471–480 of 1001 posts
Earlier quoted context omitted.
These are not low/mid tier machines when talking about "consumer-grade".
Within the MacBook Pro lineup, they are objectively the low and mid-grade pricing tiers.
Earlier quoted context omitted.
Sure, that's the title, but at least in this PR they immediately show a graph with a comparison to both. Presumably it makes more marketing sense to compare to the M1 family up front because most people that bought an M2 last year are probably not going to be upgrading to M3. They are speaking to the people most likely to upgrade.
fwiw, i cant remember the last time i saw a company go back more than a generation in their own comparison. Apple is saying here as much as they're not saying here. M2->M3 may not be a compelling upgrade story.
I’m currently at 4 to 5 years on laptops and 3 to 4 years on phones, and even then I hand them over to kids/friends/family who get a bit more use out of them.
Earlier quoted context omitted.
Why do I need to select the most expensive M3 Max version (the one with 16 cores) to get 48 or 64 gigs of RAM. The M3 Max with 14 cores only allow 36 or 96 gigs. Just let me choose 14 cores and 64 gigs.
This kind of blows my mind. I'm running the lowest spec m1 with 8 gigs of ram for my daily driver. I frequently have tens of chrome browser tabs open across multiple profiles, multiple chrome debuggers, at least two or three VS code instances, multiple high memory nodejs processes, video conferencing, screen sharing and building code all at the same time - and the thing runs smoothly without a hiccup. And does it on…
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.
The presentation mentioned dynamic GPU caching: that seems like something transformer models would like.
Llama.cpp (and MLC) have to read the all the model weights from RAM for every token. Batching aside, there's no way around that.
"No matter what your passions are in work or in life, there's a Macbook Pro that's perfect for you." For the vast majority of Apple customers, Macbook Airs are plenty powerful. They're also lighter weight, and significantly cheaper.
Looks like the latest MBA is only 0.1 lbs less than the equivalent MBP at 3.4 lbs. Feels like they used to be lighter.
I don't like my M2 MacBook Air nearly as much as the M1, in terms of the physical feel and dimensions.
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…
I'm wondering now if Apple tracks this sort of stuff in their released machines. I know my iPad always asks me if I want to send stats to Apple (and I always say no). So let's say that enough people do, then do they have a good idea of how often all the performance cores are used? Max memory B/W consumed? Stuff like that. Back when I was at Intel there were always interesting tradeoffs between available silicon/therm…
Given that there has never been any public incidents about it and what we know about similar defaults I would be surprised if Apple is getting less than 95% opt-in rate.
But I suspect at high-end they only really care about the performance of a few dozen professional apps e.g. Logic or Final Cut. And at the low-end it's likely just efficiency.
I'd prefer if there'd be higher RAM limits for each variant: * ~~Ultra~~ Max probably has an adequate limit with "up to" 128GB * Pro would be nicer to have "up to" (say) 64GB, rather than 'only' 36GB * Base perhaps "up to" 32GB, rather than 24GB There are some situations (running a VM or three with different OS(es)) where there really isn't a substitute for more bits.
IMO, if I'm doing that heavy of a workload, I'd rather SSH to a server instead of trying to do it on hardware optimized to work on a battery.
Earlier quoted context omitted.
Faster than what? M1 Pro? Just barely.
2.5x is "just barely"? lol k.
That's only rendering speed, and M3 Max vs M1 Max (not Pro). M3 Pro is only 30 percent faster:
> The 12-core CPU design has six performance cores and six efficiency cores, offering single-threaded performance that is up to 30 percent faster than M1 Pro.
> 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 20…
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.
Earlier quoted context omitted.
Sure, that's the title, but at least in this PR they immediately show a graph with a comparison to both. Presumably it makes more marketing sense to compare to the M1 family up front because most people that bought an M2 last year are probably not going to be upgrading to M3. They are speaking to the people most likely to upgrade.
fwiw, i cant remember the last time i saw a company go back more than a generation in their own comparison. Apple is saying here as much as they're not saying here. M2->M3 may not be a compelling upgrade story.