Review, compare and test Apple M1 Pro
- 2021-11-10 06:00
- Recompile By Brian - CPU Review
Today, we're taking our benchmarks where they've never been before with an in-depth look at Apple Silicon products. A couple of weeks ago we bought a new Apple MacBook Pro 16 powered by the M1 Pro SoC and have been passing its steps ever since.
So this review is all about benchmarking the new M1 Pro and properly comparing it with the best x86-based CPUs out there right now from AMD and Intel. And by fair comparison, we don't mean to run Geekbench and call it a day. We're not actually running Geekbench... we mean running a bunch of real-world cross-platform apps and comparing performance on today's best laptop chips, all with run under fair and comparable conditions.
This will be interesting as to be fair there is a lot of hype surrounding the Apple M1 silicon and its supposed desktop-destroying performance. With the hype there is often excess and there is certainly a lot of fanaticism, including the gospel of Apple's internal standards, so we will see today that all of that has been How guaranteed.
The Apple M1 Pro is a very different SoC design from the chips we typically review because it uses the Arm family of architectures instead of x86. That doesn't mean it's better, it just solves things in a different way. At the heart of this processor are up to 10 CPU cores, split into 8 high-performance Firestorm cores and 2 efficient Icestorm cores in a hybrid design. The Firestorm cores are clocked at up to 3.2 GHz and the Icestorm cores are at 2.1 GHz, although this depends on the load on the system and the workload of all cores dropping that clock by a few hundred MHz.
On the GPU side, we have a 16-core design in the M1 Pro, up to 32 in the M1 Max (offered in a more expensive version of the same laptop). Connected to the CPU and GPU in a unified design is a 256-bit wide LPDDR5 memory bus that offers a little over 200 GB/s, you can think of this like if an x86 design included memory quad channel instead of dual channel, that's for sure. lots of memory bandwidth and more than you get in similar products from Intel or AMD. There's also plenty of L2 cache and 24MB of top-level caching on the M1 Pro. Both the top-level caching and caching controllers are doubled on the M1 Max design to accommodate a larger GPU.
Apple has taken Nvidia's approach to naming the M1 Pro by launching two configurations with essentially the same name. The top-tier design features a full 10-core CPU and a 16-core GPU, but the base design cuts it down to just 8 CPU cores and 14 GPU cores. While this is prominently advertised on their product page for the MacBook Pro, Apple doesn't bother to distinguish the actual product name other than the M1 Pro.
For testing, we purchased a 16-inch MacBook Pro with the full M1 Pro configuration. We also chose 16GB of unified memory to match existing laptop test data using 16GB RAM and 512GB SSD. Apple's upgrades to both RAM and storage are horribly expensive, going to cost an extra $600 to bump that up to 32GB of storage and 1TB of SSD. You also can't upgrade these components on your own, so you have no choice but to pay these prices if you need more capacity, not that you’d expect anything different from Apple.
For today's testing, we'll be comparing the M1 Pro with our available data for AMD and Intel CPUs, all of which are power normalized to those chips' default TDPs. Of course, there are some additional challenges that we spent last week working on. Specifically, not all the apps we benchmarked are available on macOS. So we're only covering cross-platform benchmarks with the macOS version, whether it's the native ARM version or the x86 version emulated on Apple Silicon via Rosetta 2.
This immediately introduces a variable where the Windows and macOS versions of the app will behave slightly differently, but that's the best we can do depending on the circumstances. Let's start testing.
In Cinebench R23 multithreading, which is a native Apple Silicon app, the M1 Pro was slightly faster than the next best CPU we tested, the Ryzen 9 5900HX. In a 10-minute period, the M1 Pro ran 4% faster, although it also consumed less power as we'll talk about later.
The M1 Pro resembles a 35W class CPU, compared to 45W for the 5900HX; The M1 Pro ended up being 12 percent faster than AMD's best 35W CPU in the Ryzen 9 5980HS, which is probably more comparable. Apple even takes the lead on Intel, where the M1 Pro is 14% faster than Intel's best silicon at 45W, the Core i9-11980HK.
