Apple M5 Pro & M5 Max CPU Analysis - M5 Max is not much faster than the M4 Max

[Redaktion]

Apple launched their new M5 Pro & M5 Max chips with different a different approach including tiles and new super cores. We test and analyze the new M5 Pro/Max CPUs, both in terms of performance as well as efficiency. Users of the new M5 Pro can be very happy, but the results of the new M5 Max is a bit sobering.

https://www.notebookcheck.net/Apple-M5-Pro-M5-Max-CPU-Analysis-M5-Max-is-not-much-faster-than-the-M4-Max.1246054.0.html

[PHVM_BR]

They're comparing the M5 Max in the 14-inch MacBook Pro with the M4 Max in the 16-inch laptop...

[Bhagat]

 

...new M5 Max CPU is not really much faster than the old M5 Max....

Typo.
Should've read "old M5 Max"

[bert22]

I've seen from other reviewers the M5 Max is faster in the 16". Ars Technica got over 2300 MT in cinebench 2024.

[dada_dave]

I think the M5 Pro Efficiency/Power analysis might be off? It has the same CPU as the M5 Max which you contend, not unreasonably is thermally constrained int the 14" model, but you claim the CPU draws over 100W? While the 14" M5 Max draw 85W and gets a better score. As @bert22 points out other reviews (with 16" M5 Maxes) get 2300 in CB R24.

Finally in your M5 Pro review article you write:

As shown in the table above, the CPU consumes up to 72 Watts under load and settles at 62 Watts under sustained workloads, so the overall performance is very stable.

So it sounds like it was running with significantly less power than you quote here - even if the above is powermetrics rather than wall power (unclear) I don't see how it gets to 100W. It seems like there may be an error somewhere.

[gescom]

"For potential customers of the M5 Max, however, the situation is different, because the new M5 Max CPU is not really much faster than the old M5 Max."

Typo / than the old M4 Max.

[Jimster]

Same nonsense posted each year with each new release. How about doing actual benchmarks with actual applications instead of running the same 2 benchmarks that are literally the most best case scenario for apple, that are the most optimized for apple?
How about being realistic with the workloads so that people can stop parroting pure propaganda by utilizing software that doesn't at all represent real world scenarios?

[dada_dave]

I think the M5 Pro Efficiency/Power analysis might be off? It has the same CPU as the M5 Max which you contend, not unreasonably is thermally constrained int the 14" model, but you claim the CPU draws over 100W? While the 14" M5 Max draw 85W and gets a better score. As @bert22 points out other reviews (with 16" M5 Maxes) get 2300 in CB R24.

Finally in your M5 Pro review article you write:

As shown in the table above, the CPU consumes up to 72 Watts under load and settles at 62 Watts under sustained workloads, so the overall performance is very stable.

So it sounds like it was running with significantly less power than you quote here - even if the above is powermetrics rather than wall power (unclear) I don't see how it gets to 100W. It seems like there may be an error somewhere.

A customer posted their 16" M5 Pro 18-core running CB R24 getting the same score (2059) here drawing around 46W. I really think the 100W must be a mistake somehow - maybe a bad chip?

[Andreas Osthoff]

As we shown in our review of the MBP 16, the CPU consumption on our test unit was higher than 46 Watts. Also keep in mind that this is the consumption of the CPU cores alone and cannot be exactly compared to the TDP for x86 processors. If you compare to Intel chips for example, the number of the M5 Pro is comparable to the IA cores value shown in HWiNFO.

In addition to that, the RAM and the rest of the system including the display consume power as well.

[dada_dave]

As we shown in our review of the MBP 16, the CPU consumption on our test unit was higher than 46 Watts. Also keep in mind that this is the consumption of the CPU cores alone and cannot be exactly compared to the TDP for x86 processors. If you compare to Intel chips for example, the number of the M5 Pro is comparable to the IA cores value shown in HWiNFO.

In addition to that, the RAM and the rest of the system including the display consume power as well.

Thanks for the response! I do understand all that. However, the Mac user I'm referencing was measuring wall power just like you do (not with as nice a multimeter, but still he had one) and on High Power mode with the same 16" M5 Pro system you were testing - they got theirs yesterday. Based on their description of performance and power draws over multiple runs, I think they were getting the same inconsistencies you were getting on the MBP14" but with the MBP16" M5 Pro - it's possible that in their case they were still indexing, but it sounds very similar (I'm assuming Apple sent you guys Macs that were done spotlight indexing, right?). Also worth noting he found CB R26 to be more stable (though still highly variable).

It just seems very improbable that the M5 Pro would draw 100W on average in your CB R24 test, which is 20% more than you measured with the M5 Max which has the same CPU and get the same/slightly lower score than that M5 Max. And again then there's this Mac user's results which have some questionable properties in and of themselves but had similar scores (I think) to yours at seemingly much lower power draw. Finally other review sites are getting much higher CB R24 scores though none of them had exactly your model (16" MBP M5 Max, but again the CPU is supposed to be the same with your M5 Pro).

I'm really not sure what's going on and I want to make it clear that I am not suggesting that you did something wrong in your testing. If it's faulty MBPs or something with Apple's firmware and driver settings (a couple of the 14" GPU results are similarly odd here, again maybe thermals, but ...), then the machine's results are what they are. As per what you wrote in the 14" MBP, I just hope you are able to get are able to get more machines to test - Pro and Max chips - because these are very strange results. Admittedly all the other 16" MBP results seem sensible enough, or at least nothing stands out like the 14" results, but that one 16" result is very strange. I know you guys doing computer reviews are on tight deadlines with overloaded schedules, but if you somehow have the opportunity to retest or dig deeper with Apple, that would be cool. Again, I understand if that's not really possible.

As we shown in our review of the MBP 16, the CPU consumption on our test unit was higher than 46 Watts.

FYI, I'm not sure if this is what you referring to, but you didn't actually put the CBR24 power measurement for external monitors on your Power Consumption external monitor graph in the full 16" M5 Pro review. You just have CP2077 and idle - the same is true for the 13" and 15" M5 Air reviews too, no CB R24 power draw over time graphed. The only review of the recent batch where the CB R24 power shows up is the 14" M5 Max review. I know you did it since those results show up in your analysis article and you do talk about it in the review, so I wasn't sure if that was a bug/oversight as you usually put those results up when you have them.

Cheers.

[Andreas Osthoff]

Thanks for the note with the power draw, I will add them later in the other three reviews!

We will get the MacBook Pro 16 M5 Max tomorrow and I will start testing immediately.

[dada_dave]

Thanks for the note with the power draw, I will add them later in the other three reviews!

We will get the MacBook Pro 16 M5 Max tomorrow and I will start testing immediately.

Awesome! It'll be fascinating to see the results!

[The usual fine 12-13%]

So the usual 12-13% per core, which is fine.

Should have tested browsing benchmarks (Mozilla Kraken, WebXPRT), as web-surfing is a major activity and the single-core performance is reflected in those benchmarks. But also since it's reflected, maybe testing not needed. Still. But these benchmarks can be found if you click on any of the MacBooks in the table.