2020 $1799 MBP Unusual Benchmark Performance

[jgorman]

The new $1799 13-inch MBP has some unexpected benchmarking results so far, according to the video MaxTech made comparing it to the 2020 $1299 MBP.

In the Cinebench R20 CPU benchmark, it seems like Apple is letting the $1799 MBP draw more power than the $1299 MBP, so the more expensive model is running hotter and louder than the cheaper model. The more expensive model has two cooling fans, so perhaps Apple is relying on those fans to keep the CPU working harder. The $1799 MBP has a significant advantage in single-core performance. It also beats the $1299 model in multi-core performance, but by a more modest margin.

In the Unigine Heaven graphics benchmark, the $1799 MBP was cooler and quieter than the $1299 MBP, but the $1799 MBP was not fully using its GPU. It was only loading the GPU 75-80%. Maybe this is due to a driver issue. In any event, the $1799 MBP has a significant advantage in graphics performance, and the gap is even greater when an application can use Metal graphics.

On an unrelated note, this video also showed that the 2020 $1299 MBP has better graphics performance than the 2019 $1299 MBP in Geekbench's Metal benchmark and in the BruceX Final Cut Pro X benchmark.

 

[ctjack]

In Cinebench R20 CPU benchmark, it seems like Apple is letting the $1799 MBP draw more power than the $1299 MBP, so the more expensive model is running hotter and louder than the cheaper model.

25W vs 45W. And score in multicore is just slightly bigger, but not dramatically.

The $1799 MBP has a significant advantage in single-core performance.

Not proven. He made this statement according to Geekbench and not Cinebench r20 single core test. Geekbench also uses ssd speeds, RAM speeds and size. So comparison does not tell anything about single core performance.

but the $1799 MBP was not fully using its GPU. It was only loading the GPU 75-80%. Maybe this is due to a driver issue.

Issue is similar to MBP 16. MBP 16 has only powerhouse for 100W. So either MBP 16's CPU draw 65 watt and GPU has only 35W left and not performing well, or GPU draws more than 50W and CPU is left with less power.
MBP 16 CPU + GPU wants to draw 120 watts, but there is no battery and power to fulfill it. So Apple cut it at 100W.
The same goes here. No room for CPU and GPU at the same time in the MBP13. It is one or another. Both can't perform at full potential.

 

[jgorman]

I just watched the video again. The $1799 MBP reached 100 degrees C during the single run of Cinebench R20, even with the 10th-gen design and twin cooling fans. I wonder if Apple will allow the new MBP CPUs to sustain performance at 100 degrees C, like the 2020 MacBook Airs.

 

[Spectrum]

The new $1799 13-inch MBP has some unexpected benchmarking results so far, according to the video MaxTech made comparing it to the 2020 $1299 MBP.

In Cinebench R20 CPU benchmark, it seems like Apple is letting the $1799 MBP draw more power than the $1299 MBP, so the more expensive model is running hotter and louder than the cheaper model. The more expensive model has two cooling fans, so perhaps Apple is relying on those fans keep the CPU working harder. The $1799 MBP has a significant advantage in single-core performance. It also beats the $1299 model in multi-core performance, but by a more modest margin.

In Unigine Heaven graphics benchmark, the $1799 MBP was cooler and quieter than the $1299 MBP, but the $1799 MBP was not fully using its GPU. It was only loading the GPU 75-80%. Maybe this is due to a driver issue. In any event, the $1799 MBP has a significant advantage in graphics performance, and the gap is even greater when an application can use Metal graphics.

On an unrelated note, this video also showed that the 2020 $1299 MBP has better graphics performance than the 2019 $1299 MBP in Geekbench's Metal benchmark and in the BruceX Final Cut Pro X benchmark.

The 1299 model has an 8th generation 15Watt TPD CPU. The 1799 model has a 10th gen 28 Watt CPU.
So of course the latter model draws more power...it is designed exactly to do that.

 

[jgorman]

The 1299 model has an 8th generation 15Watt TPD CPU. The 1799 model has a 10th gen 28 Watt CPU.
So of course the latter model draws more power...it is designed exactly to do that.

