3 ms·
So an update to anyone who's interested This issue caused a low end machine that a reviewer built to shut down: https://www.youtube.com/watch?v=rhjC_8ai7QA htt
by dogma1138 10y ago
So an update to anyone who's interested
This issue caused a low end machine that a reviewer built to shut down:
https://www.youtube.com/watch?v=rhjC_8ai7QA https://www.youtube.com/watch?v=rhjC_8ai7QA
Reported to cause damage to older motherboards:
https://community.amd.com/thread/202410 https://community.amd.com/thread/202410
AMD has published a short message saying that they are looking in ways to "fix" this via the driver.
The drivers have a profile for clock and voltages at any given setting so it looks like these cards are going to be "downclocked" or the voltages going to the memory and GPU adjusted down for given clocks in order to prevent the card from peaking above 150W.
This doesn't look good for AMD to be fair they should recall these cards and redesign the boards.
GPU cards tend to stay away from topping the power consumed from the PCIE bus for good reasons they don't want to bet on the stability of motherboards especially with everything and their mother being interconnected via PCIE these days.
NVIDIA cards usually top at ~30W or so, single power connector AMD cards from the previous gen used to top at 50-60W.
I really don't understand how this passed internal testing or even the design phase, pretty much every engineer would prefer sucking more power directly from the PSU than from the much thinner tracers on the motherboard...
EDIT: Apparently the motherboard tracers aren't the big problem especially on high end motherboards which have a very thick power plane.
The power spike can however damage the PCIE slot itself mainly the pin's as the electromechanical spec for the PCIE sig dictates the exact specification for the pins on the slots even higher end motherboards might be damaged (slot damage like the ones reported over on the AMD forums) if there is too much current draw as the pins are likely to overheat and can either melt down or melt the plastic around them and cause the mechanical connection to fail.
Higher end motherboards however do have a better power phase so they are less likely to shutdown with these cards.
My personal recommendation is if you have purchased the reference design of this card use the 30 day money back policy and send it back.
The driver update from AMD might fix it but it also means that your card would likely suffer performance wise at least in some circumstances.
Wait for a new reference design or for non-reference cards to come out if you are still interested in the RX 480.
You'll get a higher performing card or if AMD decides to gimp all RX 480's via the drivers you could at least safely overclock your non-reference card to the original clocks and higher.