer to make a proper data-driven decision.

 

[leman]

So again one can wonder what’s the purpose of such posts other than personal amusement and satisfaction. Who even enjoys playing Cyberpunk at 1080p Low with FSR Ultra Performance on a high-res screen? That’s the lowest and worse quality you can choose to play a AAA game. A real gamer wouldn’t for sure buy such a laptop for gaming or even less recommend it. A Mac gamer wouldn’t certainly buy such a PC laptop for gaming either.

Why would I pay $1300 for such an inferior Lunar Lake laptop when I can buy a gaming console for 400-500 dollars? I can also get 52 years of renewal of Crossover. If that is not good enough I could get 6 and a half year of GFN 4K subscription. After that you have to replace that PC anyway so I could keep paying for GFN instead. I could also easily pay $1300 for a Mac Mini with M4 Pro with 20c GPU with double the performance of MBA or pay only 300-500 dollars extra for a Mac Studio with M4 Max and get at least 60 fps at 1440p High with frame generation in Cyberpunk with Crossover.

Any solution is better than that Lunar Lake laptop if you have or want to buy a Mac. You get what you pay for.

Amen to that. It is just such an odd thing to claim performance superiority using software that is not available on the other platform. If I want to play Cyberpunk on my laptop, obviously I wouldn't get a Mac. I wouldn't get a Lunar Lake platform either. But that's so michi-boy, mixing some technical knowledge with half-truths and drawing ridiculous comparisons.

P.S. Great post btw, really shows how far ahead Apple is. It is insane that a passively cooled laptop can have so much performance advantage on such a wide range of applications. The M4 models are going to be ridiculous in comparison. I wouldn't be surprised if the M4 Max laptops will trade blows with fastest Zen5 and Arrow Lake desktops.

 

[MayaUser]

I thought people by now knows who mi7chy is and what hes trying to do and fails every time. Not to mention that this topic should have been created in mac alternatives section

 

[komuh]

Not to be this guy but level of cope is insane in this topic.
Apple is overpricing ram and ssd as crazy, top lunar lake cost less than m3 air with performance around m3 pro chip and intel have 1TB of ssd plus 32 GB of ram.

It is not some sort of conspiracy while some folks prefer windows some Linux other Macs, only we Mac enjoyers are getting farmed by big corporation and defending it 🫣

 

[mi7chy]

There are very few native MacOS games but even Shadow of the Tomb Raider runs about half of the fps on M3 compared to Lunar Lake granted you didn't fall victim to 8GB = 16GB scam where it's less than half of the fps while paying over 50% more and only getting 24GB vs 32GB RAM, no 120Hz display, no WIFI 7, no user upgradable/replaceable storage, etc. but decent for synthetic benchmarks.

Lunar Lake 258V 1080p lowest on battery

M3 1080 lowest (8GB left, 16GB right)

Here Lunar Lake 16GB to 32GB RAM upgrade is only +$60 difference ($1299.99 vs $1359.99) and 1TB SSD is included vs overpaying for M3 +$800 8GB->24GB and 256GB->1TB which is the point of this discussion.

 

[Ethosik]

I think people get the wrong idea because some scenarios show Proton on Linux can run a game better than it does natively on Windows. So they crap on Apple for doing a horrible job.

There is no universe where a translated/emulated app runs better than native unless you deal with operating system specific bugs.

 

[leman]

Not to be this guy but level of cope is insane in this topic.
Apple is overpricing ram and ssd as crazy, top lunar lake cost less than m3 air with performance around m3 pro chip and intel have 1TB of ssd plus 32 GB of ram.

Apple's upgrade options were always ridiculously expensive, it is part of their business model. This is not cope, just being realistic. They are unlikely to change this business model any time soon, so it is not like we can do much about it. Historically, consumers are happy to pay Apple's high prices because they perceive Apple products to have higher value.

Which is reflected by real-world experience. Homy quotes a bunch of benchmarks in #25. Look at the photo/video benchmarks. The Lunar Lake laptops they use have 32GB, unfortunately they don't mention the MacBook Air M3 config (they only mention it is an 8-core, so it might be the 8GB version actually). Anyway, the Air is substantially faster in all creator workloads, often by a large margin.

