Oof these scores are terrible given the power. The rest of your message I don’t comprehend, sorry.
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.
M3 Chip Generation - Discussion Megathread
- Thread starter scottrichardson
- Start date
- Sort by reaction score
Total system usage != cpu power usage. 78W for single cpu core is unrealisticAccording to the recent review at Anandtech of the i9-14900K, between 50 and 70 watts on a loaded single threaded task, with an observation of 78 Watts.
View attachment 2305693
I think this is socket power. And I don’t know if it’s unrealistic. Coffee lake mobile Intel CPUs pulled 25 watts on single core (and that was core power, not socket power), and the power consumption skyrocketed since then.
Only because Geekbench doesn't utilise all cores at 100%.
Its a fair comparison as both chips have the same objective to compute and one did it in less time.
To quote GB:
"...n Geekbench 6, one workload is used and all the cores work together on that one shared objective. It is still true that the more cores you have, the quicker it will complete. However, there is now interaction between the cores."
Yeah, most multi threaded tasks don’t utilize all the cores at 100%. Use Geekbench 5 if you prefer the simpler, less realistic measure.
I find it hard to belive that 78W is realistic CPU power consumption even for 6GHz single core with whole package.
Only reviews that i can find with measurments of CPU power usage for 14900k are https://www.techpowerup.com/review/...ke-tested-at-power-limits-down-to-35-w/8.html and they hit top of 36W with encoding.
And even in mult-threaded scenarios with stupid "Turbo" total CPU package seems to draw 280W tops so 78W single core seems urealistic if this isnt total power draw.
As I'm only using CPU's that are not apple on servers atm its hard to tell how to feed single core almost 80W of power without thermal limiting it in like few ms (maybe in some ultra overclocking scenario with 9GHz)
But I'm sure there are people who would answer this and maybe idle power usage of whole CPU package is just around 25W+ in modern intel cpus just like my old ryzen 3900x, so it would make extra 40-50W for single core plausible.
What specific issues do you have with the evidence I showed from the Anandtech article? I am happy to hear criticisms of their methods. At the moment all I can see is “it can’t be that, because it can’t be.”
To be fair, they don’t really document the methodology. It’s not clear what they are measuring exactly, for example. Or maybe I missed it? It is possible that @komuh is correct in suspecting that these scores are inflated.
What evidence this is just article that make claim about 78W power usage but without methodology and every other review/article dosen't even come close to that number.
And only article with measurments for CPU power draw tops at about 38W so 2x less.
There are many variables possible. I’m trying to find out the exact methodology, but my understanding is that it’s similar for all their tests. Motherboards vary in addition, as it says in the article. It stated 78 Watts was an observation. Most results were between 50 and 70 W. Which articles are you referring to? Please list them.
On many multicore tasks that warrant using an M2 Ultra; the ultra would beat the M3 Max handily.
You don't by an M2 Ultra to just run the algorithms used by GB6. You use for rendering or scientific applications that really use all the cores.
One could as well claim that the M3 beats the M2 Ultra in Geekbench single-core. It's true but it doesn't mean much.
I'm not arguing against that. I made a rough calculation based on the numbers I found. I have corrected my previous post but that didn't change much. Previously M3 used about 8 times less power (125/15). Now it uses 7-15 times less power. So a Macbook with base M3 8 cores at 4.05 GHz and 5W with the single-core score 3076 is close to the best Intel desktop system with i9-14900K 24 cores at 6 GHz and 35W to 78W with 3409.
Last edited:
This measurement is based on mp3 encode. This does not push the cpu like the anandtech test, which uses y-Cruncher and Cinebench-R23.
This is what I found for Anandtech’s method.
“ For our power consumption metrics, we performed our usual testing: using an affinity mask to limit the cores in use, implement a high-powered workload, and then measure the power readings 30 seconds in. We take the power readings from the processor itself, using the internal registers that are designed to regulate how much the processor does a form of turbo but also regulate temperatures and so forth. This method is broadly accurate, assuming the motherboard supports the external reporting of these values, but depending on the processor family it also gives us insights into how much power is being derived from the cores individually and the package as a whole.”
Seems reasonable to me.
“ For our power consumption metrics, we performed our usual testing: using an affinity mask to limit the cores in use, implement a high-powered workload, and then measure the power readings 30 seconds in. We take the power readings from the processor itself, using the internal registers that are designed to regulate how much the processor does a form of turbo but also regulate temperatures and so forth. This method is broadly accurate, assuming the motherboard supports the external reporting of these values, but depending on the processor family it also gives us insights into how much power is being derived from the cores individually and the package as a whole.”
