M3 core counts and performance

Attachments

  • Screenshot_20231107-025454.png
    Screenshot_20231107-025454.png
    117.4 KB · Views: 40
What an impressive release. M3 family punches way beyond its weight when just looking at the specs. Also, a very strong first-generation RT offering from Apple. Now if it turns out that the GPU can run faster in a desktop chassis, Apple could pose some serious challenge for Nvidia in this market.
 
leman said:
What an impressive release. M3 family punches way beyond its weight when just looking at the specs. Also, a very strong first-generation RT offering from Apple. Now if it turns out that the GPU can run faster in a desktop chassis, Apple could pose some serious challenge for Nvidia in this market.
Click to expand...
What would we expect in Blender. I was thinking between 5000 and 6000, but now I think that was far too optimistic. Maybe 4500?
 
The fact that we are seeing these improvements when Apple barely improved raster this gen is awesome but when they do improve raster the difference will be even greater.
 
Jimmyjames said:
Ok that’s good but it really puts into perspective how far ahead Nvidia is here.
Click to expand...

Look at it in this way: before M3, Nvidia's main advantage was higher clocks and advanced features. With M3, Nvidia's advantage is only higher clocks. Apple reaching performance parity on complex shaders and hardware RT with Nvidia on iso-TFLOPS is a very big deal. Also, if we look at GPU performance on complex shaders without hardware RT, the RTX 4070 is only 22% faster than M3 Max — and that's a GPU which is nominally 60-80% faster in compute!
 
leman said:
What an impressive release. M3 family punches way beyond its weight when just looking at the specs. Also, a very strong first-generation RT offering from Apple. Now if it turns out that the GPU can run faster in a desktop chassis, Apple could pose some serious challenge for Nvidia in this market.
Click to expand...
We’ll have to wait and see, but it’s interesting that under the “High Performance Mode” the M3 Max MacBook Pro’s fans can be heard (as per the reviews above). IIRC, the difference between normal mode and high performance on M1 generation was so minimal that I wondered why was that a feature at all. It’s possible that the M3 SoC can be clocked higher but Apple doesn’t want to compromise the silent operation, so there’s a chance we could see higher clocks on desktop. Higher power consumption on desktop would also explain the huge heatsink of the Mac Studio (which seemed a bit overkill por the power consumption of the M1 Max / Ultra).
 
Andropov said:
Oh, this is a very interesting point I had missed. I was focusing mostly on performance, so the M3 Pro looked like the “weakest” upgrade. However, being 6P+6E instead of 8P+4E means it’s going to be so much more power efficient than the M2 Pro. Improved process node + newer µArch + higher fraction of the multi core performance achieved by the more efficient E cores. It’s now “weaker” it’s just balanced more towards efficiency this time. Hmm, interesting.
Click to expand...

The low power consumption/better battery life of the new Pro SOC design should be a killer feature for the 14” Pro/Pro model and even the Mac Mini/Pro (especially when used as a server).
 
leman said:
Look at it in this way: before M3, Nvidia's main advantage was higher clocks and advanced features. With M3, Nvidia's advantage is only higher clocks. Apple reaching performance parity on complex shaders and hardware RT with Nvidia on iso-TFLOPS is a very big deal. Also, if we look at GPU performance on complex shaders without hardware RT, the RTX 4070 is only 22% faster than M3 Max — and that's a GPU which is nominally 60-80% faster in compute!
Click to expand...
Plus, RTX 4070 has a TBP (total board power) of 200W. Haven’t seen power numbers for the M3 Max GPU yet, but I’m guessing it’s somewhere in the <50W range.
 
dada_dave said:
The low power consumption/better battery life of the new Pro SOC design should be a killer feature for the 14” Pro/Pro model and even the Mac Mini/Pro (especially when used as a server).
Click to expand...
This. Can’t wait to test it for myself.
I would be interested to see Apple push efficiency even further in the future - e.g. an M4 Pro with 4P+12E for example 🤤
 
dada_dave said:
Can you imagine the Macrumors forums if they did that?
Click to expand...
lol it would be pandemonium 🔥

4P+12E would be quite a beast though. We know the new efficiency cores are no joke from A17 Pro low power mode benchmarks. The multi thread performance would be amazing for just ~25W (assuming 4.1W per P core and 0.5W per E core)
 
It almost feels like in terms of gpu, Dynamic caching is more impressive than RT. That might reveal my lack of knowledge regarding RT however. I expected a 2x increase. What is the normal increase for Optix? Judging from Blender data, it seems to vary from over 2x for the 4070 laptop, to around 50% for the desktop 4090.
 
The M3 gen is so impressive. This is definitely not ina stop gap generation. I now know what Gurman meant by stopgap. The M2 didn't add any new features while the M3 did.

This generation will take weeks of research to see the full deep dive of the benefits that M3 brings. How it impacts productivity, games, emulation and battery life.
 

Similar threads

T
9to5Mac rumor: "New 13- and 15-inch MacBook Air models will both use M3 chips"
Replies
15
Views
966
Clix Pix
Clix Pix
T
ARM-based NVIDIA Grace server chip with 500 GB/s UMA claims lower power and higher performance than competing Intel & AMD designs
2
Replies
32
Views
2K
mr_roboto
M
Eric
Meta signals more layoffs could be coming after 'thousands' of workers are reportedly given low performance reviews
Replies
11
Views
386
Nycturne
Nycturne
Back
Top