Today we're taking our benchmarking to places it's never been before with a detailed look at an Apple Silicon product. A few weeks ago we bought a new Apple MacBook Pro 16 with the M1 Pro SoC and have since put it through its paces.
So this test is about comparing the new M1 Pro and comparing it to the best x86-based CPUs from AMD and Intel that we have today. And when we compare correctly, we don't mean running Geekbench and putting an end to it. In fact, we're not going to run Geekbench at all … we mean run a wide range of real-world multi-platform applications and compare the performance of the best laptop chips available today, all of which run under fair and equitable conditions.
That'll be interesting because it's fair to say there's a lot of hype surrounding the Apple M1 silicon and its supposed desktop-damaging performance. With hype there is often overhype and certainly a lot of fanboyism, including using Apple's internal benchmarks as gospel, so we'll see how justified all of this is today.
The Apple M1 Pro is a very different SoC design than the chips we usually test because it uses the ARM family of architecture instead of x86. That doesn't mean it's better, it just works differently. The heart of this processor is a maximum of 10 CPU cores, divided into 8 high-performance Firestorm cores and 2 efficient Icestorm cores in hybrid design. The Firestorm cores are clocked at up to 3.2 GHz and the Icestorm cores reach 2.1 GHz, although this depends on the load on the system and with all-core workloads these clocks drop by a few hundred MHz.
On the GPU side, we've got a 16-core design in the M1 Pro that goes up to 32 cores in the M1 Max (which is offered in a more expensive version of the same laptop). Connected to the CPU and GPU in a uniform design is a 256-bit LPDDR5 memory bus with a little over 200 GB / s. You can think of it as if an x86 design includes quad-channel memory instead of dual-channel, it is certainly a lot of memory bandwidth and a lot more than similar products from Intel or AMD. There's also plenty of L2 cache and 24MB top-level cache on the M1 Pro. Both the top-level cache and memory controllers are doubled in the M1 Max design to power the larger GPU.
Apple has taken the Nvidia approach to naming the M1 Pro by launching two configurations under the same name. The top-level design includes the full 10-core CPU and 16-core GPU, but the base design reduces this to just 8 CPU cores and 14 GPU cores. While this is prominently advertised on their product page for the MacBook Pros, Apple doesn't go out of the way to differentiate the actual product name beyond the M1 Pro.
For our tests, we bought a 16-inch MacBook Pro with the full M1 Pro configuration. We also opted for 16GB of unified memory to match our existing laptop test data that uses 16GB of RAM and a 512GB SSD. Apple's RAM and storage upgrades are disgustingly expensive; it would have cost an additional $ 600 to increase that to 32GB of memory and 1TB SSD. You also can't upgrade these components yourself, so you have no choice but to pay these prices when you need higher capacity, not that you would expect otherwise from Apple.
For testing purposes, today we're comparing the M1 Pro with our existing data for AMD and Intel CPUs, all of which are normalized to the standard TDPs of these chips. There are, of course, some additional challenges that we worked on over the past week. Not all of the applications we compare are available on macOS. Therefore, we only include cross-platform benchmarks with a macOS version, be it a native ARM version or an x86 version that is emulated via Rosetta 2 on Apple Silicon.
This immediately introduces a variable where Windows and macOS versions of apps work a little differently, but it's the best we can do in the circumstances. Let's get to testing.
In Cinebench R23 multithreading, a native Apple Silicon app, the M1 Pro is slightly faster than the second best CPU we tested, the Ryzen 9 5900HX. Over a period of 10 minutes, the M1 Pro was 4 percent faster, but also used less power, as we'll talk about later.
The M1 Pro is more like a 35W class CPU compared to the 45W of the 5900HX; the M1 Pro is 12 percent faster than AMD's best 35 W CPU in the Ryzen 9 5980HS, which is probably more comparable. Apple has an even bigger lead on Intel, where the M1 Pro is 14% faster than Intel's best silicon with 45W, the Core i9-11980HK.
The Intel Core i7-10875H is also noteworthy. This CPU is the same as the 9th Generation Core i9-9980HK in terms of multi-thread performance, which was available in the best MacBook Pro configurations of 2019. The M1 Pro is almost 60 percent faster in this benchmark, which is why Apple users are happy about the performance boost of this new generation with Apple Silicon.
When it comes to single-thread performance, not much is different between the M1 Pro and the best CPUs from Intel or AMD. Apple's chip sits exactly between the Ryzen 9 5980HS and the Core i9-11980HK, with these CPUs in this workload that's all a bit much. However, the performance is more than 20% faster than the previous Intel CPUs that Apple used.