Also of note is the Intel Core i7-10875H. This CPU is most comparable in multi-threaded performance to the 9th Gen Core i9-9980HK available in the best 2019 MacBook Pro configurations. The M1 Pro is almost 60% faster in this benchmark, which is why Apple users are excited about boosting performance with this new generation by using Apple Silicon.
In terms of single-threaded performance, there's not much of a difference between the M1 Pro and the best CPUs from Intel or AMD. Apple's chip sits right between the Ryzen 9 5980HS and the Core i9-11980HK, all of which are a bit too much for these CPUs in this workload. However, performance is 20% faster than previous Intel CPUs that Apple has used.
Handbrake is available as a native Apple Silicon app on macOS, and shows us CPU-based video encoding using x265. While this app does run natively on the M1 Pro, it benefits significantly on x86-based CPUs from AVX instructions. As such, the M1 Pro isn’t that impressive compared to its x86 equivalents, such as the Ryzen 9 5980HS or Core i9-11980HK. While performance isn’t bad, the M1 Pro does come in 9 percent behind Intel’s best Core i9, and 24% behind AMD’s best Ryzen 9 processor, though Apple’s power consumption is lower.
Handbrake is available as a native Apple Silicon application on macOS and shows us CPU-based video encoding using x265. Although this application runs natively on the M1 Pro, it benefits significantly on x86-based CPUs from AVX instructions. As a result, the M1 Pro is not very impressive compared to its x86 equivalents, such as the Ryzen 9 5980HS or the Core i9-11980HK. While the performance isn't bad, the M1 Pro falls 9 percent behind Intel's best Core i9 and 24 percent behind AMD's best Ryzen 9 processor, though Apple’s power consumption is lower.
The M1 Pro is a powerful CPU for CPU-based Blender rendering, outperforming the best chips from Intel and AMD. The performance lead over the Ryzen 9 5900HX is small at just 5%, but it’s a more substantial 23% faster compared to the 11980HK. The main issue MacBooks face in Blender is that while the app is available for Apple Silicon natively, it doesn’t support GPU rendering. When using an RTX 3050 with Optix, GPU rendering absolutely obliterates the M1 Pro with performance over 3x faster. Until Blender is updated with GPU support, MacBook Pros are not the best option.
The M1 Pro is a powerful CPU for CPU-based Blender rendering, outperforming the best chips from Intel and AMD. The performance lead over the Ryzen 9 5900HX is just 5%, but it's 23% faster than the 11980HK. The main problem MacBooks have in Blender is that even though the app is available for Apple Silicon, it doesn't support GPU rendering. When using the RTX 3050 with Optix, GPU rendering completely blurs the M1 Pro with up to 3x faster performance. Until Blender is updated with GPU support, the MacBook Pro is not the best choice.
While the M1 Pro has so far traded with the best x86 CPUs in rendering and encoding tests, Apple Silicon has a big advantage in compiling the code. I should note here that for our Chromium compiler, we benchmarked the compilation of the Windows version on Windows and the Mac version on macOS, using Google's recommended process for each operating system. onion. Unfortunately I tried but couldn't get the MacBook Pro to compile the Windows version, although it can be done on Intel Macs. So this benchmark isn't as bold as I'd like, but it's still a real-world use case in my opinion.
In any case, the M1 Pro beats its x86 rival, delivering 68% faster compile times than the Ryzen 9 5900HX, the next fastest we've reviewed. The M1 Pro has a huge advantage in these workloads, especially with twice the memory bandwidth compared to other laptops in these charts. I doubt these new MacBooks will be great for programmers.
Matlab is currently only available through Rosetta 2 emulation, and performance on the M1 Pro suffers as a result. The SoC will eventually be equivalent to a Ryzen 7 4800H or Intel Core i7-1165G7 CPU, which are good chips, just not the fastest on the market today. The M1 Pro is 30% slower than Intel's Core i9-11980HK in this application, so for now you're better off using a modern Windows machine for your engineering work in Matlab.
