At the beginning of this year, Intel officially released the 12th-generation Core mobile processor, and launched a total of 45W H series for the main performance, 28W P series for all-round notebooks, and 15/9W U series for thin and light notebooks. Later, Intel expanded its product line and launched HX series processors with up to 16 cores/24 threads.
The 12th-generation Core is a product based on the Alder Lake architecture. The CPU core is changed to a large and small core architecture, which consists of a small core E-Core focusing on energy efficiency and a large core P-Core focusing on performance. At the same time, a series of functions of DDR5 and Thunderbolt 4 have been added. As originally expected, the E-Core delivers decent performance. The performance is close to Skylake, and the power consumption is lower, which greatly improves the battery life of the laptop. However, probably because E-Core and P-Core are designed on the same Ring, E-Core does not significantly reduce power consumption. Therefore, the battery life of the 12th generation platform has not been qualitatively improved compared to the 11th generation platform, and it is difficult to compete with the Macbook series products based on the M1 processor.
Before the 12th generation Core platform, we compared the 11th generation Core platform and Apple M1, and the result is that the former wins the conclusion due to the Windows ecosystem. What if the new 12th-generation Core platform meets M1 again? This time I used the Huawei MateBook 16S to compare it with the Apple Macbook Pro. The key configurations of the two products are as follows:
MateBook 16S: Intel 12900H, UHD770, 16GB DDR5, 1TB SSD, Windows 11.
Macbook Pro: M1 Pro, 16GB RAM, 1TB SSD, macOS Monterey.
The following tests focus on the two main productivity directions of image processing and video editing. As for games and battery life, the former is a must-win for Windows, and the latter is a must-win for M1.
CPU Performance Test
Before the test starts, we first use CINEBENCH R23 to test the CPU of two laptops. The MateBook 16S scored 12,624 pts in multi-core and 1,659 pts in single-core, while the Macbook Pro scored 12,366 pts in multi-core and 1,531 pts in single-core. The single-core advantage of the Intel 12900H is more prominent, about 8% higher than that of the M1 Pro, but the multi-core results are very close.
Then there’s CrossMark, a cross-platform benchmark tool for productivity, creativity, responsiveness, and the main experience of CPU performance. The Macbook Pro scored 1583 overall, 1365 for productivity, 2074 for creativity, and 1100 for responsiveness. The MateBook 16S scored 1827 overall, 1712 for productivity, 1944 for creativity, and 1838 for responsiveness. The latter scores higher.
IO Performance Test
Next is the IO test, using CrystalDiskMark software on the MateBook 16S and AmorphousDiskMark on the Macbook Pro. Although the names of the two softwares are different, the test methods are the same, and the test data can be directly compared.
The first is the built-in SSD performance. From the test data, it can be judged that a PCIe 3.0 SSD used by the MateBook 16S (also confirmed to be a PCIe 3.0 SSD by the CrystalDiskInfo test), the peak (sequential) read and write speeds are 3515.47MB/s and 2873.91MB/s respectively. The Macbook Pro’s built-in SSD peak (sequential) read and write speeds are 6967.77MB/s and 5282.80MB/s respectively, so it can be judged that this is a PCIe 4.0 SSD. It’s just that the random read performance of the built-in SSD of the Macbook Pro is very poor, especially the random write performance, which is only about 20% of that of the MateBook 16S.
The random read and write speed of the built-in SSD of the Macbook Pro is surprisingly poor. In order to further find out the problem, we used KIOXIA PSSD mobile hard disk and Lexar CFexpress Type B Professional memory card (with the original card reader) to further test the IO performance of the two notebooks.
Lexar CFexpress Type B
The results were very unexpected. Using the same memory, the IO of the Macbook Pro is almost completely defeated by the MateBook 16S. In particular, the random read and write speed lags far behind, which is likely to be the problem caused by the M1 Pro chip.
The Macbook Pro’s poor performance in the IO test is surprising, so how does it perform in image processing? The following three sets of tests are carried out, and the specific test settings are as follows:
Lightroom I: Convert 233 photos in RAW format with a total of 6.11GB to JPG without any adjustment of brightness, noise reduction, or sharpening. The software version of the two notebooks is 11.4.1.
Lightroom II: Converts 233 RAW photos totaling 6.11GB to JPG. Settings: Highlights -48, Shadows 21, White Levels 12, Black Levels -17, Sharpening Amount 40, Radius 1.2, Detail 30, Mask 0, Noise Removal Brightness 25, Detail 50, Contrast 15, Color 25 , Detail 50, Smoothness 50.
Topaz DeNoise AI: Perform noise reduction processing on a 16512*11008 (180 million pixels) image, set to Standard, Remove Noise 16, Enhance Sharpness 60. The MateBook 16S software version is 3.6.2, and the Macbook Pro software version is 3.3.4.
The first thing to note is that the MateBook 16S equipped with UHD770 integrated graphics card cannot perform GPU acceleration when Lightroom outputs photos. Therefore, the MateBook 16S relies on Intel 12900H to output photos. The time-consuming in Lightroom I and Lightroom II is 1.45 times and 1.4 times that of Macbook Pro respectively.
In Topaz DeNoise A, which emphasizes AI computing power, the gap between the two notebooks has been greatly reduced. The MateBook 16S takes 1.08 times as long as the Macbook Pro, and the Intel 12900H has huge room for improvement. When processing DeNoise A, the CPU is used first and then the GPU is used for calculation. In the computing stage of the CPU stage, there is a problem of one-core working and multi-core onlookers.
The final test is the M1 processor’s traditional advantage: video editing. However, it is unexpected that the Macbook Pro equipped with the M1 Pro chip has a problem here, and it is very serious. Let’s take a look at the test setup first:
H264 to H264: Use Premiere Pro version 22.5 (Note: Adobe has optimized the performance of M1 since version 22.4, and the speed of exporting 10bit 4:2:0 HEVC video will be up to 10 times faster.). Compress an H264 video with a duration of 19 minutes and 57 seconds, 4K60P, a bit rate of 157Mbps, and 10bit 420 into an H264 video of 30Mps, 4K60P, and 10bit 420. MateBook 16S uses hard decoding (the same below).
H265 to H264: Also use Premiere Pro version 22.5. Compress an H265 video with a duration of 24 minutes and 38 seconds, 4K50P, a bit rate of 200Mbps, and 10bit 422 into a 30Mps, 4K50P, 8bit 420 H264 video after loading the LUT.
ProRes to H265: Mabook Pro uses Final Cut Pro 10.6.3, MateBook 16S uses Premiere Pro 22.5. Convert a ProRes LT of about 700Mbps, 4K50P to a H265 video of 22.5Mbps, 4K50P, 8bit 420.
In the video transcoding test, the Macbook Pro with the M1 Pro was beaten by the MateBook 16S with the Intel 12900H. In ProRes to H265, it was thrown away by the latter by a large margin, and the time was 1.45 times that of the latter. The smaller difference is H265 to H264, which takes 1.10 times as long as MateBook 16S.
When Apple released the Macbook series notebooks based on the M1 processor, the audience was amazed. Powerful battery life and hard decoding/encoding capabilities have successfully overwhelmed the Windows + X86 platform. However, Intel added the H265 hard decoding capability to the 11th generation Core, and the performance of the 12th generation Core platform has been further improved. As a result, not only the H264 and H265 encoding performance exceeds the M1 Pro, but also the performance of ProRes to H265 is not as good as 12900H. Intel beats hard decoding with hard decoding.