On October 29, 2020, AMD officially released a technology called “智能显存访问” in Chinese, which translates to “Smart Access Memory” in English, abbreviated as “SAM.”
In traditional architectures, due to various limitations imposed by the operating system and hardware design, processors typically can only access a portion of the graphics card’s memory at a time, often based on a preset value like 256MB. In such cases, even if the graphics card has more memory available, the processor cannot efficiently utilize the portion beyond the limit, leading to some negative impact on performance.
However, the Smart Access Memory technology breaks this limitation. By optimizing hardware design, and BIOS settings, and closely collaborating with the Windows operating system, Ryzen series processors can directly access all the memory available on the Radeon RX 6000 series and above graphics cards. This allows the processor to read more data at once, significantly improving data processing efficiency and ultimately graphics performance. The principle is similar to NVIDIA’s Resizable BAR technology.
AMD claims that enabling Smart Access Memory can result in a performance boost of 10% to 20% in gaming performance, with the exact improvement depending on various factors, making it quite attractive.
Now, four years have passed, and this article will use a configuration consisting of a Ryzen 7 5800X3D paired with a Radeon RX 7900 GRE to verify the actual effects of Smart Access Memory technology. Four years is a considerable amount of time, providing ample opportunity for AMD to refine and optimize the technology. Testing its real-world effectiveness now should provide a more objective assessment.
01
Test Platform Configuration
The main configuration is as follows:
- Processor: Ryzen 7 5800X3D
- Motherboard: ASUS TUF GAMING B550M-PLUS
- Memory: 16GB DDR4 3600MHz
- Graphics Card: Radeon RX 7900 GRE
- Storage: Samsung 850 EVO SSD + Seagate 2TB HDD
- Operating System: Windows 10 Pro
02
Gaming Performance Test about Smart Memory Access
The following seven games will be tested, all running at 1440P with the highest graphics settings. In the series of images below, the left side represents the platform with SAM disabled, while the right side represents the platform with SAM enabled.
① The first game is “Mindkiller 2”.
With SAM disabled
- Average Frame: 74 FPS
- Minimum Instantaneous frame rate: 64 FPS
- Frame time: 13.1ms
- CPU usage: 22%
- CPU power consumption: 51W
- CPU Temperature: 61 degrees
- Memory usage: 7.3GB
- GPU usage: 100%
- VRAM usage: 8.6GB
- GPU power consumption: 260W
- GPU temperature: 53 degrees
With SAM enabled
- Average Frame: 75 FPS
- Minimum Instantaneous frame rate: 63 FPS
- Frame time: 11.8ms
- CPU usage: 21%
- CPU power consumption: 55W
- CPU Temperature: 62 degrees
- Memory usage: 7.1GB
- GPU usage: 100%
- VRAM usage: 9.3GB
- GPU power consumption: 264W
- GPU temperature: 54 degrees
In this game, with SAM enabled, CPU usage is 1% lower, memory usage is 0.2GB lower, GPU usage is the same, and VRAM usage is 0.7GB higher compared to SAM disabled.
In terms of average frame rate and minimum instantaneous frame rate, both platforms perform similarly, with a slightly lower frame time with SAM enabled. CPU and GPU power consumption and temperature are also similar, with no significant differences overall.
② The second game is “Cyberpunk 2077”.
With SAM disabled
- Average Frame: 91 FPS
- Minimum Instantaneous frame rate: 78 FPS
- Frame time: 10.5ms
- CPU usage: 34%
- CPU power consumption: 60W
- CPU Temperature: 65 degrees
- Memory usage: 8.5GB
- GPU usage: 100%
- VRAM usage: 6.5GB
- GPU power consumption: 262W
- GPU temperature: 54 degrees
With SAM enabled
- Average Frame: 96 FPS
- Minimum Instantaneous frame rate: 79 FPS
- Frame time: 10.2ms
- CPU usage: 36%
- CPU power consumption: 61W
- CPU Temperature: 63 degrees
- Memory usage: 9.6GB
- GPU usage: 100%
- VRAM usage: 8.6GB
- GPU power consumption: 260W
- GPU temperature: 53 degrees
In this game, with SAM enabled, CPU usage is 2% higher, memory usage is 1.1GB higher, GPU usage is the same, and VRAM usage is 1.9GB higher compared to SAM disabled.