So this becomes a conversation about specs rather than about value. Why would I buy a computer with 32GB RAM if another 16GB model is faster for my workloads? I feel that a more differentiated approach is needed, one that takes into account the concrete needs and workflows of the user. Frankly, I don't see why anyone buying this kind of platform would need 32GB RAM, the hardware capacity to effectively process large datasets is just not there. Intel decided to offer a large amount of RAM for cheap, good for them and for their customers. I fail to see practical value in this, as it does not make the laptop faster for what people buy it for.

It is not some sort of conspiracy while some folks prefer windows some Linux other Macs, only we Mac enjoyers are getting farmed by big corporation and defending it 🫣

I don't really see where you are getting this. The PC world is plagued by cost-cutting measures, malware installed by default, fraudulent advertising, and all kinds of issues. For many years PC customers were hostages of increasing power inflation and lack of transparency regarding power and performance levels, and PC laptops are still sabotaged by poor performance on battery. It's good that Intel is moving in the right direction with Lunar Lake, which takes a much smaller performance hit on battery. Still, consistent performance whether you are plugged in or not have long been the main value proposition with the Mac laptops, and it is something other vendors are still unable to offer.

 

[leman]

There are very few native MacOS games but even Shadow of the Tomb Raider runs about half of the fps on M3 compared to Lunar Lake

Oh please. That Shadow of the Tomb Raider is a poor port is well known. There are plenty of modern demanding titles available natively on macOS, why don't you use any of those?

If you already insist on cherry-picking look at something like Frostpunk 2. The subreddit is full of people complaining about endgame lags on their 4070 gaming desktops, while my M3 Max is happy to deliver smooth 4K gameplay on battery. This is all silly anyway.

 

[Ethosik]

So this becomes a conversation about specs rather than about value. Why would I buy a computer with 32GB RAM if another 16GB model is faster for my workloads?

Because people state you need more RAM not a better overall system. Then people mock Apple for their statements on RAM where technically it’s not true but user experience is true. My clients can’t use an 8GB Windows system with Photoshop and premiere for 4K editing. But they can use their M1 iMac with 8GB.

 

[mi7chy]

Source for Frostpunk 2 MacOS claim? Developer's Steam forum shows differently.

https://steamcommunity.com/app/1601580/discussions/0/4853281047676597710/

https://steamcommunity.com/app/1601...5755024051043051474&include_deleted=1&q=macos

Lunar Lake competes with M3 so M3 Max and 4070 are kind of irrelevant. How does Frostpunk 2 run on M3?

"Sorry if this sounds dumb but I purchased the game through the Apple App Store. I’m not a PC gamer at all but wanted to play FP2 so I got it on my MacBook Air M3 w/ 8GB RAM. The game is running really choppy and looks terrible at high setting. Is it because the spec can’t handle the game?"

Can compare on Lunar Lake when the game goes on sale since it doesn't look that interesting.

Meanwhile, real world Blender comparison on Party Tug scene so not synthetic benchmark.

Lunar Lake 39.51 secs (lower is better) on battery

M3 8GB and 16GB

 

[__main__]

Hmmm. From the official Blender site.

Blender - Open Data

Blender Open Data is a platform to collect, display and query the results of hardware and software performance tests - provided by the public.

opendata.blender.org

 

[leman]

Meanwhile, real world Blender comparison on Party Tug scene so not synthetic benchmark.

I find these results odd because all other available tests show M3 Air being faster in Blender. That is not to say that you are wrong, I just think this warrants additional investigation.

I briefly tried the Party Tug scene on my M3 Max (30 cores) and it took 7 seconds in low power mode. So I find it very odd that the M3 Air would take almost 26x longer. In the official Blender benchmark the 10-core M3 is 3.7x slower than the 30x core M3 Max, which would put the expected time at around 25-30 seconds. Which would be around ~30% faster than the Lunar Lake platform and consistent with all other available Blender benchmarks.

Edit: I double-checked it and it turns out I was using the default EEVEE renderer. Using cycles it takes 40 seconds, so consistent with MaxTech results. @mi7chy which rendering engine are you using for your Lunar Lake render? The shadows on your image are blocky and the lighting is off, this does not look ray-traced. Your image looks very similar to my EEVEE result, while Cycles looks very different. I am attaching two images for comparison.

EEVEE

Cycles

 

[mi7chy]

Except for switching from CPU->GPU rendering always default Blender scene settings unless noted otherwise. For Party Tug default is EEVEE.

