m3 max 14c powermetrics wattage wrong? 14" vs 16"

[djcristi]

maybe that's why on the video encoding test on post #2 , the diff is that high also. because it uses more mem i guess.

 

[leman]

maybe it's more on browser test because of memory usage that maybe draws a lot of power also?!

That is possible. The thing is, we don't really know what that web test is doing. The difference could be cause of memory, or it could be anything else. My test only stresses the CPU, it doesn't really touch the memory subsystem at all.

 

[djcristi]

and a gpu test , https://www.matthew-x83.com/online/gpu-test.php , 1000 spheres


14” m3 max 14c

low-power 1.03 ghz, score 9-10 fps
total 35, backlight-pdbr 2, phpc 12-32, phps 22-24, powermetrics pkg 18-20, gpu 16.5-17.5

auto , gpu 1.38 ghz, score 13 fps
total 63-66, backlight-pdbr 2, phpc 40-60, phps 40-43, powermetrics pkg 36, gpu 34

idle (any mode) 0.34 ghz

 

[TommyYOyoyo]

why powermetrics show such low power usage when the actual (phpc) is much higher?

also the low-power mode on 14 has lower frecq than on 16 (2.17 vs 2.5 Ghz) , so lower performance and power consumption.

so i have both on base model Max , 14c , 36gb , 1tb.

=========
test: https://cpux.net/cpu-stress-test-online , 10 thread at 100%

----

14” m3 max 14c

Low power

freq : p-core 2.6 idle/ 2.17 load ; e-code 1-2 idle/ 2.55 load
load:
power: smc reports 26W phpc (9.5 display pdbr, 42 total system), and powermetrics only 15W cpu/pkg.
performance: 1300 h/s

View attachment 2327327

Auto/high-power

freq : p-core 2.7-4 idle/ 3.5 at start load ; e-code 1-1.3 idle/ 2.4 load
load:
power: smc reports 62W phpc (9.5 display pdbr, 85 total system), and powermetrics only 46W cpu/pkg.
performance: 2100 h/s


but as you can see in the graph after short time it stats throttling, and fans slowly increase
after I maxed the fans the power/perf increased, keeping a consistent on this values:

freq : p-core 3.2
load:
power: smc reports 50W phpc (9.5 display pdbr, 75 total system), and powermetrics only 35W cpu/pkg.
performance: 1900 h/s

View attachment 2327328

===================

16” m3 max 14c

Low power

freq : p-core 2.62 idle/ 2.5 load ; e-code 1-2 idle/ 2.55 load
load:
power: smc reports 33W phpc (12w display pdbr, 52 total system), and powermetrics only 21W cpu/pkg.
performance: 1600 h/s

View attachment 2327350

Auto/high-power

freq : p-core 3.5-3.7 idle/ 3.58 load ; e-code 1-1.3 idle/ 2.55 load
load:
power: smc reports 60W phpc (12w display pdbr, 84 total system), and powermetrics only 48W cpu/pkg.
performance: 2120 h/s


does not throttle at all, at some point I’ve set the fans to maximum but that decreased temps , so the 16 has a lot more case+fans+heatpipe cooling power

View attachment 2327330

Hey, I've just noticed the same pattern with my M4 Pro 12c MacBook Pro.

While benchmarking on Cinebench 2024, Powermetrics showed that my CPU+GPU+ANE are consuming alltogether 40W, while the "PHPS" sensor was showing 53W of consumption. The system total power consumption during this was around 68W.

When running the script written by Leman, Powermetrics showed that my CPU+GPU+ANE are consuming alltogether 48W, which was already higher than the M4 Pro TDP I found online, but "PHPS" read an even higher 54W power consumption, and the system total consumption was around 58W.

Does this mean that the M4 family's TDPs published online (e.g. M4 Pro 12-core has a 38W TDP) are not measured accurately, and that the true package TDP is much higher?

I don't know which sensor to trust anymore... Powermetrics or PHPS?