In terms of average frame rate, SAM enabled performs 5 FPS higher than SAM disabled, with similar minimum instantaneous frame rate and frame time, resulting in no significant advantage overall.
③ The third game is “Resident Evil 4 Remake”.
With SAM disabled
- Average Frame: 119 FPS
- Minimum Instantaneous frame rate: 111 FPS
- Frame time: 8.3ms
- CPU usage: 17%
- CPU power consumption: 50W
- CPU Temperature: 58 degrees
- Memory usage: 8.0GB
- GPU usage: 100%
- VRAM usage: 8.6GB
- GPU power consumption: 265W
- GPU temperature: 55 degrees
With SAM enabled
- Average Frame: 122 FPS
- Minimum Instantaneous frame rate: 112 FPS
- Frame time: 8.3ms
- CPU usage: 16%
- CPU power consumption: 50W
- CPU Temperature: 57 degrees
- Memory usage: 8.5GB
- GPU usage: 100%
- VRAM usage: 9.5GB
- GPU power consumption: 264W
- GPU temperature: 55 degrees
In this game, with SAM enabled, CPU usage is 1% lower, memory usage is 0.5GB higher, GPU usage is the same, and VRAM usage is 0.9GB higher compared to SAM disabled.
In terms of average frame rate, SAM enabled performs 3 FPS higher than SAM disabled, with similar minimum instantaneous frame rate and frame time, resulting in no significant advantage overall.
④ The fourth game is “Horizon Forbidden West”.
With SAM disabled
- Average Frame: 87 FPS
- Minimum Instantaneous frame rate: 78 FPS
- Frame time: 10.9ms
- CPU usage: 26%
- CPU power consumption: 55W
- CPU Temperature: 59 degrees
- Memory usage: 9.5GB
- GPU usage: 100%
- VRAM usage: 6.2GB
- GPU power consumption: 265W
- GPU temperature: 54 degrees
With SAM enabled
- Average Frame: 91 FPS
- Minimum Instantaneous frame rate: 82 FPS
- Frame time: 10.6ms
- CPU usage: 24%
- CPU power consumption: 57W
- CPU Temperature: 59 degrees
- Memory usage: 10.4GB
- GPU usage: 100%
- VRAM usage: 7.7GB
- GPU power consumption: 265W
- GPU temperature: 55 degrees
In this game, with SAM enabled, CPU usage is 2% lower, memory usage is 0.9GB higher, GPU usage is the same, and VRAM usage is 1.5GB higher compared to SAM disabled.
In terms of average frame rate, SAM enabled performs 4 FPS higher than SAM disabled, with a slightly higher minimum instantaneous frame rate, and similar frame time, resulting in a very slight advantage overall.
⑤ The fifth game is “Avatar: Pandora Border”.
With SAM disabled
- Average Frame: 53 FPS
- Minimum Instantaneous frame rate: 43 FPS
- Frame time: 18.2ms
- CPU usage: 32%
- CPU power consumption: 58W
- CPU Temperature: 61 degrees
- Memory usage: 13.0GB
- GPU usage: 100%
- VRAM usage: 11.8GB
- GPU power consumption: 265W
- GPU temperature: 54 degrees
With SAM enabled
- Average Frame: 53 FPS
- Minimum Instantaneous frame rate: 44 FPS
- Frame time: 17.7ms
- CPU usage: 34%
- CPU power consumption: 61W
- CPU Temperature: 59 degrees
- Memory usage: 12.3GB
- GPU usage: 100%
- VRAM usage: 12.4GB
- GPU power consumption: 265W
- GPU temperature: 53 degrees
In this game, with SAM enabled, CPU usage is 2% higher, memory usage is 0.7GB lower, GPU usage is the same, and VRAM usage is 0.6GB higher compared to SAM disabled.
