Does the non-TouchBar 13-inch MacBook Pro throttle at all?

[TigerMSTR]

The 2016 non-Touchhbar 13-inch MacBook Pro CPU did not throttle. Is this also true of the 2017 model?

I'm asking primarily because the the non-Touchbar models score somehow slightly higher than the Touchbar models of the notebook in Geekbench. If it is true that the 2017 model still does not throttle, that would mean that the non-Touchbar model somehow outperforms the more expensive Touchbar model.

For reference, the base CPU scores of the 2017 13-inch MacBook Pros:
Multicore
Without Touchbar: 9293
With Touchbar: 8934

Singlecore
Without Touchbar: 4328
Touchbar: 4216

(No, I didn't make a typo. The non-touchbar models outperform the touchbar model on Geekbench.

 

[davidcmc]

Where did you get those scores?

Edit

Nevermind, just found them in Geekbench Browser page. Indeed, the nTB model scores higher than the TB model. That's probably because Max Turbo Frequency of the nTB is higher (3.60 GHz) than the TB (3.50 GHz).

But I'm curious about the Multiscore results, since Max Turbo Frequency is "the maximum single core frequency at which the processor is capable of operating". It makes sense for Singlecore results, but I don't get how those 2.3 GHz score higher than 3.1 GHz on Multicore tests.

 

[HiRez]

According the Intel Power Gadget, with both cores loaded near 100% continuous, my 2.3 NTB spikes to 3.6 GHz, then after a few seconds of that scales back to 3.3 GHz and stays there indefinitely (I tested for over 45 minutes). I'm not quite sure how a 2.3 GHz processor can maintain a 3.3 GHz speed, but that's what Intel says.

 

[TigerMSTR]

According the Intel Power Gadget, with both cores loaded near 100% continuous, my 2.3 NTB spikes to 3.6 GHz, then after a few seconds of that scales back to 3.3 GHz and stays there indefinitely (I tested for over 45 minutes). I'm not quite sure how a 2.3 GHz processor can maintain a 3.3 GHz speed, but that's what Intel says.

Interesting. The Geekbench tests takes a little less than two minutes to complete on these notebooks. So even when considering the small amount of throttling that happens in the first few seconds, the nTB model CPU appears to be very slightly faster than the TB model. I wonder how GPU compares

 

[HiRez]

Interesting. The Geekbench tests takes a little less than two minutes to complete on these notebooks. So even when considering the small amount of throttling that happens in the first few seconds, the nTB model CPU appears to be very slightly faster than the TB model. I wonder how GPU compares

I also noticed that most of the Geekbench tests either weren't using 100% CPU or had idle time in between many short tests, so I don't think Geekbench is a great test to observe throttling. I used a large modo render, ffmpeg would also work well (generally maxes out all available cores).