Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
M4+ Chip Generation - Speculation Megathread [MERGED]
- Thread starter GoetzPhil
- Start date
- Sort by reaction score
Do keep in mind that the Oryon runs at ~ 8% higher clock to achieve this result. Knowing the power consumption of these cores will be key.
Got my iPhone Air
Geekbench 6 results:
3577 / 9166 🥵 (warm, adaptive power mode enabled, some background activity) https://browser.geekbench.com/v6/cpu/13926502
3767 / 9588 😎 (cool, adaptive power mode disabled, no background activity) https://browser.geekbench.com/v6/cpu/13937751
3934 / 10479 🥶 (on an ice pack, adaptive power mode disabled, no background activity) https://browser.geekbench.com/v6/cpu/13938931
Geekbench 6 results:
3577 / 9166 🥵 (warm, adaptive power mode enabled, some background activity) https://browser.geekbench.com/v6/cpu/13926502
3767 / 9588 😎 (cool, adaptive power mode disabled, no background activity) https://browser.geekbench.com/v6/cpu/13937751
3934 / 10479 🥶 (on an ice pack, adaptive power mode disabled, no background activity) https://browser.geekbench.com/v6/cpu/13938931
I just learned that the A19 on the iPhone 17 has only 5 GPU cores. I was convinced that it had a 6 core GPU.
Also the A19 has smaller caches than the A19 Pro.
All of this is kinda discouraging me from getting an iPhone 17, but I don’t want or need the 17 Pro…
Hopefully the base M5 will be based on the A19 Pro.
Also the A19 has smaller caches than the A19 Pro.
All of this is kinda discouraging me from getting an iPhone 17, but I don’t want or need the 17 Pro…
Hopefully the base M5 will be based on the A19 Pro.
A19 Pro in the iPhone Air also has only 5 GPU cores. Furthermore, when I checked several days ago, it seemed to perform no better or maybe even slightly worse on average in Geekbench 6.5 CPU than the A19 non-Pro in the iPhone 17, perhaps because of throttling. Also, see @aaronage's post above.
It appears that each SKU is configured separately, to meet the target goals in performance, power efficiency, and costs. I am sure similar will apply to M5 family.
As to the cache size difference - why does one need that much cache in a phone? Do you plan to run complex productivity apps or something?
Actually, no. But I purchase an iPhone every several years and I thought more cache would translate into better memory management, improved AI performance (for local models coming in the future) and overall better performance in the long term.
Yeah, maybe a 6 core GPU on the iPhone 17 could make it throttle and cause excessive heat and/or power draw. As long as the UI doesn’t become sluggish over time I’m okay with that.
I’d be very surprised if the software got as complex in the next few years for the cache sizes to make a difference. To put things in perspective - L2 of 8MB per large core in the A19 Pro is 8x larger than that of server and workstation CPUs. You’ll be fine
The biggest effect is probably battery life - larger caches mean fewer request to DRAM, which in turn means less energy spent.
You're worrying waaaayyyyy too much if this is just about UI sluggishness. Even my A14 iPhone 12 Pro Max doesn't get very sluggish, and that has just over half the CPU performance (both ST and MT) of A19 non-Pro. The main concern would be RAM, as A19 Pro is paired with 12 GB RAM whereas A19 is 8 GB, but even then it's not going to be a significant issue in the near term, since iPhone 16 Pro Max has "only" 8 GB RAM. My 12 Pro Max has 6 GB RAM, and I'd be perfectly happy using it as my main driver for a couple more years from the OS peppiness standpoint. My main reason for upgrading is to get the vastly upgraded telephoto camera.
Okay okay, gotcha, the A19 is as awesome as I thought initially and there’s not a lot of difference between the three SoCs.
Impressive. I’m sensing a business here… I’m sure there’s market for a MagSafe IcePack. You put it into the freezer, and whenever you need that extra performance, you attach it like you do with the MagSafe Battery.
Last edited:
Or just a magnetic piece of cold, that uses power to stay cold, so you have to put a magsafe battery on the ColdMag too.
Which reminds me:
Whatever happened to ball-gate piezoelectric thermal heat-exchange tech?
Was that a Delorean Moment, or has it just become the ball-holder under the pile?
In my mind, exothermic transference == metropolitan traffic: no matter how many roads and connectors are built, congestive buildup will always happen....
Whatever happened to ball-gate piezoelectric thermal heat-exchange tech?
Was that a Delorean Moment, or has it just become the ball-holder under the pile?
In my mind, exothermic transference == metropolitan traffic: no matter how many roads and connectors are built, congestive buildup will always happen....
You can already purchase peltiers (thermoelectric heat pumps) with magnets that can attach on phones and tablets.
Some also come with MagSafe charging, like the AUKEY MagFusion GameFrost.
Some also come with MagSafe charging, like the AUKEY MagFusion GameFrost.
Will Macs get Apple’s new memory protection?
Malicious software often relies on corruption and other problems with memory. Latest iPhones come with new hardware-based protection that’s also likely to come in a future M-series chip. Here…
eclecticlight.co
With respect to memory protection: My (perhaps faulty) understanding of OS/2 v1.x was that it used the 80286's descriptor tables for memory allocation to assign a unique segment for each allocation request. The segment descriptor included the size of the memory allocation. Any attempt to access memory beyond the assigned limit would generate an exception. There were reports that MS Windows developers preferred to do initial development wok on OS/2 as it caught a lot of memory errors. The downside of the 80286 was the maximum size for a segment was 64K.