Handbrake is available as a native Apple Silicon app on macOS and shows us CPU-based video coding with x265. While this app runs natively on the M1 Pro, it benefits significantly from AVX instructions on x86-based CPUs. Hence, the M1 Pro isn't that impressive when compared to its x86 equivalents like the Ryzen 9 5980HS or the Core i9-11980HK. While its performance isn't bad, the M1 Pro is 9 percent behind Intel's best Core i9 and 24 percent behind AMD's best Ryzen 9 processor, even though Apple's power consumption is lower.
The M1 Pro is a powerful CPU for CPU-based blender rendering that outperforms the best chips from Intel and AMD. The performance advantage over the Ryzen 9 5900HX is only 5%, but 23% faster than the 11980HK. The main problem with MacBooks in Blender is that while the app is natively available for Apple Silicon, it doesn't support GPU rendering. When using an RTX 3050 with Optix, the GPU rendering absolutely wipes out the M1 Pro with over 3x faster performance. Until Blender is updated with GPU support, MacBook Pros aren't the best option.
While the M1 Pro has so far struggled with the best x86 CPUs in rendering and encoding tests, Apple Silicon has a massive advantage in code compilation. I should note here that for our Chromium compilation, we compared compiling the Windows version on Windows and the Mac version on macOS to the Google recommended process for each operating system. Unfortunately, I tried to fail to get the MacBook Pro to compile the Windows version, even though it can be done on Intel Macs. So this benchmark wasn't as apples to apples as I'd like it to be, but it's still a real world use case in my opinion.
In any case, the M1 Pro destroys its x86 competition and delivers 68% faster compile times than the Ryzen 9 5900HX, the next fastest product we tested. The M1 Pro has an enormous advantage with this workload, especially with its high memory bandwidth, which is far more than twice as high as other laptops in these charts. I suspect these new MacBooks will be great for programmers.
Matlab is only available through Rosetta 2 emulation for now, and the M1 Pro's performance suffers as a result. The SoC ends up being equivalent to a Ryzen 7 4800H CPU or an Intel Core i7-1165G7, which are decent chips, just not the fastest on the market today. The M1 Pro is 30 percent slower in this app than Intel's Core i9-11980HK, so it would be better to use a modern Windows computer for your engineering work in Matlab for the time being.
Microsoft Excel is available natively for Apple Silicon and our number processing benchmark works perfectly. The M1 Pro ends up being a bit faster than the best 35W CPU we've tested, the Ryzen 9 5980HS, but it can't compete with the best x86 that a mobile form factor has to offer. The Core i9-11980HK is 23% faster, the Ryzen 9 5900HX is just ahead.
The M1 Pro is a beast for 7-zip compression, using the native Apple Silicon version against the same benchmark on Windows. The M1 Pro is 11% faster than the i9-11980HK, although it uses significantly less power on this burst workload, and a whopping 27% faster than the 5900HX, which is AMD's best performance.
However, the tables rotate when 7-Zip decompression is used. This benchmark loves simultaneous multi-threading on x86 CPUs and is particularly fast on Zen, which leads to a sovereign advantage over the M1 Pro. This is a poor result for Apple Silicon as it is 32% slower than the Ryzen 9 5900HX and more in the range of Intel 10th generation processors.
The PDF export is a single-threaded application and Acrobat Pro is only available through the Rosetta 2 emulation on macOS. But don't worry, because the M1 Pro still easily beats its x86 competition with a performance in the range of 25% faster than the best that AMD or Intel have to offer.
FL Studio is a new addition to our benchmarking and something we've been researching for a while. While there is a native Apple version of the app, it unfortunately runs terribly on the M1 Pro. I quadrupled this because I couldn't quite believe it compared to our Windows-based systems, but it's representative of the latest version of the app and exports the default track. Even though the MacBook Pro reached 100% CPU usage during the track export, in my opinion something is broken with this app and requires serious work on macOS.
For Adobe Premiere, we've updated our benchmarks to use the latest 2022 version that runs natively on Apple Silicon. With the latest version of PugetBench, the overall result for the M1 Pro looks very impressive, outperforming very powerful Windows-based machines like one with a 11800H and RTX 3070 configuration. This is an excellent result without a doubt, but let's dig deeper into the partial values.
In terms of export, Premiere's result is similar to what we saw earlier when we benchmarked Handbrake, as the M1 Pro isn't the most impressive product. PugetBench uses a mix of exports, and while the SoC supports a number of hardware accelerated encoding, Nvidia GPUs are extremely good at accelerated encoding and therefore do quite well in this benchmark as well.
