The new Nvidia GeForce 20-series GPUs are going to start shipping from next week and each of this next-gen GPU is going to feature a new overclocking system named Scanner.
Overclocking is not a simple process by any means. It’s more of a hit and trial process, there’s a lot of different parameters to take care of. Increasing the clock speed requires you to run some intensive workloads to check if the system works, adjusting the GPU voltage may provide a little more stability to the system. If the clock speed is increased too high, the system may lock up and crash altogether.
Finding just the correct and most optimal settings requires lots of rebooting and adjustments. Even after finding the correct settings you can’t be sure, as not all workloads use all parts of the chip. A clock speed which is working fine for one software may make the PC unstable for others.
This is where Scanner comes into the picture. Scanner is the new overclocking system for the new GeForce 20-series cards and it’s going to make the overclocking process more automated and reliable. It will run incrementally and increase the clock speed and GPU voltage to make a profile of a specific card’s capabilities, it’ll also test GPU’s computing ability at each clock speed to make sure it’s stable. Most of the times if a GPU is running at a slightly faster clock speed than it can support, there will only be arithmetic errors rather than complete crashes. So by making these small adjustments, Scanner can make sure that the GPU runs smoothly without any crashes or reboots, and without any human involvement. This process takes just about 20 minutes.
One thing to note is that Scanner only works for GPU overclocking. For memory overclocking you still need to do it the old way. You have the option to use the same traditional methods for the GPU itself too.
Scanner isn’t an application, instead, it’s an API. Support for Scanner will be integrated into different overclocking apps built by different GPU vendors. Currently, only the new 20-series cards are supported by Scanner but support for older cards can be expected to ship in the future.
This is surely going to make the overclocking process a lot easier. Nvidia is known for inducing hot-spots and thermal instability in its cards, so we can expect the Scanner’s ability to determine the optimal speed and voltage to be much more accurate than any other overclocking method. Since it’s all automated, Scanner is able to make the micro changes in the speed any number of times till it finds the optimal value, which is something a human would find to be a very tiring and boring job.
It’s a nice approach to overclocking by Nvidia for its new GeForce 20-series cards which Nvidia claims are ‘built to overclock’.