In terms of average frame rate and minimum instantaneous frame rate, both platforms perform identically, with negligible differences in frame time, resulting in an overall tie.
⑥ The sixth game is “Hogwarts: Legacy”.
With SAM disabled
- Average Frame: 96 FPS
- Minimum Instantaneous frame rate: 79 FPS
- Frame time: 11.5ms
- CPU usage: 24%
- CPU power consumption: 55W
- CPU Temperature: 60 degrees
- Memory usage: 14.2GB
- GPU usage: 100%
- VRAM usage: 8.9GB
- GPU power consumption: 264W
- GPU temperature: 55 degrees
With SAM enabled
- Average Frame: 100 FPS
- Minimum Instantaneous frame rate: 82 FPS
- Frame time: 10.5ms
- CPU usage: 23%
- CPU power consumption: 56W
- CPU Temperature: 60 degrees
- Memory usage: 13.9GB
- GPU usage: 100%
- VRAM usage: 9.2GB
- GPU power consumption: 265W
- GPU temperature: 55 degrees
In this game, with SAM enabled, CPU usage is 1% lower, memory usage is 0.3GB lower, GPU usage is the same, and VRAM usage is 0.3GB higher compared to SAM disabled.
In terms of average frame rate, SAM enabled performs 4 FPS higher than SAM disabled, with a slightly higher minimum instantaneous frame rate and slightly lower frame time, resulting in a very slight advantage.
⑦ The seventh game is “The Last of Us Part I”, a resource-intensive game with significant implications for comparison results.
With SAM disabled
- Average Frame: 91 FPS
- Minimum Instantaneous frame rate: 79 FPS
- Frame time: 12.1ms
- CPU usage: 56%
- CPU power consumption: 73W
- CPU Temperature: 68 degrees
- Memory usage: 11.8GB
- GPU usage: 100%
- VRAM usage: 9.8GB
- GPU power consumption: 252W
- GPU temperature: 55 degrees
With SAM enabled
- Average Frame: 103 FPS
- Minimum Instantaneous frame rate: 86 FPS
- Frame time: 10.6ms
- CPU usage: 69%
- CPU power consumption: 84W
- CPU Temperature: 72 degrees
- Memory usage: 10.1GB
- GPU usage: 100%
- VRAM usage: 11.9GB
- GPU power consumption: 238W
- GPU temperature: 54 degrees
In this game, with SAM enabled, CPU usage is 13% higher, memory usage is 1.7GB lower, GPU usage is the same, and VRAM usage is 1.1GB higher compared to SAM disabled.
In terms of average frame rate, SAM enabled performs 12 FPS higher than SAM disabled, with a 7 FPS higher minimum instantaneous frame rate and slightly lower frame time, resulting in a slightly more noticeable advantage.
03
In Conclusion
The comparison results are quite awkward. After enabling Smart Access Memory (SAM), the overall gaming performance doesn’t show significant improvement compared to when SAM is disabled. There are only slight advantages in a few games, making the overall performance somewhat underwhelming.
Nevertheless, enabling SAM does have some effect. The VRAM usage is noticeably higher when SAM is enabled compared to when it’s disabled. However, this advantage of higher VRAM usage doesn’t necessarily translate into noticeable improvements in gaming performance. Nonetheless, it would be premature to conclude that SAM is entirely ineffective based on this observation.
Therefore, the conclusion of this article is as follows: with the configuration of Ryzen 7 5800X3D paired with Radeon RX 7900 GRE, under 1440P highest graphics settings, enabling SAM doesn’t lead to a significant improvement in gaming performance compared to not enabling it.
From this, it’s evident that conducting a single test is insufficient to make a fair comparison of the actual effectiveness of SAM. It’s necessary to arrange multiple combinations of processors and graphics cards and conduct multiple rounds of comparison tests under higher graphics settings to arrive at a more objective conclusion.
Disclaimer: This article is created by the original author. The content of the article represents their personal opinions. Our reposting is for sharing and discussion purposes only and does not imply our endorsement or agreement. If you have any objections, please contact us through the provided channels.