Microsoft Excel is available natively for Apple Silicon, and our numerical processing benchmarks perform well. The M1 Pro ended up being slightly faster than the best 35W CPU we tested, the Ryzen 9 5980HS, but it can't compare to the best x86 has to offer in a portable form factor. The Core i9-11980HK is 23% faster and the Ryzen 9 5900HX has a slight lead.
The M1 Pro is a monster for 7-Zip compression, using the original Apple Silicon version against the same benchmark on Windows. The M1 Pro is 11% faster than the i9-11980HK despite using significantly less power during this continuous workload and 27% faster than the 5900HX, which is AMD's best display.
However, the tables will change when 7-zip decompression is used. This benchmark loves concurrent multithreading on x86 CPUs and is exceptionally fast on Zen, resulting in a command lead over the M1 Pro. This is a weak result for Apple Silicon, as it is 32% slower than the Ryzen 9 5900HX and more on Intel's list of 10th Gen processors.
PDF export is a single-threaded application, and Acrobat Pro is only available through the Rosetta 2 emulation on macOS. Don't worry, though, because the M1 Pro still easily beats its x86 rival with 25% faster performance than the best AMD or Intel has to offer.
FL Studio is a new addition to our benchmark and something we've been exploring for a while. Although there is a version of the original Apple app, it unfortunately runs very badly on the M1 Pro. I've quadrupled this as I can't believe it against our Windows based systems but it represents the latest version of the app, outputting the default track. Despite hitting 100% CPU usage on my MacBook Pro during the trace export, I think the app has gone wrong and it needs serious repair on macOS.
For Adobe Premiere, we've updated our benchmarks to use the latest 2022 version, running natively on Apple Silicon. Using the latest version of PugetBench, the overall results look very impressive for the M1 Pro, outperforming very capable Windows machines like the 11800H and the RTX 3070. It's certainly an excellent result, but let's go deeper into the sub-points.
On the export front, the results from Premiere are similar to what we've seen in the previous Handbrake benchmark, where the M1 Pro isn't the most impressive. PugetBench uses a mixture of exports, and while the SoC supports a wide range of hardware-accelerated encoding, the Nvidia GPU is extremely good at accelerated encoding and therefore performs quite well in this benchmark as well.
Where Apple is dominating is the live playback feature, which refers to how smooth the footage is in the timeline. The M1 Pro with features like hardware-accelerated ProRes ends up being a bit faster than competing options, with 11th Gen Intel-based machines also doing well due to the powerful decoding support with their iGPU
Effects scores are also quite good on the M1 Pro, lying between the 5900HX and 5800H in this mostly CPU limited score. Then we get the GPU score, where the performance is below the RTX 3050 as expected. I haven't actually had any low-powered GPUs in this benchmark, and the M1 Pro's 16-core GPU isn't really comparable to the Nvidia 80W+ GPU; it ended up being 23% less powerful than the RTX 3050 but using much less power.
While Premiere has an original Apple Silicon version, After Effects does not, and requires the use of Rosetta. The M1 Pro still gets a decent score in PugetBench, outperforming the Ryzen 7 4800H configuration with RTX 2060 graphics, but it falls short of Intel's flagship machines.
This benchmark doesn't use the GPU but the final score relies heavily on CPU performance, with the 11800H performing exceptionally well thanks to its excellent single-threaded performance. As for Photoshop, unfortunately we are unable to rate this app because the PugetBench engine we use is not yet compatible with the original Apple Silicon version of Photoshop and I felt using the Intel version of Photoshop instead would be unfair.
A new addition to our computer benchmark is the Gigapixel AI image upscaling, using the popular Topaz Labs engine. We ran this benchmark using GPU acceleration in the fastest possible mode, which usually runs great on Nvidia RTX 30 series GPUs. The M1 Pro has to use Rosetta 2 emulation but still performs excellently for the low-power configuration, which falls between the RTX 3060 and RTX 3050 Ti results. The best Windows machine we tested was 26% faster, but that system is more comparable to the M1 Max.
