So my 16" has arrived this morning and since I was able to spend a few hours running some basic CPU and GPU tests, and I though I'd share my initial impression with you guys. Please keep in mind that these tests are a) very basic and b) were done in macOS, so it's possible that the scores don't translate 100% to what Windows PCs are scoring, but still, it gives us a basic idea. I will be focusing at the following areas: initial impressions/usability, CPU performance, thermals and GPU performance. As for my testing methodology: I have installed the Supplemental Update and I am using an otherwise blank laptop with minimal software installed and a new user. So nothing is indexing etc.
Model being tested: 16" MacBook Pro with i9-9880H (the 2.3Ghz CPU), 16GB RAM, Radeon Pro 5500M with 8GB VRAM
Quick summary: the sustained CPU TDP is 60Watts, short-term TDP is around 80 watts (up to 10 seconds), single-core max turbo can be maintained pretty much indefinitely, the CPU performance is better than a Dell XPS with 2.4Ghz i9-9980HK and the GPU is a big upgrade of the Vega Pro 20.
Initial impressions: it is almost the same size as the 15", just a tiny bit wider and longer. The thickness is virtually the same. It can fit into a really small backpack designed for 13" laptops (even though it's a tight fit). It feels much heavier than the 15" if you hold it in your hand despite the difference only being 170g or so, but I couldn't notice any difference when carrying them in a bag. The display is larger and very nice. I don't think it is worth the upgrade if you find the real estate of the 15" limiting, but it's a nice thing to have.
The keyboard is ok: so much has been said about keyboards that I think it deserves a separate point. Obviously I need to spend a few days typing on it to get a solid feel of the keyboard, but at the first glance, well, it's a perfectly ok keyboard. If feels like a hybrid between butterfly and magic keyboard, with slightly less stable keys than butterfly but noticeably more key travel. I couldn't compare it to the pre-2016 MBP keyboard yet, will do so next week, but it certainly feels better than the 13" 2014 model I had to briefly use last week. So yeah, I guess the keyboard is ok. Butterfly is a more precise keyboard for me and I think it's easier on the fingers (since you need to use less pressure), but let's see how I feel about it in a couple of days. Having the the escape key back is nice. Overall, I like the new Touch Bar design more, it just looks neater. A funny detail: since the keys now rise much higher from the body, it is much more difficult to tap the Touch Bar on accident. So I guess that solves this.
The CPU: finally let us get to the interesting bits — performance and thermal management! To look at these things, I used Cinebench (both R15 and R20) while monitoring the output from Intel Power Gadget , nothing too fancy.
In multi-core Cinebench R20, the MBP with i9-9880H scores steady 3280 points. I did six consecutive runs and the result was always the same ±2 point. It is actually impressive how consistent these results are. Similarly, the R15 multi-core results are steady 1400 points. The CPU power draw is levelling out at stable 60W (15W higher than all 15" MBP I've seen) and the clocks stabilise around 3.1Ghz (with promised nominal clock of 2.3Ghz, which is not too shabby). I'm sure that there will be people who claim that it "throttles" since it is not running on maximal boost, but people are weird.
To put these results in perspective, this graph might be useful. The nice thing about the graph is that is shows where the scores stabilises (and thus where the sustained performance is). Most laptops start out fast but then drop down after second or third run. And some of the scores are just plain weird. For example, an Asus Stix Scar III is reported to have the R15 multi-core score of 1400, but when we actually look at the graph is clear that is stabilises somewhere around 1260. Similarly, the Dell with5.0Ghz 8-core CPU (see the graph) has the average score of 1300, but the graph looks much sadder. Go figure.
So how does the i9-9880H in the 16" MBP perform? With consistent 1400 points it actually does very well. It seems to outperform any laptop of its size and weight, save for MSI Creator P65 (with the same CPU) which performs similarly. And it can even outperform a 3.3 kg HP Omen 17. Of course, it's not the fastest laptop out there. An MSI GE75 Raider 9SG is still 15% faster, but that is a different beast (I mean, look at this heatsink)!