Where Apple is dominant is live playback, which refers to how smoothly the playback of footage is in the timeline. The M1 Pro with features like hardware-accelerated ProRes ends up being quite a bit faster than competing options, with Intel-based 11th generation machines.
The effect score is also pretty good on the M1 Pro and is between the 5900HX and 5800H in this mainly CPU-limited score. Then we come to the GPU score, where the performance is, as expected, below that of the RTX 3050. I don't have any really low-power GPUs in this benchmark, and the M1 Pro's 16-core GPU isn't exactly the equivalent of 80W + Nvidia GPUs; it even ends up 23% behind the RTX 3050, but uses a lot less power.
Premiere has a native Apple Silicon version, but After Effects does not and requires the use of Rosetta. The M1 Pro still gets a decent score in PugetBench, outperforming a Ryzen 7 4800H configuration with RTX 2060 graphics, but lags behind the top Intel-based computers.
Although this benchmark uses the GPU, the final score depends heavily on the CPU performance, with the 11800H doing particularly well thanks to its excellent single-thread performance. As for Photoshop, unfortunately we couldn't compare this app as the PugetBench tool we were using is not yet compatible with the native Apple Silicon version of Photoshop and I thought it was unfair to use the Intel version of Photoshop instead.
A new addition to our computing benchmarking is the gigapixel AI image upscaling with the popular tool from Topaz Labs. We run this benchmark with GPU acceleration in the fastest possible mode, which usually runs very well on Nvidia RTX 30 series GPUs. The M1 Pro is required to use Rosetta 2 emulation, but still offers excellent performance for a low-power configuration that is in between the results of the RTX 3060 and RTX 3050 Ti. The best Windows machine we tested is 26% faster, but this system is closer to the M1 Max.
Our latest compute benchmark is another new addition, Agisoft Metashape, which performs photogrammetry to convert still images into 3D models. This is a GPU accelerated benchmark that goes through the entire multi-step process to build a model from sample images. We are using the latest beta version of the app with native Apple Silicon support.
In this benchmark, the M1 Pro took the last of the configurations we tested, but that's not to be sniffed at, as the SoC only lags slightly behind significantly performance-hungry configurations with larger GPUs. For example, the overall M1 Pro package is only 4% slower than a system with Nvidia's RTX 3050, which runs at up to 80-95 W, although the entire M1 Pro only achieves a peak performance of around 60 W in strong CPU + GPU applications .
Next, we have gaming performance in macOS native games, and frankly, the macOS gaming ecosystem is pretty grim. Of the 23 games we compare on laptops, only 5 have macOS versions, 4 of which worked on the M1 Pro and none of them have native Apple Silicon implementations. This is a pretty similar breakdown of my entire game library, only 16% of the games of which have macOS versions. That alone means you probably shouldn't buy a MacBook to play with, but we'll test the supported games anyway.
We specifically compare Shadow of the Tomb Raider to the last section of the in-game benchmark, not to the entire run. In this segment, the M1 Pro landed between the GTX 1650 Ti and the GTX 1060 6GB in terms of performance, which is probably roughly to be expected given its low power consumption. This Apple Silicon configuration is 22% slower than the RTX 3050 with full performance from Nvidia and thus keeps it firmly in the entry-level class of the GPUs we tested.
I should note here that the game benchmarking tools available for macOS are extremely crude compared to what is currently possible on Windows, which has access to powerful software like OCAT. So while these results are representative of in-game performance, the accuracy is not what I would like it to be, but we just don't have the resources to develop our own macOS FPS counting tools.
The M1 Pro can run Borderlands 3 but it runs terribly and I don't know if that's because the game needs to be emulated through Rosetta 2 or just the GPU in that chip is weak. The performance is well below that of the GTX 1650 Ti and, to be honest, the game is practically unplayable with the ultra setting combination we tested.
Metro Exodus is the best example of game performance on the M1 Pro, although I'm not sure how equivalent the game settings are as the macOS version lacks tessellation and hairwork options. But if we assume for a moment that the game is the same, then the M1 Pro does very well, sliding between the RTX 3050 and RTX 3050 Ti at 1080p. Unfortunately, Apple Silicon doesn't support hardware-accelerated ray tracing or DLSS, so the Feature support in this game is worse than what you would get on a laptop equipped with RTX 3050.