Our ultimate computer benchmark is another new addition, Agisoft Metashape that performs photogrammetry to convert static images into 3D models. This is a GPU accelerated benchmark that goes through the entire multi-step process to create a model from a sample image. We are using the latest beta version of the app with native Apple Silicon support.
In this benchmark, the M1 Pro ranks last among the configurations we've tested but that's not worth mocking as the SoC is only marginally worse than the more power-hungry configurations with larger GPUs. For example, the overall M1 Pro package was only 4% slower than a system with Nvidia's RTX 3050 running at 80-95W max, even though the entire M1 Pro only peaked around 60Wf power usage in heavy CPU + GPU applications.
Next, we have the gaming performance in the original macOS games, and to be honest, the macOS gaming ecosystem is pretty bad. Of the 23 games we reviewed on the laptop, only 5 of them have a macOS version, 4 of them work on the M1 Pro, and none of them have a native Apple Silicon implementation. This is a pretty similar issue with my entire game library, where only 16% of games have a macOS version. That means you probably shouldn't buy a MacBook for gaming, but we'll be testing supported games.
Shadow of the Tomb Raider was specifically evaluated by us using the last part of the in-game benchmark, not the entire process. In this segment, the M1 Pro ranks between the GTX 1650 Ti and GTX 1060 6GB in terms of performance, which is perhaps to be expected given its low-power nature. This Apple Silicon configuration is 22% slower than Nvidia's full-powered RTX 3050, keeping it solid at the input level of the GPU we tested.
I should note here that the game benchmarking tools available for macOS are extremely rough compared to what is possible on Windows right now, which has access to powerful software like OCAT. So while these results are representative of in-game performance, the accuracy is not where I’d like it to be, but we just don’t have the resources to develop our own macOS FPS counting tools.
I should note here that the game benchmarking tools available for macOS are difficult compared to what is possible on Windows currently, which has access to powerful software like OCAT. So while these results represent in-game performance, accuracy isn't where I want it to be, but we don't have the resources to develop our own macOS FPS counting tools.
The M1 Pro can run Borderlands 3, but it runs terrible and I don't know if it's because the game needs to be emulated via Rosetta 2 or just the GPU in this chip is weak. Performance is lower than even the GTX 1650 Ti, and to be honest the game was basically unplayable using the combination of Ultra settings we tested.
Metro Exodus is the best example of game performance on the M1 Pro, although I'm not sure how comparable game settings are since the macOS version lacks options for tessellation and Hairworks. But if we assume for a moment that the game is the same, the M1 Pro performs very well, falling between the RTX 3050 and RTX 3050 Ti when playing at 1080p. Unfortunately, however, Apple Silicon doesn't support hardware-accelerated ray tracing or DLSS, so its feature support in this game is worse than what you'd get on a laptop that's been built in equipped with an RTX 3050.
Finally, we have Civilization VI, which we test using low settings mainly to compare integrated graphics. With a more powerful GPU, it tells us about the CPU's limited performance. Here's another game that ran poorly on the M1 Pro, failing to match the GTX 1650 Max-Q we've seen in thin-and-light laptops before. This game doesn't seem to be optimized at all for Apple Silicon and it can get muffled due to the fact that it has to run an emulator.
The final area of performance to assess is power consumption, as measured at the wall but excluding idle power usage. This gives us a rough idea of how much power the various system components are using under heavy usage, while accounting for differences in displays and so on.
The final area of performance to evaluate is power consumption, as measured on the wall but excluding idle power usage. This gives us a rough idea of how much power the various system components are using under heavy use, while also taking into account screen differences and so on.
First, we have Cinebench R23 multithreading, which highlights the great performance of the M1 Pro. While slightly outperforming the Ryzen 9 5900HX, the power consumption is 13W lower in the long-run of the benchmark, more than in the 35W CPU class.
