Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Jamalogo10

macrumors member
Original poster
Jun 13, 2017
89
32
I recently purchased a customized MBP. (2017 13in nTB space grey with an i7, 512 SSD, 16GB of RAM)

It looks as though the score is higher than almost every 2017 13in TB MBP. It doesn't make sense.

https://browser.primatelabs.com/v4/cpu/3245525


I came from a mid 2012 non retina 15in quad core i7 with 4gb of ram and 512ssd and my god does this thing feel tiny in comparison. In a good way! I'm 6ft and the 15in MBP did not fit comfortably on my lap, never mind the tray table of an air plane.

If you're like me and felt worried about missing out on the quad core i7 and dGPU, don't be. This thing is still lightning fast and can handle every thing I throw at it, minus extremely intensive tasks or high level gaming. One more thing... THE BATTERY LIFE is ridiculous. I can go nearly 2 days without charging it. Best travel companion ever!

Does anyone have a protective, water proof case/sleeve they would recommend?
 
Of coursed your benchmark is similar to the top end i7 TB, they both boost to the same speed, over a short benchmark they are the same. Key difference is the i7 in the TB models have more cooling and a higher base clock speed. Meaning that they can perform at those levels for far longer before throttling, and that the throttle speed is faster on the TB than the nTB. If I start my MacBook Pro cold it has roughly the same scores, after it heats up it decreases a bit due to throttling and if I'm on battery, that is the source of the variation you see in Geekbench.

That being said, the difference isn't much unless you do work such as processing arrays with millions of floating point digits into vector graphics.
 
Of coursed your benchmark is similar to the top end i7 TB, they both boost to the same speed, over a short benchmark they are the same. Key difference is the i7 in the TB models have more cooling and a higher base clock speed. Meaning that they can perform at those levels for far longer before throttling, and that the throttle speed is faster on the TB than the nTB. If I start my MacBook Pro cold it has roughly the same scores, after it heats up it decreases a bit due to throttling and if I'm on battery, that is the source of the variation you see in Geekbench.


But wouldn't you say the i7 TB needs more cooling because of the different CPU and the addition of the touch bar? So wouldn't basic functions require that higher level of base CPU performance to fully maximize the touch bar?


That being said, the difference isn't much unless you do work such as processing arrays with millions of floating point digits into vector graphics.


The higher base clock measurement doesn't mean a whole lot because whenever the turbo boost function is needed it will run. The 28 vs 15 argument makes sense but I've also heard the nTB can sustain a fairly high turbo output around not too far below the 4ghz ceiling almost indefinitely. (I need to check my sources but I remember reading that)

I slapped my knee when I saw the new nTB mbps cpu benchmarks increase 25+% over the course of 8 months. Upon comparing my 9975 benchmark to the 2016 i7 ntb the differences are stark.

https://browser.primatelabs.com/v4/cpu/3271767
(a benchmark I stumbled across)

This increase in "measurable" performance is unheard of. Especially from Apple.
 
The higher base clock measurement doesn't mean a whole lot because whenever the turbo boost function is needed it will run.

Only when you're running a high-demand, single-thread process and only until your CPU thermal throttles - which is the point Michael was making. The TB will outperform the nTB during real-world, sustained, demanding workloads. It's likely only a minority that will experience this, but that's how it is.

The two machines will perform similarly in synthetic benchmark tests, as you've discovered.

The benchmark figures for 2017 vs. 2016 are indeed quite stark - around 10-15% higher in favour of 2017, sometimes more in the case of the nTB models.
 
Only when you're running a high-demand, single-thread process and only until your CPU thermal throttles - which is the point Michael was making. The TB will outperform the nTB during real-world, sustained, demanding workloads. It's likely only a minority that will experience this, but that's how it is.

The two machines will perform similarly in synthetic benchmark tests, as you've discovered.

The benchmark figures for 2017 vs. 2016 are indeed quite stark - around 10-15% higher in favour of 2017, sometimes more in the case of the nTB models.

