A few things I noticed, as I'm seeing the variety of SKUs becoming more complex.
- Note that memory bandwidth is down. M2 Pro had 200GB/s, M3 Pro only has 150GB/s. M3 Max only has 400GB/s on the higher binned part.
- Just like the low-spec M3 14" has one fewer Thunderbolt port, it also doesn't officially support Thunderbolt 4 (like M1/M2 before it)
- The M3 Pro loses the option for an 8TB SSD. Likely because it was a low volume part for that spec.
- The M3 Pro actually has more E-cores than the Max (6 vs 4). Interesting to see them take this away on a higher-specced part; seems like Intel wouldn't do this
One thing that left me a bit confused was the comparison against Intel Macs. Although I am still using an Intel 16-inch MacBook, I really wanted to see how the M3 fared against the M2, not Intel and M1. I think it's no surprise the M3 exceeds the Intel Core i7-9750H in basically all of Apple's own benchmarking. My real question, which will probably be answered next week, is how it compares to the generation right before it.
My work laptop is a 14-inch MacBook Pro and I've been impressed with the battery life considering all of the containers that run (on Kubernetes!) as part of my dev workflow. I guess Apple deciding to compare Intel and M1 was to try to convince existing MacBook users to upgrade to the latest generation CPUs.
I feel very disappointed that the laptops lack decent i/o. They spend all this time talking about how advanced the GPU is, but it's a $1,600 laptop that can only power one external display or a 2,000 laptop that can only power two displays?
Plus they are nerfing the desktops by only offering the all-in-ones with m2, so you can walk into a store and want to buy any other desktop now you are paying a huge premium for outdated tech :(
I was really hoping for FaceID and better display support, but I don't feel there was any compelling reason to upgrade anything. If you preferred Windows before, you probably still prefer Windows. If you have an M processor there is no compelling reason to upgrade. If you were thinking about getting a Studio or a mini, you should probably just wait until they update the processor.
An interesting thing about the M2 Pro and M3 Pro is that they shifted away from being mostly performance cores. The M1 Pro was either 6+2 P+E or 8+2 P+E. The M2 Pro was either 6+4 or 8+4. The new M3 Pro is either 5+6 or 6+6.
Apple has been shifting away from performance cores with each generation. M1 Pro: 75-80% P-cores; M2 Pro: 60-67% P-cores; M3 Pro: 45-50% P-cores.
This shows up when you look at Geekbench results. A 10-core M2 Pro (6+4) gets 12,100 while a 10-core M1 Pro (8+2) gets 12,202. The 12-core M2 Pro (8+4) gets 14,221. That's a 16.5% increase from having 20% more cores. In some ways, this feels like an odd result. Adding two additional M2 P-cores gets Apple a comparatively small result over the 10-core M2 Pro (less than the average core performance). However, adding two efficiency cores over the M1 Pro gives them the same 16.5% boost over the 8+2 M1 Pro.
If I had to guess, maybe it's thermal throttling when running the benchmark. If the additional P-cores can't truly be P-cores under 100% load, then their impact shows up less. Likewise, if the E-cores can match P-core performance under heavy thermal load, then it could show up as being roughly equivalent in the benchmark.
I wonder if real-world scenarios end up differing from benchmarks in a meaningful way around this. For example, core-pinning can be useful for warm caches and in a real-world scenario you might have a process the OS tries to pin to P-core-1 that has spikes in utilization while another process is pinned to P-core-2 with similar spikes. So both get the performance of a P-core and warm caches while the thermal load isn't that high so the P-cores are still at their peak performance (unlike a benchmark that's trying to use all cores as much as possible at the same time).
Maybe this is a business decision more than something based around how the chip performs. The big selling point of the M1/M2 Max was graphics (maybe the extra RAM as well). You could get the same CPU in the Pro or Max. Now the M3 Pro is a 5+6 or 6+6 CPU while the M3 Max is 10+4. 67-100% more performance cores becomes a selling point for the M3 Max even for those who might not care about graphics as much.