Seems reasonable to me.
That's because the Geekbench 6 multicore benchmark is a software performance benchmark. Rather than measuring multicore performance, it measures how well various kinds of software can take advantage of that performance.
If you have a CPU with many cores, you probably bought it because the software you use can take advantage of those cores. A better benchmark might run several copies of Geekbench in parallel and report the sum of of multicore scores from those runs as the overall multicore score. For example, if you have M2 Ultra, you may want to run 3 copies of Geekbench and limit them to 8 cores each (using virtual machines if necessary).
Seems like the motherboard could be the issue, they redid some of the MT tests with a different board (same CPU) and got lower power usage.
Well it’s an “issue” in that it allows that much power without overclocking for popularity reasons. Other motherboards are less generous. Even so those results are valid. They measured those power levels
Oh I am not saying the results are not valid. Just found it interesting that different motherboards give different results, when they should all act the same(you know following specs and all).
Yeah, definitely. It’s the Wild West out there with some of these motherboards!
I have updated the Conversation from 'Speculation' to 'Discussion' Megathread 
Colour me mildly impressed by the M3 Max. Like others I find the M3 Pro an oddity, but only when looking at it next to the M2/M1 Pros. I guess Apple is trying to go for more differentiation between the models, which is fine.
I will say, I'm a little bit peeved as I only just upgraded to a fully maxed out M2 Max 16" MPB as my main workstation just 4 months ago. Kinda have matte black ray-tracing envy at at the moment, but likely can't really take much advantage of the ray-tracing, or the extra CPU cores at this point since mine doesn't break a sweat.
Keen to see more benchmarks. But by all accounts Apple continues to dominate with power/efficiency vs performance. To be right up there in single core and likely faster than almost everything with multicore all while unplugged on battery, using minimal power, is amazing!
GPU raster performance is the next thing I'm keen to see benchmarks for. Early notes suggest the GPU isn't actually much better?
Colour me mildly impressed by the M3 Max. Like others I find the M3 Pro an oddity, but only when looking at it next to the M2/M1 Pros. I guess Apple is trying to go for more differentiation between the models, which is fine.
I will say, I'm a little bit peeved as I only just upgraded to a fully maxed out M2 Max 16" MPB as my main workstation just 4 months ago. Kinda have matte black ray-tracing envy at at the moment, but likely can't really take much advantage of the ray-tracing, or the extra CPU cores at this point since mine doesn't break a sweat.
Keen to see more benchmarks. But by all accounts Apple continues to dominate with power/efficiency vs performance. To be right up there in single core and likely faster than almost everything with multicore all while unplugged on battery, using minimal power, is amazing!
GPU raster performance is the next thing I'm keen to see benchmarks for. Early notes suggest the GPU isn't actually much better?
If it's like the A17 Pro, it's not much better than its predecessor for rasterization.
If I may posit some ideas on Apple Silicon as a whole.....
I believe that the M1, M2 and M3 are all, essentially, the same base design. Sure the core designs have evolved a little, and the GPU has been optimised, and most recently we have added new features. But the core, 'CORE' design concept has vastly remained the same.
- Arm v8.x instructions for the CPU cores.
- The raster engine in the GPU hasn't changed much.
I believe between M1 and M2 they 'fixed' some GPU scaling issues to do with cache (correct me if I'm wrong).
Other than that, what we're seeing here is M1 with two generations of clock speed boosts thanks to process improvements, and more cores, also thanks to process improvements and die space.
I know I'm vastly oversimplifying, but we haven't seen MAJOR changes.
So - my point - I believe that, like Intel, they stick with the same foundational design concepts for a few 'generations' before making a large jump to the next thing. Assuming in our case it will be Arm v9, and new GPU cores.
Maybe M4 will be that new generation. A larger leap in IPC, and jump in single-core performance. That said - no other CPU manufacturer is getting huge IPC improvements these days, and for Intel even shrinking the process will just result in faster clocks and more cores more than likely.
It appears that Apple does need to improve raster performance on the GPU cores for them to be able to compete with the higher end parts of nVidia and AMD. While Apple has never really tried to compete at the top with GPU, there's no reason why they shouldn't want to, given they already do with CPU.
It’s a completely new GPU core design. Not an iteration.
Gokhan Avkarogullari (@gavkar) on X
It's been a long & satisfying journey to build this brand new scary fast GPU: never before seen architecture, HW ray tracing & mesh shading. Congratulations to all teams at Apple for making this great GPU, great M3 SoCs and great MacBook Pro products that Apple introduced today.