I don't believe that's true. The CPU in nTB doesn't overheat. I've nTB one and using the i7 CPU at 400%, I am getting this:
Screen Shot 2017-06-30 at 11.57.18 AM.png
The lines are immediately flat and can be sustained forever. If you have big XCode project - real world example -, compile times should be similar to i7 TB or better.
 
  • Like
Reactions: ckeck
This really confusing me on the choice of which model to buy. I suspect in real world tests you wouldn't be able to tell much difference between any of them unless doing professional work over sustained periods.

Saying that, I've been seeing the 2 GHZ chips generally jump up to somewhere in the lower 3ghz spectrum when executing 'normal tasks' IE: opening an application. This could lend an idea that the nTB has to turbo a lot more then those in the TB who comfortably hang around their stock speed more often. If you are talking efficiency then possibly the TB are better than the nTB.

If you scroll down and look at the actual performance scores to really muddy the waters, look at the i7 nTB outpacing most of the i5 TBs.
 
Last edited:
This really confusing me on the choice of which model to buy. I suspect in real world tests you wouldn't be able to tell much difference between any of them unless doing professional work over sustained periods.

Saying that, I've been seeing the 2 GHZ chips generally jump up to somewhere in the lower 3ghz spectrum when executing 'normal tasks' IE: opening an application. This could lend an idea that the nTB has to turbo a lot more then those in the TB who comfortably hang around their stock speed more often. If you are talking efficiency then possibly the TB are better than the nTB.

If you scroll down and look at the actual performance scores to really muddy the waters, look at the i7 nTB outpacing most of the i5 TBs.
I would even say i7 TB might perform slightly better for CPU bound tasks ans slightly worse for GPU bound tasks for the benchmark we are seeing.

Both TB and nTB would be idle at the same clock speed. ~1.5-2.0ghz. Clock speed on the box doesn't seem to mean anything anymore.

What's really upsetting is the i7 from the 12" MacBook is not very far from there for CPU bound tasks. That means they could have build a i7 12" with a bigger screen without fans and with a smaller size than the 13". Without loosing much in term of performance in comparison with the 13" TB or nTB. Oh, well.
 
Hmm i still feel day to day benchmarks mean that the choice still comes down to : Do you want the more ports and touchbar, maybe slightly better sound.

I suspect most professionals will be looking at sustained performance: IE knocking time of a render. Day to day it probably isn't a blind bit of difference.

Interestingly I did do a rather simple test in a store between the two base models and opened all the same apps at the same time. I didn't notice a difference but i did notice CPU usage slightly higher in the nTB version. Reason unknown.
 
Last edited:
I slapped my knee when I saw the new nTB mbps cpu benchmarks increase 25+% over the course of 8 months. Upon comparing my 9975 benchmark to the 2016 i7 ntb the differences are stark.

https://browser.primatelabs.com/v4/cpu/3271767
(a benchmark I stumbled across)

This increase in "measurable" performance is unheard of. Especially from Apple.
I don't think that's a good representative test for that machine. Scores for the 2016 13-inch 2.4 GHz i7 seem to average around 4100/8200. Still a nice speed boost going to the 2017 (even the 2.3 GHz i5 Kaby Lake appears to easily outpace the 2.4 GHz i7 Sky Lake from 2016), but not as dramatic.

I wish Geekbench would put some longer-running tests into the portfolio, and output throttling data and temps.
 
  • Like
Reactions: Jamalogo10
I don't think that's a good representative test for that machine. Scores for the 2016 13-inch 2.4 GHz i7 seem to average around 4100/8200. Still a nice speed boost going to the 2017 (even the 2.3 GHz i5 Kaby Lake appears to easily outpace the 2.4 GHz i7 Sky Lake from 2016), but not as dramatic.

I wish Geekbench would put some longer-running tests into the portfolio, and output throttling data and temps.

Don't forget speedstep technology 2.0 is in the 2017 chips which might be making a difference.
 
I don't believe that's true. The CPU in nTB doesn't overheat. I've nTB one and using the i7 CPU at 400%, I am getting this:
View attachment 706603
The lines are immediately flat and can be sustained forever. If you have big XCode project - real world example -, compile times should be similar to i7 TB or better.

