Author | Travis
Produced by | OSC Open Source Community (ID: oschina2013)
Recently, Raspberry Pi released the long-awaited 64-bit Raspberry Pi OS. Theoretically, a 64-bit operating system can better utilize the advantages provided by a 64-bit processor (newer versions of Raspberry Pi have adopted 64-bit processors), support more than 4GB of memory (Raspberry Pi 4B can be equipped with up to 8GB of memory), and perform better when handling multimedia content.
So how do 32-bit and 64-bit perform in actual tests? Recently, the foreign media Phoronix conducted detailed tests on both. Let’s take a look at the test data and performance after switching from Raspberry Pi OS 32-bit to 64-bit.
To ensure fair testing and eliminate differences caused by different processors or memory, this test uses the Raspberry Pi 400, which has the following hardware configuration:
-
CPU: Broadcom BCM2711 Quad-core Cortex-A72 (ARM v8) 64-bit SoC @ 1.8GHz
-
Memory: 4GB LPDDR4-3200
-
Storage: SanDisk 16GB
-
Decoding: H.265 (4K@60), H.264 (1080P@60, 1080P@30), OpenGL ES 3.0
The Linux kernel version used in this test is 5.10, and the software package versions used in the tests are the same. The only difference in the tests is switching from the latest Raspberry Pi OS 32-bit to 64-bit.
In the first WebP image encoding test (lower scores are better; this is a test of Google’s libwebp using the cwebp image encoding utility), the encoding speed of the 64-bit system is about 4.5% – 7% faster than the 32-bit system, with only a slight difference between the two.
In the subsequent GraphicsMagick test (higher scores are better; this tests GraphicsMagick and its OpenMP implementation), there are significant differences in the test results between the 32-bit and 64-bit systems. The largest difference occurs during the HWB color space test, where the 64-bit system is about 47% faster than the 32-bit system, and there is also about a 32% performance gap when performing Swirl operations. Additionally, there are performance differences of about 17% – 28% during rotation, sharpening, and Gaussian noise processing.
In the FLAC audio encoding test (lower scores are better; this measures the time required to encode a sample WAV file to FLAC format), the 64-bit system also improved by about 17%; in the LAME MP3 encoding test (lower scores are better; this measures the time required to encode a WAV file to MP3 format), the 64-bit system improved by about 43%.
As seen in the previous tests, the 64-bit system shows a performance improvement of over 40% in certain test scenarios. In the Stress-NG test (higher scores are better; Stress-NG is a Linux stress testing tool that can test CPU, Memory, IO, and Disk), the performance gap between the two is further amplified, with the 64-bit system showing an approximately 232% performance improvement in Vector Math; in the Glibc C String Functions test, there is also about a 196% improvement.
In addition to multimedia encoding and system stress testing, performance tests were also conducted on Python and PHP. In the PyBench test (lower scores are better; PyBench reports the average test time of different functions, providing an estimate of Python’s average performance on the system), the 64-bit system improved by about 13% compared to the 32-bit system; in the PHPBench test (higher scores are better; PHPBench executes a large number of tests to evaluate various aspects of the PHP interpreter), the 64-bit performance improved by about 54%.
Although the above test results show that the performance of the 64-bit operating system has significantly improved compared to the 32-bit version under the same testing environment, this is not the maximum representation of the performance gap in all test scenarios.
In the Sysbench test (higher scores are better; Sysbench is a multi-threaded benchmark testing tool based on LuaJIT, specifically testing CPU and memory), the advantages of the 64-bit processor + 64-bit operating system are maximized, with the performance improvement of the 64-bit system reaching an astonishing 1380%.
Here, we have only selected some representative tests from dozens of different load tests, but looking at all the test results, switching the Raspberry Pi OS to the 64-bit version has improved its average performance by about 48%.
Users who want to see all the test data can visit the Phoronix official website (https://www.phoronix.com/scan.php?page=article&item=raspberrypi-32bit-64bit).
Compared to other Linux distributions, Raspberry Pi OS has only now released its 64-bit system, which is indeed a bit late. However, based on the test results, the significant improvement of the 64-bit system makes the wait worthwhile. At this point, you should find no reason to refuse to upgrade to the 64-bit system.
1. Why is there so much industry attention on a new processor core?
2. The second issue of “Microcontroller and Embedded System Applications” electronic journal for 2022 is out!
3. Chip price increase: Inevitable!
4. “In-depth Understanding of RISC-V Program Development” courseware and video courses are now online~
5. Don’t regret it! 18 practical experiences in embedded C summarized by industry veterans
6. Be cautious when converting hardware to software!
Disclaimer: This article is a network reprint, and the copyright belongs to the original author. If there are copyright issues, please contact us. We will confirm the copyright based on the copyright certificate you provide and pay the remuneration or delete the content accordingly.