Render time is preferred over score. Blender benchmark database lists M3 but doesn't distinguish passive cooling M3 Macbook Air from active cooling M3 Macbook Pro. With passive cooling M3 Macbook Air, performance is variable due to throttling so ok for short synthetic benchmarks, renders, workloads, etc. before it throttles from sustained workload. Furthermore, even with 16GB RAM it can only handle simpler scenes but runs on out memory on Barbershop and more complex scenes so >16GB is preferred.

https://www.theverge.com/24114684/apple-macbook-air-m3-13-15-inch-2024-review

Another win for lower cost 32GB Lunar Lake 258V iGPU rendering Barbershop scene without issue. Someone with 24GB M3 Macbook Air do the same render for comparison and make sure to use iGPU rendering.

 

[__main__]

Except for switching from CPU->GPU rendering always default Blender scene settings unless noted otherwise. For Party Tug default is EEVEE.

Render time is preferred over score. Blender benchmark database lists M3 but doesn't distinguish passive cooling M3 Macbook Air from active cooling M3 Macbook Pro. With passive cooling M3 Macbook Air, performance is variable due to throttling so ok for short synthetic benchmarks, renders, workloads, etc. before it throttles from sustained workload. Furthermore, even with 16GB RAM it can only handle simpler scenes but runs on out memory on Barbershop and more complex scenes so >16GB is preferred.

https://www.theverge.com/24114684/apple-macbook-air-m3-13-15-inch-2024-review
View attachment 2435311

Another win for lower cost 32GB Lunar Lake 258V iGPU rendering Barbershop scene without issue. Someone with 24GB M3 Macbook Air do the same render for comparison and make sure to use iGPU rendering.

View attachment 2435318

All credible evidence shows the M3 to be significantly faster than Lunar lake.

 

[leman]

Except for switching from CPU->GPU rendering always default Blender scene settings unless noted otherwise. For Party Tug default is EEVEE.

Ok. You own a Lunar Lake laptop, right? Can you post your results for Party Tug with cycles?

 

[mi7chy]

Except for switching between CPU or GPU rendering I have a policy of not changing scene defaults to discourage from willy nilly changing settings and forgetting to mention so it becomes non-repeatable, misleading and a waste of time for anyone Google searching for comparisons.

 

[Homy]

Oh please. That Shadow of the Tomb Raider is a poor port is well known. There are plenty of modern demanding titles available natively on macOS, why don't you use any of those?

There are 5,762 native Apple Silicon games only on Steam, besides the ones like RE series on Mac App Store but despite all this effort somehow only ”very few” can be found and always the ones with worst performance. Here is another x86 game running through Rosetta but with different result. Intel showed in their presentation that Metro Exodus ran with 45 fps at 1080p Medium on their fastest Lunar Lake Ultra 9 288V with Arc 140V. Here it’s running on the old M2 10c GPU at 1080p Medium with 40-60 fps. So it’ll run even faster on M3 10c GPU.

 

[eltoslightfoot]

I have a post on here I will have to dig up. It shows that by getting a refurbished Acer Predator Helios Neo 16” with an i7-13700HX with RTX 4060 and 16GB RAM and 1TB SSD for $999 with 2 year warranty (off Acer’s official EBay store) I saved literally thousands over what an equivalent MacBook Pro 16 would have been (this was a year ago). It was like $2800 or something.

Then I doubled the RAM to 32GB and added another 1TB SSD for like $120. So there I saved another thousand or so.

And I have owned many Macs over the years. But this is getting absolutely ridiculous on RAM and SSD prices. Just ridiculous. And I can run all my favorite games. I refuse to buy any more Macs—especially now that Powershell is on Windows. I can still do all my Linux-y things.

 

[eltoslightfoot]

Found it…

https://forums.macrumors.com/threads/bye-bye-for-now-apple-it-has-been-fun.2421598/page-4?post=33002004#post-33002004

 

[leman]

Except for switching between CPU or GPU rendering I have a policy of not changing scene defaults to discourage from willy nilly changing settings and forgetting to mention so it becomes non-repeatable, misleading and a waste of time for anyone Google searching for comparisons.

Ok. In that case you will surely want to update or remove your post #35 where you are comparing rasterized EEVEE render time on Lunar Lake with path-trace Cycles rendering time on M3, right?

