EVGA explains how Amazon’s MMO bricked 24 GeForce RTX 3090s

An Analysis Of Dead EVGA GeForce RTX 3090 Cards That Failed When Played From Amazon New world game indicates that a rare solder issue limited to a small batch of cards is responsible, a company spokesperson told PCWorld.

EVGA said it received around two dozen dead GeForce RTX 3090 cards that allegedly failed to play the game. New world beta. All of the boards were earlier production boards made in 2020. Based on x-ray analysis, they appear to have a “poor finish” when soldering around the board’s MOSFET circuits that feed the impacted boards.

Players entered a minor panic in early July when reports started to arrive that Amazon’s popular beta New world MMO was killing graphics cards. At the time, the bulk of the reports revolved around EVGA’s GeForce RTX 3090 GPUs, although there were unconfirmed reports of dead cards from different manufacturers and different card brands, including both Radeon and older GeForce cards.

EVGA said all failures reported to the company were limited to GeForce RTX 3090 cards. The company said it immediately shipped replacement cards to affected players rather than waiting for defective cards to return.

The company declined to say how many GeForce RTX 3090 cards it sold, but called this small batch significantly less than 1% of the total.

Amazon

Customers have told EVGA that the maps are dead in the in-game menu system or while the game is loading. No customer has reported any issues to EVGA since Amazon added frame rate limiters to the game. .

Both EVGA and Nvidia worked with Amazon Games to get the version of the game that broke the cards for further testing. EVGA said it couldn’t reproduce the issue, but was working on adding a method to filter out the particular power profile created by the beta.

The New world The beta in question did not implement a frame rate limiter in the menu system, which many released games include. This could suddenly change a GPU from an in-game frame rate of 100 fps to a roar at 800 fps. It’s like driving a car at full throttle up a hill and accidentally putting the car in neutral. If you don’t lift your foot, the motor spins in its red line.

Not the fan controller either

Early theories that EVGA’s cooling system was to blame are also incorrect, the spokesperson said.

“In no way is it related to the fan controller,” he said.

Brad Chacos / IDG

In this theory, EVGA’s temperature monitoring hardware was accused of not being able to keep up with the load of the 30-series RTX GPUs, causing gaps that triggered the brickwork.

EVGA said this was not the case, however. The fan controller microcontroller could indeed give the impression that it was not working under the extreme oscillations of New world, but EVGA said the issue was with noise on the i2c bus, forcing popular third-party monitoring tools like HWInfo and GPU-Z to incorrectly report the noise as a fan controller failure. Its own EVGA Precision X1 tool would filter out noise and report it properly, EVGA officials said.

The company has since released an update to the microcontroller that, when paired with updated versions of third-party tools, will correctly show that the fan controller is working properly. To get the latest microcontroller update, EVGA said that customers should download the latest version of EVGA Precision X1, which should indicate whether a microcontroller update is required for the GPU or not.