Apple announced its next-generation MacBook Pros this past October, and with it, Apple’s next generation of M1 silicon. Based on initial reviews, the changes to these new MacBooks seemed to excite the tech world. It wasn’t perfect at launch, far from it, but many ISVs jumped on board to write native apps for the new platform. Apple returned to a previous year’s design while seeking to blow the competition out of the water with performance per watt (PPW).
I can argue that Apple took the right short-term step forward in redesigning the new MacBooks, and the PPW on the M1, M1 Pro and M1 Max is impressive. However, without fail, there seems to be an “Apple spell” that comes over people almost every time a new Apple product is released. Again, I believe Apple did an excellent job of redesigning the new MacBooks, and the PPW is impressive. Still, there needs to be some dispelling because, unlike most of the tech industry, I do not believe the M1 Max wins in every scenario.
While I knew Apple would fare well in PPW, perform poorly in AAA games, what about raw performance in content creation scenarios? To go head-to-head against the new MacBook Pro M1 Max in terms of raw performance, the Asus Zephyrus m16 seemed like a logical option. Before we get to the benchmarks, let us lay out the test methodology, the specifications of the systems, and then jump into the logistics and results of the benchmarks.
For the testing of both systems, I ran the benchmarks on AC (plugged in its stock charger) rather than DC to give both systems the chance to perform at their best. Most systems out of the box have higher performance configurations when running on AC. For example, the MacBook I used for testing under the settings Battery>Power Adapter> Energy Mode is already set to High Power. The setting under Battery>Battery>Energy Mode is set to automatic, which “will automatically choose the best level of performance and energy usage.” Similarly, the Zephyrus m16 has settings to change in Windows 11 settings and Asus’ Armoury Crate. I kept both systems at stock settings, with the one exception of changing the screen timeout so that the systems do not go into sleep mode while benchmarking.
I also wanted to mention why I included Asus’ Turbo Mode in the benchmarks. The entire article is about both systems in their stock form. I also think it is important to show that the Intel-based system is set up in such a way that OEMs and even users can achieve greater performance. The stock mode is set up to achieve greater performance while taking into consideration energy efficiency. Turbo mode is only available when connected to AC power and maximizes power to the CPU and GPU for a bump in performance. For Blender, it was up to 7% on the longer renders, for KeyShot and V-Ray was about 2% and for Redshift, it was about 28% difference in performance.
Both systems were also in the same environment, with the same surrounding airflow. I updated both systems before starting and updated the Zephyrus from Windows 10 to Windows 11. The systems were run through each benchmark three times back-to-back with a short cool-down time in between. I then took the average of those three benchmarks.
For all of the benchmarks, all except for Redshift, the MacBook ProM1 Max could not benchmark using the GPU compute (CUDA). To give a perspective on the performance of the GPU, I ran the Blender CUDA benchmark on the Zephyrus m16 with a GeForce RTX 3060. I took the percentage difference between the Zephyrus m16 CUDA and the MacBook Pro M1 Max CPU for each of the five benchmarks and then averaged the percentage difference between the five benchmarks. The Zephyrus m16 rendered the benchmarks 89% faster than the MacBook Pro M1 Max and 78% faster for the top three of the five longest renders. I believe this is a significant performance difference that should be considered when comparing the two systems. However, the difference does not express the raw and fair performance of the MacBook Pro M1 Max since the difference is primarily a lack of support. I completely understand that the CPU and GPU benchmarks are not a one-to-one comparison and that most of the delta is due to a lack of support for GPU rendering in these programs Those that need the performance are not going to wait around for the GPU support. Blender does support the M1 Max GPU in Alpha but it is only experimental and is not recommended for studio use. That is why I am leaving it up to the Redshift benchmark to showcase the M1 Max’s GPU performance.
- 10-core CPU with 8 performance cores and 2 efficiency cores
- 32-core GPU
- 16-core Neural Engine
- 32GB unified memory
- 400GB/s memory bandwidth
- 1TB SSD
- macOS Monterey 12.1
- 16-inch display
- Intel Core i9-11900H with 8 cores
- NVIDIA GeForce RTX 3060 with 6GB GDDR6 dedicated memory
- 32GB DDR4; 16GB DDr4 onboard memory and 16GB DDR4-3200 So-DIMM memory
- 1TB M.2 NVMe PCIe 3.0 SSD
- Upgraded to Windows 11 Home 21H2
Apart from the MacBook having two extra efficiency cores, the Zephyrus and MacBook have the same number of performance cores, same 16-inch form factor, amount of memory for the CPU, and the same amount of storage. Both systems were updated to the latest OS version before testing.
Blender was the longest test that I did. Blender is a free and open-source 3D software that supports modeling, rigging, animation, simulation, rendering, compositing. Blender has an automated benchmarking tool with six benchmarks that Blender renders scenes. There are options for benchmarking different devices in the scene, i.e., CPU compute, GPU compute with CUDA, Optix with GPU, and NVIDIA Raytracing acceleration. I only did the CPU compute since that is the only device that the benchmark supports on the MacBook. The Metal API is not integrated into Blender for GPU acceleration. I ran the five benchmarks together rather than waiting for one to finish and then letting it cool down. On average, the rendering time of the benchmarks varies in length. In order of shortest to longest are bmw_27, fishy_cat, koro, classroom, pavilion_barcelona, and victor. The benchmarks are in seconds, and a lower number indicates the render speed. The lower the number the better.
