Nono it uses all cores it just uses them badly. The power of the cores is badly underutilized. It isn’t discarding one I don’t like, it’s discarding one that clearly has a problem. Again, it’s not much a problem for me because actually the perf/W for the M1 Max gets *better* relative to the Intel machine on CB23. The Intel machine has to draw even more than the 300-500% greater power here, not it because *it* is drawing more power but because the M1 is drawing so much less than it should be a single or all core workload. The difference is actually worse for the single core but still bad for multicore.
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.
Intel Alder Lake vs. Apple M1
- Thread starter leman
- Start date
- Sort by reaction score
Nono it uses all cores it just uses them badly. The power of the cores is badly underutilized. It isn’t discarding one I don’t like, it’s discarding one that clearly has a problem. Again, it’s not much a problem for me because actually the perf/W for the M1 Max gets *better* relative to the Intel machine on CB23. The Intel machine has to draw even more than the 300-500% greater power here, not it because *it* is drawing more power but because the M1 is drawing so much less than it should be a single or all core workload. The difference is actually worse for the single core but still bad for multicore.
This is far from clear. It has been pointed out that CPU power usage using Cinebench can be suspiciously low. For some reason Cinebench does not appear to utilize the CPU very well. This is why reviewers such as Anandtech use industry-standard benchmarks such as SPEC that are based on variety of different real-world workloads. And if you look at those, you'll notice that Cinebench is not a good predictor of CPU performance.
Actually it’s worse. It’s only low usage on AS (or macOS not sure which). On Wintel especially it seems to push the core(s) fully so it can do it, but something about the programming on AS/macOS prevents it from doing so.
Or perhaps the M1 simply doesn't deal well with sustained CPU-heavy loads.
But even in Geekbench, which is historically very Apple-friendly, the lowly i5 beats the M1 Max by about 8-9% in both single and multi, and the i9 by ~15% single and ~50% multi. So the notion that the M1 Max is on par with Intel's new desktop CPUs doesn't seem to hold much water.
On the mobile side, AMD's recent CPUs have kept up pretty well with the M1s too.
I think we have the next Intel ad:
“Hey Mac people, what would you say if I told you there’s a laptop that lets you percolate your own coffee, fresh, anywhere you have serious work to do? Now, what if I told you that laptop wasn’t from Apple?”
Perhaps, but this would contradict practically every single other sustained benchmark out there. In fact, there are very few CPUs that deal with sustained CPU-heavy tasks better than M1... since M1 is one of the very few CPUs out there that can maintain close to peak performance in both burst and sustained workloads.
Oh, I would never suggest that M1 Max is generally on par with Intel's desktop CPUs. At least not in burst workloads. Intel is definitely ahead here, at least in absolute terms. In sustained, depends on what you do. I have little doubt that M1 pro/max is going to be faster for, say, software development that the i5 Alder Lake (even despite the huge power consumption difference).
No we’re measuring peak power though sustained shows the same pattern and is in contrast to every other sustained benchmark. It’s using 3/5 of the power it should be in single core and only somewhat better in multicore.
Ummm the GB score I’ve seen it was more like 5% and even for single/multi but if we take your scores for single/multicore it’s still burning ~4x the power for 8% better benchmarks. Again, for Apple the Alder Lake microarchitecture is still in the rear view mirror.
Last edited:
Intel just needed to show something 'new' to sell against Zen 3. They could have discounted Comet Lake instead, at risk of further damaging the brand.
I didn't get any popups & native apps run as expected, so I don't think so. I'll keep an eye out.
Last edited:
You are not factoring the amount of work the Intel chip is doing in that amount of time. Your 500% less power draw means nothing if the Intel chip finishes the job over a min faster while the m1 is still running full load.
Depends on the context. Are you looking at pure performance? Work done per battery charge? Electricity costs? Also, what platform are we talking about? Desktop? Sure, a 10-core M1 does not look too good next to enthusiast-class 16+ core Intel or AMD CPUs if you go by raw performance only. Mobile? Now, that's something else entirely.
Based on what I have seen, I’d expect the pro M1 Macs to offer performance between the to the Alder Lake i5 and i7 models in variety of real work tasks. It’s desktop-level performance in a compact, lightweight laptop. Who else can offer something like this currently? Mobile Alder Lake i9 will be close (or even better) in burst (still consuming 2-3x more power), but it will tank in sustain.
Last edited:
I am … the difference in performance isn’t big enough to meaningfully change the joules used. 3 or 5/1.05 = 2.85 or 4.76 more joules. Even if the performance is 10% that still doesn’t change the underlying fact that the i5 is using multiples more power for a very small percentage increase in performance and often none at all.