What's with the 36 GB option? The other memory configs (16 GB, 64 GB) are still clean powers of two. Size suggests that they're using ECC-capable memory but using the extra width intended to support ECC for data... but why would this only apply to a single size? Part availability?
EDIT: Digging this a bit, it's not (one or more) 72-bit wide busses with 2^32 words as I'd expect as a gray-beard, it's (probably) six 32-bit wide busses with 6 GiB per bus; and this use of 1.5 * 2^N deep memories has become relatively common with the use of IC stacking, with 12 power-of-two sized ICs stacked in a single package (instead of the more "comfortable" 8 high or 16 high stacks of the same ICs giving powers of two).
It also appears they've dropped the memory bandwidth from 200GB/s on the M1/M2 Pro to 150MB/s on the M3 Pro, and you have to upgrade to the tippy top M3 Max chip to get the full 400MB/s bandwidth experienced on the M1/M2 Max chips.
I prefer that tbh. If I'm frequently saturating 16GB of memory in the previous gen and it is time for an upgrade, I probably don't need literally double the memory, but 24GB would be nice.
It looks like these models won't be as useful for LLM inference which are heavily memory bandwidth constrained. The Macbook Pro page shows M3 at 100GB/s,150GB/s, and 300GB/s vs M2 at 200GB/s and 400GB/s. 400GB/s is available for M3 if you opt for the high gpu config, but interesting to see it go down across all of these models.
I don't understand why they don't update the entire lineup whenever a new chip comes out, especially if really only swapping the chip in the macbook pro and imac here.
On a separate note does anyone have any insight on how unified memory compares to vram in the context of machine learning performance? Considering an H100 with 80GB costs like $30k a maxed out macbook pro with 128gb unified memory for $5k is interesting. Is it remotely comparable or compelling for large models considering realistically most prosumers are capped at the 24gb nvidia cards for anything resembling a reasonable budget.
1. Engineering bandwidth (it's work to upgrade a lineup! They probably need at least a year)
2. Manufacturing bandwidth - they probably spent the first half of the year manufacturing the 3nm iphone processor while researching the mbp processor. I expect it's difficult to ramp manufacturing on many chips simultaneously
3. Sure, consider the demand side. If my parents are going to buy a new laptop and a new phone this year, I think they would be more likely to do that 6 months apart. Similarly, I expect keeping a cadence of announce products a,b,c in this quarter and d,e,f in that quarter helps keep apple in the news (for good/exciting things)
Wrt machine learning: I can't wait for the results to come out for this once we can get our hands on it
Speculating, but I'd wager it's to clear out existing inventory by phasing out the old chips on other SKUs over time. E.g., you start with the expensive Macbook Pro models when you have fewer M3 chips in production, since those will sell slowly, and then you clear out your remaining M2 inventory in the Air lines while waiting for a "refresh" when M3 production is more ramped up.
I have no idea what Apple's sales are for the macbooks or if Airs sell more than Pros, etc., but they're definitely making profit this way to be this consistent in their approach.
Besides wanting to clear out existing M1/M2 capacity, I don't there's enough M3 stock right now.
I forget how many fabs TSMC has that have 3-nanometer capability, but it's gotta be low. Apple's supposedly got _all_ of TSMC's capacity[0], but I doubt it's enough.
They just updated their macbook air and mac studios and mac pros like 6 months ago? If they updated right now, the previous owners would feel utterly screwed over.
The M2 Ultra tops out at 800GB/s unified memory bandwidth. A 4090, on the other hand, has 1,008 GB/s. On the PC side, dual-channel DDR4-6400 offers a bandwidth of 102 GB/s.
Depends on if you're talking about training performance or inference performance. It's probably a decent deal for inference on such large models, but I doubt it's anywhere near competitive for training. In the 'consumer' space there's also the A6000 with 48GB for ~$4k.