You are right, but what I found unusual is the $1799 model ran hotter and louder in the Cinebench R20 benchmark, which might be surprising considering it is using a more power-efficient, 10th-gen chip and it has an additional cooling fan. Of course, Apple sets the thermal limit to what they want, so perhaps they are trying to get all the performance they can out of the chip.

 

[Spectrum]

Yeah, actually what I didn't realise is just how hard (sustained) the 1.4Ghz MBPro can be pushed. Maintaining 3 GHz all core turbo and 25Watts with a single fan set up is mighty impressive (On a 15 Watt TDP CPU). This is what makes the performance of the 10th gen look relatively modest. The 28Watt CPU stabilises at around 30 watts, and the same 3Ghz all core turbo, so barely above the TDP specification.

It has made me think again about the i7 1.7Ghz model. Whilst it might be 8th gen, the performance most likely exceeds the new 10th gen.

I wonder if this new TDP headroom is also what leads to the better (apparent) GPU result they report in this video?

 

[Mattyman]

I did notice the i5 in the basic 2020 model is still outperformed by the i5 in the higher 2020 model in R20. If they are are both being push to around 30 watts and similar GHz how is this the case?

 

[jgorman]

I did notice the i5 in the basic 2020 model is still outperformed by the i5 in the higher 2020 model in R20. If they are are both being push to around 30 watts and similar GHz how is this the case?

The 10th-gen CPUs can generally perform more instructions per clock than the 8th-gen CPUs, given similar conditions.

 

[Mattyman]

The 10th-gen CPUs can generally perform more instructions per clock than the 8th-gen CPUs, given similar conditions.

Ok so the 15w and 28w 10th gen when being pushed and both equal wattage and GHz should perform the same I assume?

 

[jgorman]

Ok so the 15w and 28w 10th gen when being pushed and both equal wattage and GHz should perform the same I assume?

It is hard to speak about a 10th-gen 15W chip, because Apple does not make a laptop with one. Each laptop maker can set their own voltage, power and thermal limits in firmware. Likewise, each laptop's cooling system is different. For example, a Acer laptop's CPU might have a 75 degree C thermal limit, so how well it performs greatly depends on how well the laptop's cooling system works.

Also, the 28W chip has a higher base frequency and turbo boost frequency than any 10th-gen 15W chip. Like this video showed, the 28W MBPs generally draw 5W more power under load than 15W MBPs; this was true of the 2019 MBPs too. So, unless there is a problem with cooling, the 28W MBP would probably turbo boost higher initially, and then run at a higher frequency for any given load, which would give you better performance.

 

[Mattyman]

It is hard to speak about a 10th-gen 15W chip, because Apple does not make a laptop with one. Each laptop maker can set their own voltage, power and thermal limits in firmware. Likewise, each laptop's cooling system is different. For example, a Acer laptop's CPU might have a 75 degree C thermal limit, so how well it performs greatly depends on how well the laptop's cooling system works.

Also, the 28W chip has a higher base frequency and turbo boost frequency than any 10th-gen 15W chip. Like this video showed, the 28W MBPs generally draw 5W more power under load than 15W MBPs; this was true of the 2019 MBPs too. So, unless there is a problem with cooling, the 28W MBP would probably turbo boost higher initially, and then run at a higher frequency for any given load, which would give you better performance.

thanks

 

[jgorman]

Yeah, actually what I didn't realise is just how hard (sustained) the 1.4Ghz MBPro can be pushed. Maintaining 3 GHz all core turbo and 25Watts with a single fan set up is mighty impressive (On a 15 Watt TDP CPU). This is what makes the performance of the 10th gen look relatively modest. The 28Watt CPU stabilises at around 30 watts, and the same 3Ghz all core turbo, so barely above the TDP specification.

It has made me think again about the i7 1.7Ghz model. Whilst it might be 8th gen, the performance most likely exceeds the new 10th gen.

I wonder if this new TDP headroom is also what leads to the better (apparent) GPU result they report in this video?