Qualcomm is teasing details about its next PC chip (and future M5 competitor)
www.windowscentral.com
Snapdragon X2 Elite due early next year with up to 18 (12P+6p) Oryon Gen 3 cores clocked up to 5GHz (single/dual core boost)
Qualcomm shared Geekbench 6 numbers from a reference 8 Elite Gen 5 device. It manages ~3800 points single thread. Probably safe to assume the reference device is hitting the peak 4.6GHz boost during the test (because reference device). That suggests a 5GHz Oryon Gen 3 would be in the 4100 points range.
So… X2 Elite probably sits in the M4 Max range for single thread performance. Probably a bit faster than an M4 Max 12P+4e for multi thread work. Full-fat M4 Max can chug a lot of power so it might actually compare reasonably well on multi-thread efficiency (can’t imagine Qualcomm would let this thing pull more power than M4 Max, but maybe!).
They didn’t share much about the GPU which isn’t a good sign. If I had to guess, the GPU probably continues to be a weak point (the first X Elite GPU wasn’t great).
Doesn’t look like this chip will compare favourably to M5 Max on anything but maybe CPU multi-thread performance. Exciting chip nonetheless - it’ll be interesting to see what improved on the first X Elite.
Hopefully they have Linux support ready this time because this silicon is wasted on Windows (wishful thinking? 😅).
Qualcomm is coming for Apple's M4 Max with its new Snapdragon X2 Elite Extreme SoC — coming in 2026
Snapdragon X2 Elite and X2 Elite Extreme are Qualcomm's new SoCs for the Windows PC market, with a focus on high performance workloads and great efficiency that rivals the latest from Apple, AMD, and Intel. But not until 2026.
www.windowscentral.com
Snapdragon X2 Elite due early next year with up to 18 (12P+6p) Oryon Gen 3 cores clocked up to 5GHz (single/dual core boost)
Qualcomm shared Geekbench 6 numbers from a reference 8 Elite Gen 5 device. It manages ~3800 points single thread. Probably safe to assume the reference device is hitting the peak 4.6GHz boost during the test (because reference device). That suggests a 5GHz Oryon Gen 3 would be in the 4100 points range.
So… X2 Elite probably sits in the M4 Max range for single thread performance. Probably a bit faster than an M4 Max 12P+4e for multi thread work. Full-fat M4 Max can chug a lot of power so it might actually compare reasonably well on multi-thread efficiency (can’t imagine Qualcomm would let this thing pull more power than M4 Max, but maybe!).
They didn’t share much about the GPU which isn’t a good sign. If I had to guess, the GPU probably continues to be a weak point (the first X Elite GPU wasn’t great).
Doesn’t look like this chip will compare favourably to M5 Max on anything but maybe CPU multi-thread performance. Exciting chip nonetheless - it’ll be interesting to see what improved on the first X Elite.
Hopefully they have Linux support ready this time because this silicon is wasted on Windows (wishful thinking? 😅).
In all honesty the weak point is absolutely Windows on ARM.Qualcomm is teasing details about its next PC chip (and future M5 competitor)
Qualcomm is coming for Apple's M4 Max with its new Snapdragon X2 Elite Extreme SoC — coming in 2026
Snapdragon X2 Elite and X2 Elite Extreme are Qualcomm's new SoCs for the Windows PC market, with a focus on high performance workloads and great efficiency that rivals the latest from Apple, AMD, and Intel. But not until 2026.
Snapdragon X2 Elite due early next year with up to 18 (12P+6p) Oryon Gen 3 cores clocked up to 5GHz (single/dual core boost)
Qualcomm shared Geekbench 6 numbers from a reference 8 Elite Gen 5 device. It manages ~3800 points single thread. Probably safe to assume the reference device is hitting the peak 4.6GHz boost during the test (because reference device). That suggests a 5GHz Oryon Gen 3 would be in the 4100 points range.
So… X2 Elite probably sits in the M4 Max range for single thread performance. Probably a bit faster than an M4 Max 12P+4e for multi thread work. Full-fat M4 Max can chug a lot of power so it might actually compare reasonably well on multi-thread efficiency (can’t imagine Qualcomm would let this thing pull more power than M4 Max, but maybe!).
They didn’t share much about the GPU which isn’t a good sign. If I had to guess, the GPU probably continues to be a weak point (the first X Elite GPU wasn’t great).
Doesn’t look like this chip will compare favourably to M5 Max on anything but maybe CPU multi-thread performance. Exciting chip nonetheless - it’ll be interesting to see what improved on the first X Elite.
Hopefully they have Linux support ready this time because this silicon is wasted on Windows (wishful thinking? 😅).
Windows developers are just 5 years behind at this point (despite Windows on ARM have been a thing long before Apple Silicon).
True!
I have a Surface Laptop 7 with X Elite to play around with. Windows on Arm itself is ok - x86 translation works fine, the system is fast and stable and all that. The foundations are as solid as they can be.
The issue is that developers and users are apathetic towards Windows as a whole. It’s an ******tified legacy platform that’s tolerated rather than loved.
IMO, the Arm PC guys (Qualcomm, Nvidia, Mediatek etc.) should be focusing as much on Linux as Windows. Linux is where all the exciting developments in PC are happening (e.g. Steam OS for gaming). Focusing on Windows is shortsighted.
