Some Radeon owners may have a background driver component writing to their SSD around the clock, and AMD has acknowledged it is looking into the problem.
The component is AMD's External Events Utility (EEU), which installs alongside Radeon drivers. One user said their drive's reported health fell 1% a month for 16 months. Another reported sustained idle writes of about 5MB/s. Neither claim has been independently confirmed.
On October 21, a company representative wrote on AMD's community forum that it was "looking into it with the aim to resolve it in a future driver release." A second representative, AMAD Vik, later apologized on Reddit and said it is "hopefully fixed in the next one." That October 21 reply predates the November 8 Reddit post from user darklinkpower that put the reports back in front of people, so the issue was evidently known before that thread. AMD's October 30 release notes for Adrenalin 26.20.2 do not mention it, and no fix date has been published.
The arithmetic is large. At 5MB/s, a drive takes about 18GB of host writes per hour, 432GB a day and close to 13TB a month. But host writes are what the PC sends to the drive. SSD controllers use caching and wear-leveling, so how much of it becomes actual wear on the NAND flash is unknown.
The cause is also unclear. Reports include Adrenalin 26.19.2 and older releases, so the latest update may not be the trigger.
To check your own PC, press Win + R, type resmon, open the Disk tab, sort by write activity and look for sustained writes from atieclxx.exe or atiesrxx.exe. Two workarounds have been reported: stopping or disabling the AMD EEU service in services.msc, or turning off Resizable BAR, which can improve gaming performance.
The EEU is not a new suspect. In November 2025, a Redditor found that AMD chipset driver software kept updating a log file whenever on-screen elements were moved or resized, which raised questions about whether stopping the service would affect FreeSync. That was activity tied to user input. The new reports describe writes that continue while the machine sits idle, and both workarounds carry trade-offs, which is a reason to measure first and disable second.