I agree with you that the 8th-gen model is impressive, but I would stick with the i5. I am not sure if the i7 is worth the extra cost. The 10th-gen models are good too, since they can perform more operations for any given frequency than the 8th-gen models.

Keep in mind that it only takes a short time for one Cinebench R20 run to finish, so reviewers typically run it many times in a row to simulate sustained CPU performance. If the 2020 MBPs are like the 2019 MBPs, the 28W CPU models will sustain performance at 32-33W, while the 15W CPU models will sustain at 27-29W.

This ability to use ~5W more power might not sound like much, but in the 2019 models, it could give the 28W CPU 20% better peformance when both the CPU and GPU are working hard.

As far as the new graphics, I guess it is probably a combination of it drawing more power and being able to dissipate that power with twin cooling fans.

 

[FastLaneJB]

Ok so the 15w and 28w 10th gen when being pushed and both equal wattage and GHz should perform the same I assume?

Yes I think so as they are the same chips really but just binned differently. It's a bit misleading as the 15W chips are allowed to hit 25W and there's some Windows laptops that run them at 25W. Similar in the larger laptops, some Windows laptops can put the 8 core chips up to 62W (More in a burst, I saw the new Gigabyte Aero can pump 100W into one) when they are 45W TDP chips.

The wattage a chip runs isn't normally covered but can make a huge difference really. For instance the Dell XPS 13 which I see people mention for being so slim and sleek (It is, I had one for a while) also only runs at 15W on the CPU. So in reality it'll be quite a bit slower than the MacBook Pro 28W but just don't think most people are aware of that.

Of course the downside is battery draw... Swings and roundabouts

 

[Spectrum]

What is confusing to me is that the base MBPro is still “marketed” as a 15 watt 1.4Ghz machine, when in reality, due to the firmware (And cooling) permitting 25 watts constant, it is far more similar to a CPU with a base frequency of >2Ghz.

Stating that it is a 1.4Ghz CPU is completely meaningless.

 

[Mattyman]

Yes I think so as they are the same chips really but just binned differently. It's a bit misleading as the 15W chips are allowed to hit 25W and there's some Windows laptops that run them at 25W. Similar in the larger laptops, some Windows laptops can put the 8 core chips up to 62W (More in a burst, I saw the new Gigabyte Aero can pump 100W into one) when they are 45W TDP chips.

The wattage a chip runs isn't normally covered but can make a huge difference really. For instance the Dell XPS 13 which I see people mention for being so slim and sleek (It is, I had one for a while) also only runs at 15W on the CPU. So in reality it'll be quite a bit slower than the MacBook Pro 28W but just don't think most people are aware of that.

Of course the downside is battery draw... Swings and roundabouts

In the benchmarks the xps13 generally beats the pro. I guess we are talking sustained performance only.

 

[KPOM]

What is confusing to me is that the base MBPro is still “marketed” as a 15 watt 1.4Ghz machine, when in reality, due to the firmware (And cooling) permitting 25 watts constant, it is far more similar to a CPU with a base frequency of >2Ghz.

Stating that it is a 1.4Ghz CPU is completely meaningless.

Agreed. And the “2.0 GHz” 28W chips will idle and run at well below 2.0 GHz if the OS doesn’t request a higher speed. We really should refer to the “15W” chips as 25W chips given Apple’s implementation. It explains why the base Pro is faster than the Air in CPU-intensive applications and why it is so close to the 28W Ice Lake chips.

 

[Spectrum]

Agreed. And the “2.0 GHz” 28W chips will idle and run at well below 2.0 GHz if the OS doesn’t request a higher speed. We really should refer to the “15W” chips as 25W chips given Apple’s it explains why the base Pro is faster than the Air in CPU-intensive applications and why it is so close to the 28W Ice Lake chips.

Exactly! I just struggle to see where the benefit is to market these as 1.4 and 1.7 GHz CPUs. Is it just to get a more efficient “upsell” to the 2.0Ghz model? (Which of course is also never really at 2.0Ghz either...).
It is really confusing.