Finally, we have Civilization VI that we are testing at low settings primarily for integrated graphics comparisons. With more powerful GPUs, it tells us about the limited CPU performance. This is another game that runs poorly on the M1 Pro and can't even compete with the GTX 1650 Max-Q we've seen in previous thin and light laptops. This game doesn't seem to be optimized for Apple Silicon at all and it could be stifled by having to run it emulated.
The last performance range to be assessed is the power consumption, measured on the wall, but without power consumption when idle. This gives us a rough idea of how much power the various system components will consume under heavy use, while taking into account differences in displays, etc.
First we have Cinebench R23 multithreading, which underlines the great efficiency of the M1 Pro. Although the Ryzen 9 5900HX was slightly outperformed, power consumption was 13W lower during the sustained long-term portion of the benchmark, more in the 35W class of CPUs.
What's even more impressive is that many x86 CPUs burst pretty high in order to achieve strong multi-thread performance for a short time, for example in apps like 7-Zip or Excel. This is not the case with the M1 Pro, which performs competitively in these apps even though it only consumes 43W of normalized peak power. The fact that Apple can achieve the performance they can provide without real boost or turbo clock is very eye-opening and indicates a design geared towards efficiency.
Apple sees the greatest efficiency advantage over its competitors in Cinebench single threading. The performance here is roughly on par with the best laptop CPUs from AMD and Intel, but the power consumption is an order of magnitude lower. Swallowing just 7W of power is pretty insane when the latest x86 CPUs get over 20W of power when they go above 4.0GHz.
It is for this reason that the new MacBooks offer such great battery life that they are a dominant advantage over their competitors in terms of efficiency when using light-threaded workloads and enable the best performance without sacrificing battery life.
The M1 Pro is also exceptionally efficient when it comes to gaming. Peak power consumption during an intense Shadow of the Tomb Raider area tickled 60W, which was less than half of our RTX 3050 test system and only slightly higher than a Ryzen 9 5900HX with its integrated GPU. The performance was also lower than that of the RTX 3050 in this test, 24% behind the RTX 3050, but the power consumption was 55% lower overall, so the performance per watt with this processing load is clearly in Apple's favor. For this reason, Apple has developed a monolithic SoC with unified memory that offers many advantages from a performance point of view.
Testing the new 16-inch MacBook Pro over the past two weeks has been very interesting and a fun challenge to learn more about macOS and its various apps. Even more interesting is the performance of the new M1 Pro in real world applications, so let's break it down into what may be a long conclusion.
What we learned
From a purely performance point of view, the Apple M1 Pro is an impressive SoC. Performance varies a bit from application to application, but in general the CPU is very competitive with today's best x86 laptop CPUs from Intel and AMD. In the best of cases where an app can really benefit from the specific features of the M1 Pro, such as: B. its extreme memory bandwidth, the M1 Pro can destroy everything that is available on Windows-based devices. At other times, the performance is roughly on par with CPUs like the Ryzen 9 5900HX and the Core i9-11980HK and can fall behind in the worst scenarios like CPU-based video coding. On the whole, however, you are definitely getting modern excellence.
Apple can extend its lead further if you have specially tailored workloads that take advantage of the M1 Pro's accelerators. Video editing with ProRes acceleration is an example, although the export performance is not quite as good as that of the best x86 systems equipped with a discrete Nvidia GPU. If you have these types of workloads, and I can imagine a lot of them will be Apple first-party apps like Final Cut or Logic, then you will see great benefits from using a MacBook Pro with these new M1 chips .
GPU performance is mixed though, I should note that we only tested the M1 Pro version with its 16-core GPU, not the full M1 Max which we expect to see significantly better performance. Basically, the GPU of the M1 Pro corresponds to a typical discrete Nvidia GPU with lower performance in productivity apps (i.e. slower than a RTX 3050 with full performance) and is rather mediocre when gaming due to a multitude of problems, including a lack of optimization and emulation.
Performance isn't the star of the show, it's efficiency that puts the M1 Pro in a class of its own.
The performance is good, but not the star of the show. It's efficiency that puts the M1 Pro in a class of its own. In heavy multi-threaded CPU apps, safe performance can be similar to a Ryzen 9 5900HX, but power consumption is at least 10W lower, making 35W class CPUs the most comparable option. Apple simply has the best performance at the same level of performance as today's best x86 designs, a head start that grows in low-threaded and GPU-heavy apps. The entire design is geared towards efficiency and the results are excellent for battery life.