I'm not sure what you're trying to say here. The new Intel chips have now caught up with old AMD chips. AMD is releasing THEIR new chips in a few months, and I have a feeling they will be easily on top again.
Yes but let's not forget Intel is still on 10nm. 7nm is right around the corner and it looks like it's finally ready for mass production. Next year's intels chips will finally be competitive in power efficiency with a slide showing 20 % power drop on 7nm and another 18% on 5nm.
And by the time Intel actually gets to 7 nm (instead of falsely marketing their 10 nm with the number 7), which everyone is already on, the competition will be on an even smaller process. Intel is still playing catch up at this point. AMD is going to release their new chips soon and they will still be a lot more power efficient than Intel while I can safely assume be even more powerful than their current offerings. AMD has been on a roll and isn't showing any signs of stopping. Much like Apple.
Intel’s 10nm is already equivalent if not better than TSMC’s 7nm. Supposedly AMD’s next chips out at the same time will be on TSMC 5nm+ for another 25-30% efficiency gain for them as well. Intel’s next node is too long away to reliably say when it will enter production. For instance TSMC 3nm is supposed to be next year but has already been delayed. Fabrication gets harder from here for everyone. Everyone’s roadmaps are suspect.
To be fair, the reason Intel rebranded is because everyone else’s numbers were even more BS (especially Samsung). Intel 10nm is better than TSMC’s 7nm, it’s more like 7nm+. However yes next year AMD will be on a new node as well and they currently have a microarchitecture PPW advantage.
I want it to look stock from the front and sides, I do have a plan thoughIf you're gonna be modding the case anyway, fit a radiator on the top of that bad boy and water cool it.
You can look at Intel a couple of ways. They are a business and as such has put profit ahead of innovation and they have been very successful at making a profit on their chips. I think it is actually amazing they have stayed competitive with AMD and Apple when you think about it. Apple is on 5nm process and AMD is on 7nm process while Intel is going to be on it's third or fourth gen on 10nm enhanced super fin or process 7 and they are still competitive.
Anyone who understands CPU architecture understands the slight of hand that Intel is playing. Instead of going to a true 7nm process by Intel's standards they are renaming the old architecture something new and adding a bunch of efficiency cores to help with heat and battery life. The problem is when all cores are running under load the thermals will still be a problem as others have noted.
So it is a little unfortunate that Intel is playing games with semantics but they have been doing these things for years with core branding.
The good news for Intel is that while Apple doesn't have a lot of room in terms of process reduction going forward. Intel still does. So when Intel finally gets to 5nm or 3nm with hybrid architecture they are implementing in Alder Lake they should still see pretty big performance gains and less heat.
I like the fact Intel makes chips in the USA and as a national security interest I want to see Intel succeed but when they sit on their ass instead of innovate and then just rename something to sound better while cranking up the frequency and heat it doesn't bode well for them.
I think the future of these chips is exciting because in my opinion we have not seen real computer chips yet. What I mean is that around the 5nm mark we are seeing a dramatic shift in efficiency and performance that has simply not been possible before now. Obviously the architecture such as ARM or x86 makes a huge difference as well but the process node is a fundamental shift in what is possible. Once AI neural engines are improved and new substrates are found that are as cheap as silicon but perform better like graphene etc we will have the next leap in computing but for now we are constrained by the physics of silicon.
Until I see Intel get to a better process and figure out the heterogeneous multi processing I am going to stick with Apple. AMD has a great opportunity but long term I do not think they can hold on to their advantage and they lack support in other areas like drivers, etc. Neither AMD nor Intel will be able to match ARM in efficiency so again it is interesting times to see what is ahead!
I do not agree. I have used AMD 5000 mobile processors and Intel 11th gen Tiger lake processors and there is a huge difference in terms of efficiency and heat. Tiger lake gets very hot very fast and AMD just doesn't. AMD's 7nm process is both more efficient and less thermally constrained. The Alder Lake chips will be using the same Architecture as Tiger Lake-they just renamed it. Maybe it is a slightly revised process but Intels 10nm process is not the same and AMD's 7nm and certainly nowhere near Apple's 5nm.
That probably has to do with the architecture, not the process itself. 12600k and 127000k are extremely competitive against 5800x and 5900x respectively. Alder Lake isn't using the same architecture as Tiger Lake. Alder Lake is using Golden Cove cores while Tiger Lake is using Willow Coves.
This is exactly the reason why Intel decided to rename their 10nm to 7. TSMC and Samsung has been using false marketing node names for so long but wasn't the problem when Intel had a big lead in node, but now that TSMC has caught up and surpassed Intel on process node, Intel's node name looks more behind than it was before when Intel is actually only a node behind.