Single-core performance is less interesting I think. Suffice to say that I got steady 436 points in Cinebench R2 and 180 points in R15, which puts it at the same level as any other laptop running the same CPU. It runs at steady 4.5 Ghz over multiple runs, which is about the maximum you can expect of this CPU in real-word conditions (Intel would like to top believe that this is a 4.8Ghz CPU but those top 200Mhz are a scam, since they only trigger if the CPU temperature is under 50C. Really Intel?).
By the way the 15" MBP with the 2.4GHz CPU apparently scores 3160 points at it's highest, so we have a clear improvement here as well.
Ah, I know that you people will pester me with Geekbench, so here you go: 1092/6696, which are just in the middle of the aggregate scores for comparable machines. I personally don't think Geekbench is a good benchmark, it just mixes too many things and doesn't really let you understand what it is you are measuring. Using Cinebench is also not the best, but at least I am confident that I know what is happening.
The GPU: I didn't have much time, so this will be brief. In a OpenCL based maximal throughput benchmark, we have a 30% improvement in raw compute performance and 20% improvement in the memory bandwidth over the Vega Pro 20 on my 2018 machine. For a game test, I used Total War: Three Kingdoms, which is probably one of the most demanding games with a native Mac version. I got 30 fps in all high preset running at 1920x1200 — 10 fps more then Vega Pro 20 and without any lag spikes that Vega Pro 20 suffers from when the game is played with large army sizes. This is the same as a desktop 1050 GTX Ti, which does not sound very impressive , but you need to keep in mind that this is running under macOS. I am sure this thing will outperform a GTX 1650 easily with gaming drivers on windows. I have written more about this in this post
To sum it up: make no mistake, the MacBook Pro has always been — and still is — an ultrabook. It is still as small and light as these higher-performance laptops go. But it looks to me like the 16" is indeed shaping up to be the pro ultrabook king — you will simply not find any other laptop that has this performance AND this form factor AND this battery life at the same time. While it is not dramatically different from the 15" model, the improvements are there — and most importantly, it's price/performance ratio has increased dramatically. I'd argue that it is probably a better buy than any other comparable workstation out there, talking strictly price. For example, a Dell Precision 5540 configured similarly to the higher-end MBP is $300 more.
Model being tested: 16" MacBook Pro with i9-9880H (the 2.3Ghz CPU), 16GB RAM, Radeon Pro 5500M with 8GB VRAM
Quick summary: the sustained CPU TDP is 60Watts, short-term TDP is around 80 watts (up to 10 seconds), single-core max turbo can be maintained pretty much indefinitely, the CPU performance is better than a Dell XPS with 2.4Ghz i9-9980HK and the GPU is a big upgrade of the Vega Pro 20.
Initial impressions: it is almost the same size as the 15", just a tiny bit wider and longer. The thickness is virtually the same. It can fit into a really small backpack designed for 13" laptops (even though it's a tight fit). It feels much heavier than the 15" if you hold it in your hand despite the difference only being 170g or so, but I couldn't notice any difference when carrying them in a bag. The display is larger and very nice. I don't think it is worth the upgrade if you find the real estate of the 15" limiting, but it's a nice thing to have.
The keyboard is ok: so much has been said about keyboards that I think it deserves a separate point. Obviously I need to spend a few days typing on it to get a solid feel of the keyboard, but at the first glance, well, it's a perfectly ok keyboard. If feels like a hybrid between butterfly and magic keyboard, with slightly less stable keys than butterfly but noticeably more key travel. I couldn't compare it to the pre-2016 MBP keyboard yet, will do so next week, but it certainly feels better than the 13" 2014 model I had to briefly use last week. So yeah, I guess the keyboard is ok. Butterfly is a more precise keyboard for me and I think it's easier on the fingers (since you need to use less pressure), but let's see how I feel about it in a couple of days. Having the the escape key back is nice. Overall, I like the new Touch Bar design more, it just looks neater. A funny detail: since the keys now rise much higher from the body, it is much more difficult to tap the Touch Bar on accident. So I guess that solves this.