Even more impressive is that a lot of x86 CPUs burst quite high to achieve strong multi-threaded performance in a short time, such as in applications like 7-Zip or Excel. That is not the case with the M1 Pro, which manages competitive performance in those applications despite using only 43W of normal peak power. The fact that Apple can get the performance they do without any kind of real boost or turbo clock is very understandable and indicates a design built for efficiency.
Where Apple sees the biggest lead over its competitors in efficiency is in single-threading Cinebench. Performance here is roughly on par with AMD and Intel's best laptop CPUs, but power consumption is an order of magnitude lower. Consuming only 7W of power is pretty insane as the latest x86 CPUs are pushing power above 20W when boosted above 4.0GHz.
This is why the new MacBook offers such great battery life, it leads the competition in efficiency when using light threaded workloads, allowing for the best performance without does not reduce battery life.
The M1 Pro is also particularly efficient when it comes to gaming. Peak power usage in the heavy Shadow of the Tomb Raider section hit 60W, less than half that of our test system RTX 3050 and only slightly higher than the Ryzen 9 5900HX using the company's integrated GPU. it. Performance is also lower than the RTX 3050 in this test, 24% behind the RTX 3050, but power consumption is 55% lower so performance per watt is very beneficial for Apple in this compute workload. This is why Apple designed a monolithic SoC with unified memory, it has a lot of benefits from a power perspective.
Testing out the new 16-inch MacBook Pro for the past two weeks has been a lot of fun and a fun challenge to learn more about macOS and its various apps. Even more interesting is how the new M1 Pro performs in real-world applications, so let's break it down in the lengthy conclusion.
From a pure performance perspective, the Apple M1 Pro is an impressive SoC. Performance does vary slightly depending on the application, but overall the CPU is very competitive with today's best x86 laptop CPUs from Intel and AMD. In the best cases, when an app can really benefit from specific features of the M1 Pro, like massive memory bandwidth, the M1 Pro can crush what's available on Windows-based devices. At other times, performance is roughly on par with CPUs like the Ryzen 9 5900HX and Core i9-11980HK, and in worst-case scenarios, like CPU-based video encoding, it can fall behind. Overall, you're definitely getting state-of-the-art performance.
Where Apple can extend its lead even further is when you have workloads specifically tailored to use the M1 Pro's accelerators. Video editing with ProRes acceleration is one example, although the export performance isn't quite as good as the best x86 systems equipped with Nvidia discrete GPUs. If you have those kinds of workloads, and I imagine a lot of them will be first-party Apple apps like Final Cut or Logic, then you'll see big profits from using the MacBook Pro with these new M1 chips.
However, GPU performance is a mixed bag, 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 perform significantly better. tell. Essentially, the M1 Pro's GPU equates to a typical lower-power discrete Nvidia GPU in productivity applications (thus, slower than the full-powered RTX 3050), and is pretty mediocre in gaming due to many issues including lack of optimization and emulation
Performance isn't the star of the show, it's efficiency that elevates the M1 Pro to a class of its own.
However, good performance is not the star of the show. It is efficiency that has raised the M1 Pro to a class of its own. In heavy multi-threaded CPU applications certainly the performance can be comparable to the Ryzen 9 5900HX, but the power consumption is at least 10W lower, which makes 35W-class CPUs the most comparable option. Apple simply has the best performance at the same power as today's best x86 designs, leading the way in the rise of light-threaded and GPU-heavy applications. The entire design is built around performance and great results for battery life.
This is to be expected, of course, since Apple is a full process node ahead of x86 chips: they're using TSMC's 5nm process, compared to AMD's 7nm and Intel's in-house 10nm SuperFin for their chips. With that kind of advantage, Apple should be ahead, and so should they. This gives the M1 Pro two significant bonus points: the performance drop isn't too great for the battery, because the chips are so efficient that they can run at full capacity without being plugged in. And second, the new 16-inch MacBook Pro is near silent during most workloads, while many Windows laptops crank up their fans to jet engine levels, especially in applications that use discrete GPUs.