 

[sauria]

Hey, I've just noticed the same pattern with my M4 Pro 12c MacBook Pro.

While benchmarking on Cinebench 2024, Powermetrics showed that my CPU+GPU+ANE are consuming alltogether 40W, while the "PHPS" sensor was showing 53W of consumption. The system total power consumption during this was around 68W.

When running the script written by Leman, Powermetrics showed that my CPU+GPU+ANE are consuming alltogether 48W, which was already higher than the M4 Pro TDP I found online, but "PHPS" read an even higher 54W power consumption, and the system total consumption was around 58W.

Does this mean that the M4 family's TDPs published online (e.g. M4 Pro 12-core has a 38W TDP) are not measured accurately, and that the true package TDP is much higher?

I don't know which sensor to trust anymore... Powermetrics or PHPS?

Thanks for the report. I assume there is more going on here than differences in the cooling ability between the smaller and large larger laptops?

 

[TommyYOyoyo]

Thanks for the report. I assume there is more going on here than differences in the cooling ability between the smaller and large larger laptops?

Yeah, I completely agree with you. However, I just discovered an interesting pattern that's worth checking. I might be wrong though...

I noticed that Windows laptops also have a large power overhead other than the processor power, just like macs. For instance, during Cinebench R23, the Flow Z13 with AMD Ryzen AI MAX+ 395 SoC, for which the "CPU Package Power" on HWiNFO reads around 70W of sustained power under load, uses around 110W total system power (without display!) according to notebookcheck. Similarly, the Zenbook S16 with Ryzen HX370 SoC consumed around 28-33W CPU Package Power under load, while the system total power reached around 46W without display.

I noticed the pattern of the total system power being around 150-160% the CPU Package Power on windows machines with SoCs, which correlates with the test results on my M4 Pro 12-core mac (40W CPU Power, 68W system power, with the unknown PHPS sensor being 52W).

Although, according to the same source, the M4 Pro 14-core (~46W CPU Power on powermetrics according to many sources) and the M4 Max 16-core (～50-57W CPU Power on powermetrics) were consuming accordingly around 80W and 98W total system power with Cinebench workflow. The power overhead on those machines seemed quite excessive (total system power being more than 160% the CPU Power) according to the pattern I concluded.

So, this puts me in doubt again whether I should use powermetrics or the unknown PHPS sensor to accurately read true CPU Package Power. It's sad to see Apple removing the "Package Power" metric of Powermetrics at some point in the past few year.

 

[sauria]

Yeah, I completely agree with you. However, I just discovered an interesting pattern that's worth checking. I might be wrong though...

I noticed that Windows laptops also have a large power overhead other than the processor power, just like macs. For instance, during Cinebench R23, the Flow Z13 with AMD Ryzen AI MAX+ 395 SoC, for which the "CPU Package Power" on HWiNFO reads around 70W of sustained power under load, uses around 110W total system power (without display!) according to notebookcheck. Similarly, the Zenbook S16 with Ryzen HX370 SoC consumed around 28-33W CPU Package Power under load, while the system total power reached around 46W without display.

I noticed the pattern of the total system power being around 150-160% the CPU Package Power on windows machines with SoCs, which correlates with the test results on my M4 Pro 12-core mac (40W CPU Power, 68W system power, with the unknown PHPS sensor being 52W).

Although, according to the same source, the M4 Pro 14-core (~46W CPU Power on powermetrics according to many sources) and the M4 Max 16-core (～50-57W CPU Power on powermetrics) were consuming accordingly around 80W and 98W total system power with Cinebench workflow. The power overhead on those machines seemed quite excessive (total system power being more than 160% the CPU Power) according to the pattern I concluded.

So, this puts me in doubt again whether I should use powermetrics or the unknown PHPS sensor to accurately read true CPU Package Power. It's sad to see Apple removing the "Package Power" metric of Powermetrics at some point in the past few year.

Yes, obfuscation is never good. I think specifications should be clear for cross comparison.