The CPU: finally let us get to the interesting bits — performance and thermal management! To look at these things, I used Cinebench (both R15 and R20) while monitoring the output from Intel Power Gadget , nothing too fancy.
In multi-core Cinebench R20, the MBP with i9-9880H scores steady 3280 points. I did six consecutive runs and the result was always the same ±2 point. It is actually impressive how consistent these results are. Similarly, the R15 multi-core results are steady 1400 points. The CPU power draw is levelling out at stable 60W (15W higher than all 15" MBP I've seen) and the clocks stabilise around 3.1Ghz (with promised nominal clock of 2.3Ghz, which is not too shabby). I'm sure that there will be people who claim that it "throttles" since it is not running on maximal boost, but people are weird.
To put these results in perspective, this graph might be useful. The nice thing about the graph is that is shows where the scores stabilises (and thus where the sustained performance is). Most laptops start out fast but then drop down after second or third run. And some of the scores are just plain weird. For example, an Asus Stix Scar III is reported to have the R15 multi-core score of 1400, but when we actually look at the graph is clear that is stabilises somewhere around 1260. Similarly, the Dell with5.0Ghz 8-core CPU (see the graph) has the average score of 1300, but the graph looks much sadder. Go figure.
So how does the i9-9880H in the 16" MBP perform? With consistent 1400 points it actually does very well. It seems to outperform any laptop of its size and weight, save for MSI Creator P65 (with the same CPU) which performs similarly. And it can even outperform a 3.3 kg HP Omen 17. Of course, it's not the fastest laptop out there. An MSI GE75 Raider 9SG is still 15% faster, but that is a different beast (I mean, look at this heatsink)!
Single-core performance is less interesting I think. Suffice to say that I got steady 436 points in Cinebench R2 and 180 points in R15, which puts it at the same level as any other laptop running the same CPU. It runs at steady 4.5 Ghz over multiple runs, which is about the maximum you can expect of this CPU in real-word conditions (Intel would like to top believe that this is a 4.8Ghz CPU but those top 200Mhz are a scam, since they only trigger if the CPU temperature is under 50C. Really Intel?).
By the way the 15" MBP with the 2.4GHz CPU apparently scores 3160 points at it's highest, so we have a clear improvement here as well.
Ah, I know that you people will pester me with Geekbench, so here you go: 1092/6696, which are just in the middle of the aggregate scores for comparable machines. I personally don't think Geekbench is a good benchmark, it just mixes too many things and doesn't really let you understand what it is you are measuring. Using Cinebench is also not the best, but at least I am confident that I know what is happening.
The GPU: I didn't have much time, so this will be brief. In a OpenCL based maximal throughput benchmark, we have a 30% improvement in raw compute performance and 20% improvement in the memory bandwidth over the Vega Pro 20 on my 2018 machine. For a game test, I used Total War: Three Kingdoms, which is probably one of the most demanding games with a native Mac version. I got 30 fps in all high preset running at 1920x1200 — 10 fps more then Vega Pro 20 and without any lag spikes that Vega Pro 20 suffers from when the game is played with large army sizes. This is the same as a desktop 1050 GTX Ti, which does not sound very impressive , but you need to keep in mind that this is running under macOS. I am sure this thing will outperform a GTX 1650 easily with gaming drivers on windows. I have written more about this in this post
To sum it up: make no mistake, the MacBook Pro has always been — and still is — an ultrabook. It is still as small and light as these higher-performance laptops go. But it looks to me like the 16" is indeed shaping up to be the pro ultrabook king — you will simply not find any other laptop that has this performance AND this form factor AND this battery life at the same time. While it is not dramatically different from the 15" model, the improvements are there — and most importantly, it's price/performance ratio has increased dramatically. I'd argue that it is probably a better buy than any other comparable workstation out there, talking strictly price. For example, a Dell Precision 5540 configured similarly to the higher-end MBP is $300 more.