It makes a world of sense if you think of it from a logistics point of view, especially considering Apple sells insane volumes of everything they sell.
It costs less to manufacture the current generation ship and the last generation chip concurrently, it allows them to push the newest generation chip out the door faster without running out of supply or having to open up excessive production lines.
>H100 vs apple silicon
while it's only one spec, H100 has over 2TB/s of memory bandwidth and Apple silicon caps out at 400GB/s. H100 in general kicks the pants out of Apple silicon performance and has far more support. These parts aren't really in the same league but Apple Silicon has become a popular hobbyist choice for LLM inferencing.
Also see: The Jetson AGX Orin 64gb, Nvidia RTX A6000 48gb, AMD Radeon Pro W7900 48gb, two nvlinked 24gb 3090's.
Regarding updating the lineup, I would guess that it is completely intentional to keep the carrot just out of reach across the lineup. Each new chip release rotates the product categories in just the right way to keep the consumers consuming.
Right? I’ve just setup my office and for now I’m using my MBP hooked up to a couple of Studio Displays - which I adore for software development. I was planning on picking up a Mac Mini or Mac Studio so that I could still have the MBP spare.
Maybe not. It’s a bit of a pain having to unplug and replug everything in, not to mention having windows and such shuffle around each time.
My gut tells me the Macbook Air sales have been cannibalizing Macbook pro sales.
For the M1 and M2 generations, the airs released in the fall and the pros a few months later in the spring. The airs have gotten so good that many people that would have gotten the more expensive (and more lucrative) pros got airs instead.
> Additionally, support for up to 128GB of memory unlocks workflows previously not possible on a laptop, such as AI developers working with even larger transformer models with billions of parameters.
The presentation explicitly highlighted this very randomly, without showing how AI development actually works.
I know Apple has put more effort into Apple Silicon support for PyTorch but is it there yet?
It's quite impressive that the event was filmed on iPhone 15 Pro[1], although it involved professional lighting and various equipment, which is not typical for the average user.
With the ability to capture in Log, it's possible that the next iPhone release might be filmed using the very phone that's being unveiled.
Really love the pace of Apple innovations. Nowadays, I am biased towards battery and weight (MacBook Air) if it is for portability and the other line of notebooks are converted into desktops via TB exactly because of weight and battery life. I feel the difference in battery use from one to the other. The Air seems like it has "infinite battery" evn when the others use Mx processors.
Also waiting for a great Linux port battery/weight wise.
I was seriously considering an m3 pro for the space black color until I confirmed it wasn't passively cooled. An extra monitor would be nice at work but the air is just too great as a laptop.
I think the M series laptops were the first full manifestation of the laptop idea.
Instant sleep/resume, great input devices and monitor, cpu power at will without overheating or loud fan, super fast shock resistant storage, 20+ hours of battery life, no battery drain when off.
It’s kind of strange their benchmark for “Machine Learning Programmers” is “Simulation of Dynamical Systems in MATLAB.” Seems like they could have capitalized better by using a generative ai inference benchmark.
These are still going to be used in laptops with soldered flash chips for the SSD. Apple's laptops aren't meant to be used by people who expect to repair their laptop when the storage fails. Many people don't have access to replacement chips and services to have the flash chips found on the board replaced. It's as if their hardware is meant to be e-waste.
The RAM is still not ECC. They focus so much on the shallow marketing without shipping anything which makes a difference.
My Windows desktop has generated significantly more e-waste per year than the Macs I have owned. Full stop. I've upgraded SSDs from 500gb to 1TB to now 2TB, upgraded graphics cards, and replaced a massive broken aluminum and copper heatsink. All in the last two years.
I've had the same MacBook, in a similar timeframe. It gets just as much use. Its still rock-solid.