How noisy is the single fan under full load like that? I read on a few reviews that the dual fans of the TB models was a lot nicer sounding. Sustaining boost for a long time is only a good user experience if i'm not listening to 6k RPM jet engine like fans.
 
That is not impressive. My 12.9 iPad Pro gets the same score lol.

Even my 4 year old 15" Retina Macbook Pro is faster xD.
 
That is not impressive. My 12.9 iPad Pro gets the same score lol.

Even my 4 year old 15" Retina Macbook Pro is faster xD.

I see alot of scores being compared between iOS and macOS hardware, and I think it is an Apples to Oranges comparison. Since the architecture is different and the applications are compiled differently, I believe the numbers don't paint a true picture of the performance. Comparing an Intel to an ARM processor would be more appropriate if they could be used with the same software without an emulator or the software being recompiled. They have different input methods as well, and don't run at 5K or 4K like a macOS system is able to. Peripherals are also a factor which the iOS devices run a lot slimmer on than a desktop or a laptop.
 
I see alot of scores being compared between iOS and macOS hardware, and I think it is an Apples to Oranges comparison. Since the architecture is different and the applications are compiled differently, I believe the numbers don't paint a true picture of the performance. Comparing an Intel to an ARM processor would be more appropriate if they could be used with the same software without an emulator or the software being recompiled. They have different input methods as well, and don't run at 5K or 4K like a macOS system is able to. Peripherals are also a factor which the iOS devices run a lot slimmer on than a desktop or a laptop.

I think we will see that soon. If the benchmarks are true, then the 12.9 iPad Pro has a more powerful GPU than the integrated GPU in the 13" MBP so it has already beaten Intel already at that area. So I suppose the next generation of Apple's ARM chips make the Intel chips obsolete for the low to mid-tier chips. And at that point it make sense for Apple to switch to ARM for the 13" rMBP line. Then you can compare ARM chips vs Intel on Mac OS X.
 
Last edited:
I think we will see that soon. If the benchmarks are true, then the 12.9 iPad Pro has a more powerful GPU than the integrated GPU in the 13" MBP so it has already beaten Intel already at that area. So I suppose the next generation of Apple's ARM chips make the Intel chips obsolete for the low to mid-tier chips. And at that point it make sense for Apple to switch to ARM for the 13" rMBP line. Then you can compare ARM chips vs Intel on Mac OS X.

None of us know if the GPU is more powerful, since they don't run the same software. You may say that a train have the ability to achieve higher speeds than cars, but they are limited to certain routes, and so the comparison falls short because neither the car or the trains maximum speed say anything about how your favoured means of transportation is able to reach their destination.
 
None of us know if the GPU is more powerful, since they don't run the same software. You may say that a train have the ability to achieve higher speeds than cars, but they are limited to certain routes, and so the comparison falls short because neither the car or the trains maximum speed say anything about how your favoured means of transportation is able to reach their destination.

There are plenty of benchmarks where the GPU of the iPad Pro scores higher than the 13" rMBP, so it is more powerful.

Also tablets are no different from laptops. An iPad Air can run OS X if you hack it as Max 10.9 supports ARM chips. But Apple won't allow it officially. There are also tablets that run full desktop OS like Windows 10, basically running every piece of software that a desktop PC can.
 
Last edited:
There are plenty of benchmarks where the GPU of the iPad Pro scores higher than the 13" rMBP, so it is more powerful.

Also tablets are no different from laptops. An iPad Air can run OS X if you hack it as Max 10.9 supports ARM chips. But Apple won't allow it officially. There are also tablets that run full desktop OS like Windows 10, basically running every piece of software that a desktop PC can.

If tablets are not different, why would I own an iPad, a Windows 10 desktop and a MacBook Pro? They run different programs, they have different input methods, some require more space to operate and they output different results. It's not about benchmarks, it is about the user experience and how fast you can get work done or run a game.

My iPad doesn't have the same version of Photoshop as my MacBook Pro, my MacBook Pro won't run Overwatch as smooth as my 1080 GTX Ti Win10 box will and my Windows 10 box won't run macOS without Clover filled with hacks or kexts.

To me it is not about potential, it is about being able to use your computer, not relying on theory, hacks or promises.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.