 

[leman]

I have a post on here I will have to dig up. It shows that by getting a refurbished Acer Predator Helios Neo 16” with an i7-13700HX with RTX 4060 and 16GB RAM and 1TB SSD for $999 with 2 year warranty (off Acer’s official EBay store) I saved literally thousands over what an equivalent MacBook Pro 16 would have been (this was a year ago). It was like $2800 or something.

Then I doubled the RAM to 32GB and added another 1TB SSD for like $120. So there I saved another thousand or so.

And I have owned many Macs over the years. But this is getting absolutely ridiculous on RAM and SSD prices. Just ridiculous. And I can run all my favorite games. I refuse to buy any more Macs—especially now that Powershell is on Windows. I can still do all my Linux-y things.

And that's why I think it is great one has all the options on the market. Availability of choice is good for us customers. The definition of what constitutes value is different for everyone. I think if one looks at it from this perspective, rather than taking elements of specs and their prices out of context, the entire story becomes much simpler.

 

[MayaUser]

I have a post on here I will have to dig up. It shows that by getting a refurbished Acer Predator Helios Neo 16” with an i7-13700HX with RTX 4060 and 16GB RAM and 1TB SSD for $999 with 2 year warranty (off Acer’s official EBay store) I saved literally thousands over what an equivalent MacBook Pro 16 would have been (this was a year ago). It was like $2800 or something.

Then I doubled the RAM to 32GB and added another 1TB SSD for like $120. So there I saved another thousand or so.

And I have owned many Macs over the years. But this is getting absolutely ridiculous on RAM and SSD prices. Just ridiculous. And I can run all my favorite games. I refuse to buy any more Macs—especially now that Powershell is on Windows. I can still do all my Linux-y things.

That combo cpu+gpu is not good for people like me who needs both cpu and gpu power over short and longer period of time...that $2000 difference for us is melted in 1 month with 16" M3 Max and from there.....
So again, people are different with difference needs. For gaming , for sure macs has far more alternative/competition options on the other side, better options, but gaming is for spending more money and time. For creators macs are a viable solution

 

[eltoslightfoot]

That combo cpu+gpu is not good for people like me who needs both cpu and gpu power over short and longer period of time...that $2000 difference for us is melted in 1 month with 16" M3 Max and from there.....
So again, people are different with difference needs. For gaming , for sure macs has far more alternative/competition options on the other side, better options, but gaming is for spending more money and time. For creators macs are a viable solution

Oh I agree. There are certain combos with local LLMs and the like. But it is completely unfair to compare an M3 Max to an i7-13700HX and RTX 4060 combo.

You would have to do so with an appropriate high end AMD processor and RTX 4090. Even then. The Windows gaming laptop would win in some arenas and the M3 Max would win in others (and cost thousands more again--if that matters. For these professional uses, it may honestly not).

For me, I was happy to save thousands and go with what works. I mostly do Python (now dabbling in AI), writing (Scrivener and Obsidian), etc., and of course gaming. So I had to have the 4060 anyway...

 

[mi7chy]

Lunar Lake on battery looks to be around double the fps of M3 MBA on Metro Exodus at 1080p low.

M3 MBA

 

[Homy]

I thought everybody knew that at 1080p Low the game is CPU-bound and doesn’t task the GPU much. The E-cores and P-cores in M3 have a frequency of 2.75 - 4.06 GHz. The cores in Ultra 7 258V run at 3.7 - 4.8 GHz. MBA also throttles due to being fanless to keep the temperature low. A better comparison is as I showed with M3 in MBP with fan because Lenovo Yoga has also fan. Intel’s numbers were from 1080p Medium as I already said so where it matters, at higher quality when the GPU kicks in M3 GPU performs better. For that reason the difference between Low and Medium is little on MBA (here about 7.5%) while the difference on PC is up to 43% according to HW Unboxed even on mid-range and high-end desktop GPUs. Also Metro Exodus despite being a better port and performing better than Shadow of the Tomb Raider is still a x86 app compared with a native Windows version.

So again who wants to play a AAA game with lowest quality on a high-res monitor? The PC can run the game with higher frame rate on Low, the Mac can run it with higher frame rate on Medium. I take the Mac especially in such a game where you don’t need 100 fps, but if frame rates matter you may as well run it at 360p. You may even break the speed of light with that. Some also say if you run yourself while playing the game it may run faster but I haven’t tried it myself.