I’m looking to replace a 2014 3.0 GHz 13 inch MBPro, that can turbo up to 3.5Ghz, but obviously can also ramp down to around 1Ghz when it is idle.

But at least knowing it is a 3.0Ghz CPU gives me a good indication of the expected sustained CPU frequency.

For the 8th and 10th Gen 13”MBPros, in my opinion, these should also be listed as “2.8-3.0Ghz” CPUs since this is the apparent speed where they are stable (With boosts of up to ~4Ghz and low power states of less than ~1Ghz.)

Similarly, the MBAir would make much more sense (to me) to say this is a ~1.7Ghz CPU (with boosts of ~3.5), than to say it is a 1.2 GHz CPU.

 

[jgorman]

What is confusing to me is that the base MBPro is still “marketed” as a 15 watt 1.4Ghz machine, when in reality, due to the firmware (And cooling) permitting 25 watts constant, it is far more similar to a CPU with a base frequency of >2Ghz.

Stating that it is a 1.4Ghz CPU is completely meaningless.

No, I understand how you feel. However, in this case, Apple almost hit the bullseye with the base MBP. If the MBP CPU fell below its base clock in sustained performance, or thermal throttled, people would be very upset.

When you tax the 2019 base MBP CPU and the integrated GPU at the same time, all four cores run at 1.5 GHz using 28W at 87 degrees C. This is just above Apple's marketed 1.4 GHz. Of course, if you just use the CPU without using the GPU, performance is much better at 3.0 GHz.

Truthfully, I think Apple underpromised and overperformed here. Using Intel's definition of base frequency, 1.4 GHz is the clock speed for all cores when the processor draws 15W to do a "an Intel-defined, high-complexity workload." If you give the base MBP the most taxing CPU and GPU workload, it can stay above 1.4 GHz, and the cooling solution can take care of the heat given off by the processor drawing 28W.

 

[Spectrum]

When you tax the 2019 base MBP CPU and the integrated GPU at the same time, all four cores run at 1.5 GHz using 28W at 87 degrees C. This is just above Apple's marketed 1.4 GHz. Of course, if you just use the CPU without using the GPU, performance is much better at 3.0 GHz.

Ah...I hadn't considered the GPU component. That makes sense. Since I don't really use the GPU for anything, I tend not to think about it!
Presumably it is only games that might tax the GPU and CPU at max load? (Or maybe separate GPU-bound and Cpu-bound tasks running simultaneously?). Not that I have any GPU-bound stuff going on at all I think.

 

[jgorman]

Ah...I hadn't considered the GPU component. That makes sense. Since I don't really use the GPU for anything, I tend not to think about it!
Presumably it is only games that might tax the GPU and CPU at max load? (Or maybe separate GPU-bound and Cpu-bound tasks running simultaneously?). Not that I have any GPU-bound stuff going on at all I think.

Yes, games do that, but the CPU and GPU also work together to varying degrees in any rendering (creating an image or a frame of video). So, this includes batch photo editing, Zoom calls, streaming video and exporting a video clip in iMovie.

 

[sdedalus1]

I wish Intel and more reviewers would list the "all core turbo": sustained boost for all cores. Seems more useful for comparison than single core max boost or the super low numbers, and for particular laptops provides clues about thermal/power limits. Max sort of does this with his review, showing where the boost settles under sustained cpu load. Of course, can't compare ghz across different generations without taking account of IPC.

 

[Mattyman]

In this review the 10th MacBook Pro performs worse than the 8th gen MacBook Pro on an iMovie export. Why do you think this would be the case? It’s looks a bit odd but he did the test several times!

 

[binkie]

In this review the 10th MacBook Pro performs worse than the 8th gen MacBook Pro on an iMovie export. Why do you think this would be the case? It’s looks a bit odd but he did the test several times!

Should not be the case anymore after a few updates. Has to do with optimisation.

 

[Mattyman]

this guy suggests the new pro thermal throttles after 5 minutes. Not seen this reported yet....any thoughts if this could be true?

 

[Mattyman]

Also, is there any reason why the fans are different in the same 2020 MacBook Pro 13 inch. A bit odd....