While the M1 Pro beats the competition when it comes to free roaming off the charger, it's a more complicated situation when we're talking about the best performance a laptop can offer. Some of the gains the M1 Pro won is thin enough that switching an x86 laptop into "high performance, high power consumption" mode will see the instruction set switch back 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 when it comes to pushing performance out of the water at 70W or higher and can be fast. more than 10%.
Where does this matter if you care less about battery life and more about getting your tasks and workloads done as soon as possible. Windows systems might still have the edge there, especially if you have a system with half a discrete GPU that's pretty decent, given the M1 Pro's GPU isn't that amazing and the M1 Max's ridiculously expensive.
We are simply not in a position where the M1 Pro in the MacBook is beating or coming close to a desktop workstation in most applications, I know some people are trying to claim this, but when CPUs like The Ryzen 9 5950X and Core i9-12900K exist, it's just not true in a general sense.
However, while the M1 Pro may not be faster than what today's best Windows systems can achieve (especially on desktops), it's a lot faster than the previous generation MacBook Pros that use Intel CPUs. The 9th generation engine that Apple is using used to be slow by modern standards, AMD's Ryzen 4000 series easily beat it in multi-threaded applications in early 2020 and performance has improved further more since then. So Apple owners are definitely getting a big upgrade, at least 50% more in pure CPU heavy workloads and even more in niche apps that benefit from M1 Pro design.
So the performance and efficiency of the M1 Pro is very good, but I have some complaints.
Using the MacBook Pro across a wide range of productivity apps, it still feels like you're testing Apple Silicon. There are a lot of apps that still haven't been updated to use the M1 series natively, even though the first M1-enabled computers went on sale a year ago. Rosetta 2 works great, but it costs you performance and efficiency. On top of that, some updated apps are missing features (like Blender lacks GPU rendering support) or don't run as well (like FL Studio) compared to the same app on Windows. That's not good for a production-centric system, where users may need a specific app for their workflow and need those apps to perform well for maximum productivity.
Gaming is also a bit funny on macOS, both from a compatibility and performance standpoint. One of the best things about buying a high-performance Windows laptop to stay productive is that you can also get some gaming on the side, making those systems pretty snappy with the mail. Huge library of supported games. Macs don't have that flexibility, and the M1 Pro is a bad choice for gaming.
Finally, pricing. The full M1 Pro starts in the laptop for $2,500, with the full M1 Max going for as little as $3,300 (or more if you want good SSD capacity). Memory and storage upgrades are disgustingly expensive: an extra $400 to go from 16 to 32GB of RAM and an extra $200 to go from a 512GB SSD to 1TB, it's ridiculous and even more horrible due to people lack of upgradeability. Apple is giving you all you got with this release.
But if we take the base 16-inch M1 Pro configuration we bought, we're looking at a $2,500 laptop. You can get some pretty powerful Windows machines for that price, such as the Gigabyte Aero 15 OLED, or even what I want an expensive laptop like the Razer Blade 15 Advanced. Apple could justify the exorbitant price tag if they were outright faster in everything, but it's not, especially not with the M1 Pro configuration compared to equivalent models with this powerful discrete GPU. And I'm not even talking about more value-oriented Windows laptops, which is a completely separate issue.
Apple is running like a monopoly at this price point, which I guess is fair enough considering the majority of MacBook buyers are existing Mac users. From that pool, professionals who run specific applications that get the most out of the hardware will rejoice with the M1 Pro. But if Apple wants to lure Windows users to the Mac, setting such a high price isn't the way to do it - for the entry-level price point, the $999 MacBook Air M1 is a good system, but It's a completely different ball game.
Those issues aside, Apple Silicon is certainly in an interesting, compelling position, and there's a lot of room for x86 rivals to improve, especially on the efficiency front. For example, I wanted to see what AMD could do with a large APU design with their chip's existing efficient and powerful GPU. There will be a big battle in their hands and from here, hopefully we can continue to test Apple's latest chips as they come out.