Your concern is a hypothetical one. The inability to upgrade has, in a VERY REAL sense, meant for me: I overbuy specs upfront, so I don't have to upgrade (and thus, generate e-waste). It also experientially means that the machines are more reliable. The least wasteful machine is one that doesn't have to be upgraded; not one that can be and needs to be. The 500gb of storage I have in my MacBook sometimes feels limiting; but the cost of upgrading (a whole new machine) has stopped me from actually going out and doing it; and thus less e-waste is generated.
Eventually I will brainstorm what to do with this machine once its outlived its useful life as a laptop. I'm thinking, server rack. We've got Asahi, its got thunderbolt so wiring up super fast storage drives is a cinch.
> It's as if their hardware is meant to be e-waste.
But the chassis is made from 100% recycled aluminum!~
On a serious note, Apple has a simple and working strategy, to which they're 100% committed: repair only in authorized service points, otherwise recycle. Making devices user-repairable adds costs and compromises on specs; you can't put a number on the disks being replaceable as easily as you can put a number on size, IO bandwidth, or number of write cycles, and numbers is what looks good on benchmarks.
You also have to consider Apple's scale: they have literally billions of deployed devices (everything from AirPods to Mac Pros) that they need to support, so what could work for a different vendor won't necessarily work for them.
(I don't necessarily agree with that strategy from end-user POV, merely pointing out the context.)
The point is, they don't want people to fix or upgrade their hardware, and they will fight to defend this strategy. Give me one reason why they should to it, from their perspective. If they have them glued, the equipment has shorter life and customers need to purchase a new one quicker. And since they can't replace components themselves, they have to pay exorbitant prices for 16 GB RAM or 1 TB storage that is cheaper than ever.
These are nice machines, though, and I buy one for building iOS apps every couple of years - but usually 1-2 generations later so that the pricing is more reasonable.
What's the deal with "dynamic caching"? I wish they would have talked about it more. It sounds like they're reducing memory allocations by only allocating what is actually needed.
You got it. Instead of letting software developers allocate a static cache with overhead that eats a little into the unified memory, the Mac gets to decide how to dynamically allocate and release cache for graphics. They said it would be "transparent to developers," but not really sure what that means exactly.
From what I understand this is about assigning GPU resources (such as register files and other on-core memory) to shaders. Imagine you have a complex shader that has two paths, a frequently taken fast one that needs X bytes of on-chip memory to function and a rarely taken slow path that needs Y bytes (Y > X). Usually this stuff is statically partitioned, so you have to provision Y bytes for the slow path to run this shader. With dynamic partitioning, the shader will only allocate Y bytes if the slow path is hit, which frees the resources to load up more concurrent shader programs and improve the shader occupancy.
This stuff is only really relevant if you are dealing with complex uber-shaders and recursive function invocations, both of which are fairly common in raytracing pipelines.
They kept mentioning performance relative to intel Macs which makes me think there is a large cohort of people sticking to their x86 rigs due to compatibility. Being able to run an x86 linux or windows VM is still a requirement for me.
> With the M3 family of chips, hardware-accelerated ray tracing comes to the Mac for the first time. Ray tracing models the properties of light as it interacts with a scene, allowing apps to create extremely realistic and physically accurate images. This, along with the new graphics architecture, allows pro apps to deliver up to 2.5x the speed of the M1 family of chips.
Does this mean we can finally play Cyberpunk 2077 on a laptop? Is it going to be anywhere near as powerful as desktop Nvidia cards?
The performance is not that great, but you can already play Cyberpunk 2077
on a M2 Pro laptop using Game Porting Toolkit https://www.youtube.com/watch?v=sPJpkRmsceU
Keep in mind, it has to do x64 -> ARM64 transition and translate Direct X into Metal. I would expect a native version to work better.
[+] [-] dang|2 years ago|reply
Apple unveils the new MacBook Pro featuring the M3 family of chips - https://news.ycombinator.com/item?id=38078065
Apple supercharges 24‑inch iMac with new M3 chip - https://news.ycombinator.com/item?id=38078068
[+] [-] kivlad|2 years ago|reply
- Note that memory bandwidth is down. M2 Pro had 200GB/s, M3 Pro only has 150GB/s. M3 Max only has 400GB/s on the higher binned part.
