Youtuber overclocks SATA SSD, but without much practical benefit
Overclocking is done with a CPU or a GPU. However, YouTuber Gabriel Ferraz looked a little further and tried to overclock an SSD. That only partially worked; the SSD died at the end of the process. Moreover, overclocking appears to make little sense in practice.
Ferraz shows in a YouTube video that he managed to increase the maximum clock speed of a SATA SSD from the unknown brand RZX. He does not explain exactly how he does that. He does warn that there are ‘safety reasons’ not to overclock SSDs and says there is a good chance that the device will not survive. Which was also the case with him.
The video shows Ferraz taking a 2.5″ SATA SSD for his project. He uses a SATA drive and not an NVMe drive, because the SSDs Ferraz had at his disposal were already at their maximum capacity. clock frequency. The RZX SSD contains an SM2259XT2 controller from Silicon Motion and TLC NAND from Kioxia.
The controller in question has a single one ARC-32bitcore, which supports clock speeds up to 550MHz. However, in the RZX SSD, the clock speed has been reduced to 425MHz by default. Kioxia’s NAND memory can officially run at a maximum of 400MHz, but has been clocked down to 193.75MHz in the RZX SSD. Ferraz had to put the SSD in safe mode by shorting two contact points on the PCB. He was then able to overclock the controller and the flash memory of the SSD with SMI’s MPtool, although he does not explain exactly how he does this in the video.
The results are not very surprising, says Ferraz. He managed to increase the maximum frequency of the controller to 500MHz, an increase of about eighteen percent and slightly lower than the maximum speed of 550MHz that Silicon Motion advertises. The flash memory frequency was also increased to Kioxia’s advertised 400MHz. However, the sequential read speeds remained virtually the same and the random read speeds went from 134 to 170MB/s. The random write speeds actually dropped from 155 to 140MB/s.
In practice, according to Ferraz, this did not seem to be so bad. In various benchmark tools the number of points increased slightly, but not much. The loading times of games and heavy Adobe Premiere projects also appeared to hardly improve in practice. At the end of the journey, it also appears that the SSD consumes a lot more energy: an average of about 63 percent according to Ferraz. The SSD also stops working; the nand memory breaks down. Ferraz gives no specific explanation for this.