Dynamic random access memory (DRAM) overclocking has existed for decades. Initially, companies like Corsair chose memory IC that can work safely at high frequency and/or low latency. Then, it establishes a module to ensure that it works at a specific frequency, timing and voltage, but these settings need to be manually entered into the BIOS. To simplify things, Intel introduced the Extreme Memory Profile (XMP) in 2007-the settings stored in the module's serial presence detection (SPD) chip can be activated on supported platforms-and now almost all fever-level memories have XMP profiles.
However, the guarantee/recommendation settings and XMP configuration file settings are usually relatively conservative to maximize compatibility with various systems. At the same time, each computer is unique, so die-hard enthusiasts tend to decide for themselves the maximum data transmission rate and the minimum delay that their machines can achieve. However, this is time-consuming and requires knowledge, experience and skills.
AMD's automatic memory overclocking technology tries to make ordinary performance enthusiasts have the ability to adjust the maximum performance of their memory modules and CPU like professionals, but it does not require a lot of manual testing. In addition, AMD's automatic memory overclocking should also allow the owner of the cheap memory module set by JEDEC standard in SPD to overclock it to the limit. Performance-oriented enthusiasts who tend to use the best memory module with XMP3.0 configuration files will also be able to maximize performance and reach a level that their construction can handle without a lot of manual work.
AMD's patent US20220 155982 wrote: "The setting of the memory module can be configured according to the configuration file of the supplier or the specifications input by the user". These configuration files are usually generated and tested by a system configuration different from the user system. In addition, the specifications entered by users may be limited by the margins determined and tested using these different system configurations. "
AMD's automatic memory overclocking technology needs to determine the frequency setting of overclocked memory, determine the most active time setting through the memory stability test, and generate a configuration file including the frequency and delay setting of overclocked memory. The generated configuration file will eventually be loaded by the program. Interestingly, the patent does not touch the memory voltage, which means that the module must maintain the SPD voltage (that is, the voltage from the SPD configuration file), or must manually set the voltage in UEFIBIOS, or use a third-party application.
The patent was published on May 9, 2022/KLOC-0, and we can only guess when AMD's plan will be realized. However, we can safely speculate that AMD's next-generation Ryzen7000 series processor will play a role once it comes out.