- Just like the low-spec M3 14" has one fewer Thunderbolt port, it also doesn't officially support Thunderbolt 4 (like M1/M2 before it)
- The M3 Pro loses the option for an 8TB SSD. Likely because it was a low volume part for that spec.
- The M3 Pro actually has more E-cores than the Max (6 vs 4). Interesting to see them take this away on a higher-specced part; seems like Intel wouldn't do this
[+] [-] koito17|2 years ago|reply
My work laptop is a 14-inch MacBook Pro and I've been impressed with the battery life considering all of the containers that run (on Kubernetes!) as part of my dev workflow. I guess Apple deciding to compare Intel and M1 was to try to convince existing MacBook users to upgrade to the latest generation CPUs.
[+] [-] daft_pink|2 years ago|reply
Plus they are nerfing the desktops by only offering the all-in-ones with m2, so you can walk into a store and want to buy any other desktop now you are paying a huge premium for outdated tech :(
I was really hoping for FaceID and better display support, but I don't feel there was any compelling reason to upgrade anything. If you preferred Windows before, you probably still prefer Windows. If you have an M processor there is no compelling reason to upgrade. If you were thinking about getting a Studio or a mini, you should probably just wait until they update the processor.
[+] [-] mdasen|2 years ago|reply
Apple has been shifting away from performance cores with each generation. M1 Pro: 75-80% P-cores; M2 Pro: 60-67% P-cores; M3 Pro: 45-50% P-cores.
This shows up when you look at Geekbench results. A 10-core M2 Pro (6+4) gets 12,100 while a 10-core M1 Pro (8+2) gets 12,202. The 12-core M2 Pro (8+4) gets 14,221. That's a 16.5% increase from having 20% more cores. In some ways, this feels like an odd result. Adding two additional M2 P-cores gets Apple a comparatively small result over the 10-core M2 Pro (less than the average core performance). However, adding two efficiency cores over the M1 Pro gives them the same 16.5% boost over the 8+2 M1 Pro.
If I had to guess, maybe it's thermal throttling when running the benchmark. If the additional P-cores can't truly be P-cores under 100% load, then their impact shows up less. Likewise, if the E-cores can match P-core performance under heavy thermal load, then it could show up as being roughly equivalent in the benchmark.
I wonder if real-world scenarios end up differing from benchmarks in a meaningful way around this. For example, core-pinning can be useful for warm caches and in a real-world scenario you might have a process the OS tries to pin to P-core-1 that has spikes in utilization while another process is pinned to P-core-2 with similar spikes. So both get the performance of a P-core and warm caches while the thermal load isn't that high so the P-cores are still at their peak performance (unlike a benchmark that's trying to use all cores as much as possible at the same time).
Maybe this is a business decision more than something based around how the chip performs. The big selling point of the M1/M2 Max was graphics (maybe the extra RAM as well). You could get the same CPU in the Pro or Max. Now the M3 Pro is a 5+6 or 6+6 CPU while the M3 Max is 10+4. 67-100% more performance cores becomes a selling point for the M3 Max even for those who might not care about graphics as much.
[+] [-] addaon|2 years ago|reply
EDIT: Digging this a bit, it's not (one or more) 72-bit wide busses with 2^32 words as I'd expect as a gray-beard, it's (probably) six 32-bit wide busses with 6 GiB per bus; and this use of 1.5 * 2^N deep memories has become relatively common with the use of IC stacking, with 12 power-of-two sized ICs stacked in a single package (instead of the more "comfortable" 8 high or 16 high stacks of the same ICs giving powers of two).
