GPUsNews

NVIDIA RTX 30 Series GPUs Won’t Crash Upon Unstable Overclocking

Memory overclocking on graphics cards just became more problematic, thanks to a new Error Detection and Replay (EDR) technology. Traditionally, when you overclocked the GPU memory past the stable limits, you’d start seeing artifacts and eventually result in crashes. With the new updated version of EDR used in NVIDIA’s RTX 30 series GPUs, instead of artifacts and crashes, you’ll start seeing a reduction in performance once you reach the limits of your GPU memory.

Source: Andreas Schilling

This is because the new EDR tech upon detecting data transmission errors retires the faulty transmission and/or replays it till it’s successful. These retired/replayed transmissions reduce the memory bandwidth, thereby reducing the performance upon an unstable memory overclock.

With EDR, users no longer have to overclock their GPU’s memory until it crashes – as soon as plateauing performance is observed, the user has reached the limits of the memory and should stop increasing the clock frequency.

This is one of the reasons early samples of the RTX 3080 given to reviewers seem to handle memory overclocking so easily. In reality, there’s a good chance that you’ve already overclocked the memory well past the stable limits, but because of EDR, you’ll no longer know as crashes and artifacts will only be produced in extreme scenarios.

Areej

Computer Engineering dropout (3 years), writer, journalist, and amateur poet. I started my first technology blog, Techquila while in college to address my hardware passion. Although largely successful, it was a classic example of too many people trying out multiple different things but getting nothing done. Left in late 2019 and been working on Hardware Times ever since.
Back to top button