This is of course expected because Apple is a complete process node in front of the x86 chips: They use 5 nm from TSMC, compared to 7 nm for AMD and Intel's in-house 10 nm SuperFin for their chips. With that kind of advantage, Apple should come out on top, and they are. This gives the M1 Pro two major advantages: The performance does not drop as much in battery mode, as the chips are so efficient that they can run at full power without plugging in. Second, the new 16-inch MacBook Pro is nearly silent during most workloads, while many Windows laptops turn up their fans at the Jet Engine level, especially in apps that use a separate GPU.
… the new 16-inch MacBook Pro is almost noiseless for most workloads, while many Windows laptops run their fans to the Jet Engine level, especially in apps that use a separate GPU.
While the M1 Pro destroys its competitors when it comes to free roaming away from a charger, it's a more complicated situation when it comes to getting the best performance from a laptop. Some of the M1 Pro's margins are so small that if you put an x86 laptop in "high performance, high power" mode, the script reverts to AMD or Intel.
For example, while the M1 Pro is 4% faster than the 5900HX in Cinebench when the 5900HX is configured at 45W, AMD can regain the lead by throwing it out of the water at 70W or more and can be up to 10% faster be.
This is important when you are less concerned about battery life and more concerned with getting your tasks and workloads completed as quickly as possible. Windows systems can still have the edge there, especially if you get one with a halfway decent discrete GPU, as the M1 Pro's GPU isn't amazing and the M1 Max is ridiculously expensive.
We're also just not in a position where the M1 Pro in a MacBook beats or comes close to a desktop workstation in most apps. I know some people claim this, but when CPUs like the Ryzen 9 5950X and Core i9 -12900K exist, it's just not generally true.
While the M1 Pro may not be orders of magnitude faster than what today's best Windows systems can do (especially on the desktop), it is massively faster than last-gen MacBook Pros that use Intel CPUs. The stuff of the 9th Apple owners will definitely get a huge upgrade, at least 50% more for pure CPU-heavy workloads and even more for niche apps that benefit from the M1 Pro design.
So the M1 Pro's performance and efficiency are very good, but I have some complaints.
Using the MacBook Pro in a wide variety of productivity apps still feels like users are testing the Apple Silicon Beta. There are many apps that have still not been updated to use the M1 family natively, even though the first M1-based computers hit the market a year ago. Rosetta 2 works very well, but it costs performance and efficiency. Additionally, some updated apps either lack features (like Blender without GPU rendering support) or run poorly compared to the same app on Windows (like FL Studio). That's not good for a production-centric system where users may need a specific app for their workflow and those apps need to work well for maximum productivity.
Gaming is a bit ridiculous on macOS too, from both a compatibility and a performance standpoint. One of the best things about buying a powerful Windows laptop for productivity is that you can sneak in with games on the side too, which makes these systems pretty versatile given the huge library of games supported. Macs don't have this versatility, and the M1 Pro is a poor choice for gaming.
Finally, the pricing. The full M1 Pro starts at $ 2,500 in laptops, with the full M1 Max costing at least $ 3,300 (or more if you want decent SSD capacity). Storage and storage upgrades are disgustingly expensive: $ 400 more to move from 16GB to 32GB of RAM and $ 200 more to move from a 512GB SSD to 1TB is ridiculous and due to the fact that lack of upgrade option for the user even more terrible. Apple rubs you in for absolutely everything you have with this version.
But if we take the base 16-inch M1 Pro configuration we bought, we're looking at a $ 2,500 laptop. There are some pretty powerful Windows machines out there for that price, like the Gigabyte Aero 15 OLED or even what I think is an expensive laptop like the Razer Blade 15 Advanced. Apple could justify exorbitant prices if they were outright faster on everything, but that's not the case, especially not with the M1 Pro configuration when compared to these equivalent models with beefy discrete GPUs. And I'm not even talking about value-driven Windows laptops, which are a completely different thing.
Apple is acting like a monopoly with this pricing, which I think is fair enough considering the vast majority of MacBook buyers are existing Mac users. From this group, professionals who run certain apps that take full advantage of the hardware will be delighted with the M1 Pro. But if Apple is looking to attract Windows users to the Mac, pricing this high isn't the way to go – for the mainstream price point, the $ 999 MacBook Air M1 is a nice system, but that's a whole different ball game.
With these issues aside, Apple Silicon is certainly in an interesting, compelling position and there is plenty of room for x86 competitors to improve, especially in terms of efficiency. For example, I would like to see what AMD could do with a large APU design with a more powerful GPU and the existing efficiency of their chips. There will be a big fight going on and from here we hopefully can continue to test the latest Apple chips as they hit the market.