[+] [-] stetrain|2 years ago|reply
M2 and M3 go up to 24GB
M3 Pro is available in 18GB and 36GB
M3 Max is available in 36GB, 48GB, 64GB, 96GB, and 128GB
[+] [-] kccoder|2 years ago|reply
[+] [-] ksec|2 years ago|reply
[+] [-] fastball|2 years ago|reply
[+] [-] easton|2 years ago|reply
[+] [-] LeoPanthera|2 years ago|reply
Clive Sinclair did that trick back in the 80s.
[+] [-] sharms|2 years ago|reply
[+] [-] jdprgm|2 years ago|reply
On a separate note does anyone have any insight on how unified memory compares to vram in the context of machine learning performance? Considering an H100 with 80GB costs like $30k a maxed out macbook pro with 128gb unified memory for $5k is interesting. Is it remotely comparable or compelling for large models considering realistically most prosumers are capped at the 24gb nvidia cards for anything resembling a reasonable budget.
[+] [-] wrsh07|2 years ago|reply
1. Engineering bandwidth (it's work to upgrade a lineup! They probably need at least a year)
2. Manufacturing bandwidth - they probably spent the first half of the year manufacturing the 3nm iphone processor while researching the mbp processor. I expect it's difficult to ramp manufacturing on many chips simultaneously
3. Sure, consider the demand side. If my parents are going to buy a new laptop and a new phone this year, I think they would be more likely to do that 6 months apart. Similarly, I expect keeping a cadence of announce products a,b,c in this quarter and d,e,f in that quarter helps keep apple in the news (for good/exciting things)
Wrt machine learning: I can't wait for the results to come out for this once we can get our hands on it
[+] [-] Willish42|2 years ago|reply
I have no idea what Apple's sales are for the macbooks or if Airs sell more than Pros, etc., but they're definitely making profit this way to be this consistent in their approach.
[+] [-] CaliforniaKarl|2 years ago|reply
I forget how many fabs TSMC has that have 3-nanometer capability, but it's gotta be low. Apple's supposedly got _all_ of TSMC's capacity[0], but I doubt it's enough.
[0]: Apple is saving “billions” on chips thanks to unique deal with TSMC | https://news.ycombinator.com/item?id=37040722
[+] [-] osti|2 years ago|reply
[+] [-] nateb2022|2 years ago|reply
[+] [-] dotnet00|2 years ago|reply
[+] [-] smith7018|2 years ago|reply
[+] [-] faeriechangling|2 years ago|reply
It costs less to manufacture the current generation ship and the last generation chip concurrently, it allows them to push the newest generation chip out the door faster without running out of supply or having to open up excessive production lines.
>H100 vs apple silicon
while it's only one spec, H100 has over 2TB/s of memory bandwidth and Apple silicon caps out at 400GB/s. H100 in general kicks the pants out of Apple silicon performance and has far more support. These parts aren't really in the same league but Apple Silicon has become a popular hobbyist choice for LLM inferencing.
Also see: The Jetson AGX Orin 64gb, Nvidia RTX A6000 48gb, AMD Radeon Pro W7900 48gb, two nvlinked 24gb 3090's.
[+] [-] bmitc|2 years ago|reply
[+] [-] EthicalSimilar|2 years ago|reply
Maybe not. It’s a bit of a pain having to unplug and replug everything in, not to mention having windows and such shuffle around each time.
[+] [-] dcchambers|2 years ago|reply
For the M1 and M2 generations, the airs released in the fall and the pros a few months later in the spring. The airs have gotten so good that many people that would have gotten the more expensive (and more lucrative) pros got airs instead.
[+] [-] minimaxir|2 years ago|reply
The presentation explicitly highlighted this very randomly, without showing how AI development actually works.
I know Apple has put more effort into Apple Silicon support for PyTorch but is it there yet?
[+] [-] nextos|2 years ago|reply
M3 Max might be a contender, but Metal is far from CUDA yet? A laptop M3 Max looks like it can compete with a Nvidia 3070.