The Zephyrus m16 performs better on four out of the six benchmarks, with one benchmark going to Apple and another being within seconds. The Zephyrus was, on average, almost 3 minutes faster (174 seconds) than the MacBook M1 Max at rendering Blender’s benchmark for victor, which is the longest render. It is a minute and a half faster (91 seconds) than the MacBook Pro M1 Max on the second-longest render, the pavilion_barcelona. I took the percentage difference of each of these benchmarks, averaged them out, and came to an 8.7% difference in favor of the Zephyrus. If I take the three longest renders of the six benchmarks, classroom, pavillon_barcelona, andvictor, the Zephyrus, on average, outperforms by a margin of 13.7%. For professionals that understand the difference that 10+% in render performance can bring, I believe this is a significant margin to consider and one that makes a difference.
V-Ray is another rendering tool that is used on different 3D modeling and design applications. It is used to create photo-realistic imagery and animation, and it has a free benchmarking tool called V-Ray 5 Benchmark. Like the Blender render, there is a choice between CPU, Cuda, and Optix. Again, I am only doing CPU renders since macOS only supports the CPU-based encoding option. According to V-Ray, the final score is based on internal statistics of the calculations per minute and is readable in vsamples for V-Ray.
The V-Ray 5 Benchmark is like Blender’s, where the Zephyrus outperforms the MacBook Pro M1 Max in rendering. To provide a reference point, I found the score of the 2019 MacBook Pro 16 with an Intel Core i9-9980HK. It has a score on the V-ray 5 Benchmark of 7183 vsamples. The Zephyrus m16 outperformed the MacBook Pro M1 Max by a delta of 1807 vsamples. The percentage difference between the Zephyrus m16 and the MacBook Pro M1 Max is 20.9%, which is twice as much of a percent difference than the Blender benchmarks.
Keyshot is another free 3D software for creating renders and animation. It has a free standalone application for viewing interactive and photo-realistic 3D models made in KeyShot. KeyShot Viewer has a free benchmark within it that, like the other two benchmarks, has the option for CPU compute, Cuda, and Optix but not for macOS. The way that Keyshot measures its benchmark is also different. According to the Keyshot Viewer website, the results are scored in multiples based on render time. Higher scores are better and scores higher than a 1.0 are better than the reference system, an Intel Core i7-6900K CPU.
When compared to the Intel Core i7-6900k as reference, the MacBook Pro M1 Max was 1.5 times faster, while the Zephyrus came in at 2.24 times faster. Zephyrus m16 outperforms the MacBook Pro M1 Max by a margin of about .66 points. Rather than making a straight percentage difference with respect to the reference points, I took the 1:X ratios from the reference score of the Intel Core i7-6900K and set both ratios equal to give us a percentage difference between the two systems. The Zephyrus m16 then is 1.43 times faster than the MacBook Pro M1 Max or 43% faster than the MacBook Pro M1 Max. The MacBook Pro M1 Max almost becomes to the Zephyrus m16 what the Intel Core i7-6900K is to the MacBook Pro M1 Max.
I saved the best benchmark for last. It is not the best benchmark in terms of superiority. It is the best benchmark because Apple used it in its official press release of the MacBook Pro with M1 Pro and M1 Max. It is also the only benchmark out of the four that has support for GPU-encoding on macOS. RedShift is a GPU-based render from Maxon that can be used as a standalone rendering tool or an integrated rendering option for Cinema4D. It is a command-line tool that can load a scene, render it, and measure the time it took to render.
Apple’s claim is that the 32-core GPU in the M1 Max offers up to 4 times faster render in Maxon Cinema 4D with Redshift. If you read the fine print, this is in comparison to the previous generation’s 2.4 GHz 8-core Intel Core i9-based 16-inch MacBook Pro systems with Radeon Pro 5600M graphics with 8GB HBM2, 64GB of RAM, and 8TB SSD. This 2019 MacBook Pro is the same one that I used as a reference in the V-Ray benchmark, the Intel Core i9-9980HK. The MacBook Pro M1 Max is a whole three seconds slower than the Zephyrus m16 with an NVIDIA GeForce RTX 3060, and the Zephyrus in Turbo mode is almost two minutes faster than that.
Battery life and price
My biggest takeaway from doing these benchmarks on the new MacBook Pro M1 Max and its closest competition is that a lot of what you are paying for is the extra battery life and lower fan noise that comes with its impressive performance. I saw a small difference in running some of these benchmarks on DC compared to AC, which is a good thing. The MacBook Pro is also noticeably quiet and rarely did I hear the fans kick on. But the MacBook is not the best at everything, and as a mobile workstation, the Zephyrus was able to outperform the MacBook Pro M1 Max in raw performance—even if it must hug an outlet to do so. I also must mention that the Zephyrus m16 is significantly less expensive than the MacBook Pro M1 Max. The Zephyrus m16 that I tested with was about $1,800, while the 16-inch MacBook Pro M1 Max was about $3,400. That is almost twice the price for the benefit of not needing to hug an outlet or hear the fans.
There are countless other benchmarks out there that could have said something different, but I think the Blend, V-Ray 5, Keyshot, and Redshift content creation benchmarks represent a considerable portion of the professional content creation community. The ASUS Zephyrus m16 outperformed the MacBook Pro M1 Max by a considerable percentage in terms of raw performance. That is not to say that Apple’s silicon is bad, but it does not win in every scenario. 3D rendering is one of those scenarios for both the GPU and CPU compute.
I think all competition is good competition, and just as I would like to see Apple improve its raw performance, likewise, I would like to see Intel improve on its PPW. Intel, AMD, and Apple are all pushing innovation in silicon, and I love it.
Note: Moor Insights & Strategy co-op Jacob Freyman ran all benchmarks and contributed to writing this article. Special thanks to Intel Corp for lending us the test equipment.