It's an unfair comparison as Nvidia is a huge desktop heat/energy hog and Apple M chips are really efficient. Let's see how desktop M3 chips fare.
[+] [-] selectAll|2 years ago|reply
With the ability to capture in Log, it's possible that the next iPhone release might be filmed using the very phone that's being unveiled.
[1] Source: https://www.youtube.com/live/ctkW3V0Mh-k?t=30m02s
[+] [-] wslh|2 years ago|reply
Also waiting for a great Linux port battery/weight wise.
[+] [-] kmlx|2 years ago|reply
[+] [-] Eldandan|2 years ago|reply
[+] [-] starshadowx2|2 years ago|reply
[+] [-] whatever1|2 years ago|reply
Instant sleep/resume, great input devices and monitor, cpu power at will without overheating or loud fan, super fast shock resistant storage, 20+ hours of battery life, no battery drain when off.
It took us 30 years but we finally have it.
Now let’s hope WIntel can also catch up soon.
[+] [-] faizshah|2 years ago|reply
[+] [-] unclejack|2 years ago|reply
The RAM is still not ECC. They focus so much on the shallow marketing without shipping anything which makes a difference.
[+] [-] 015a|2 years ago|reply
I've had the same MacBook, in a similar timeframe. It gets just as much use. Its still rock-solid.
Your concern is a hypothetical one. The inability to upgrade has, in a VERY REAL sense, meant for me: I overbuy specs upfront, so I don't have to upgrade (and thus, generate e-waste). It also experientially means that the machines are more reliable. The least wasteful machine is one that doesn't have to be upgraded; not one that can be and needs to be. The 500gb of storage I have in my MacBook sometimes feels limiting; but the cost of upgrading (a whole new machine) has stopped me from actually going out and doing it; and thus less e-waste is generated.
Eventually I will brainstorm what to do with this machine once its outlived its useful life as a laptop. I'm thinking, server rack. We've got Asahi, its got thunderbolt so wiring up super fast storage drives is a cinch.
[+] [-] rollcat|2 years ago|reply
But the chassis is made from 100% recycled aluminum!~
On a serious note, Apple has a simple and working strategy, to which they're 100% committed: repair only in authorized service points, otherwise recycle. Making devices user-repairable adds costs and compromises on specs; you can't put a number on the disks being replaceable as easily as you can put a number on size, IO bandwidth, or number of write cycles, and numbers is what looks good on benchmarks.
You also have to consider Apple's scale: they have literally billions of deployed devices (everything from AirPods to Mac Pros) that they need to support, so what could work for a different vendor won't necessarily work for them.
(I don't necessarily agree with that strategy from end-user POV, merely pointing out the context.)
[+] [-] benterix|2 years ago|reply
These are nice machines, though, and I buy one for building iOS apps every couple of years - but usually 1-2 generations later so that the pricing is more reasonable.
[+] [-] anticensor|2 years ago|reply
[+] [-] mkl|2 years ago|reply
[+] [-] sosodev|2 years ago|reply
[+] [-] iAMkenough|2 years ago|reply
[+] [-] ribit|2 years ago|reply
This stuff is only really relevant if you are dealing with complex uber-shaders and recursive function invocations, both of which are fairly common in raytracing pipelines.
[+] [-] bischofs|2 years ago|reply
[+] [-] coverband|2 years ago|reply
[+] [-] miohtama|2 years ago|reply
Does this mean we can finally play Cyberpunk 2077 on a laptop? Is it going to be anywhere near as powerful as desktop Nvidia cards?
[+] [-] KAdot|2 years ago|reply
[+] [-] brucethemoose2|2 years ago|reply
My 2020 14" RTX 2060 laptop played it just fine, with decent settings (and copious DLSS). At 4K!
In the state I played it (earlier this year), it was great looking even without raytracing.
[+] [-] etchalon|2 years ago|reply
It's about as powerful as a PS5.
[+] [-] jayd16|2 years ago|reply