HWG90

Arrowhead2010
Didn't Amazon patch the issue today and cap the frames in the menu? Going to be hard to reproduce now unless you've still haven't updated.

Oh my... I'm gonna snap. It has nothing to do with the software. I can't say that enough.

The game did NOT cause the problem, the problem was ALWAYS there.

liud21
Jay is just trying to save himself, EVGA probably sent him a email or Jay realize what he might have done and is doing damage control... Those 30 series SUPER cards are coming out soon. And How bad would it be if Jay didn't get one from EVGA to do his video... 😂. This isn't a new issue, early in the year the 3090FTW3 we're getting Red LED of death, and there was a thread on it with over 50 pages... EVGA was silent on the issue, I wouldn't be surprised if they remain so on this issue too...

Elawarai
Still no issue on my RTX 3090 FTW3 UG. 
 
I'm currently in queue screen and the GPU temp, usage and the power drain are not worrying. However, the temp is quite high for a menu screen (65 °C). 

HWG90

degenerate
 
 
Well frankly because I have a Kingpin which is overbuilt for daily gaming and also that the framerate cap seems to mitigate the issue. I'm also of the mind that it's being blown a little out of proportion and it seems somewhat connected to the FTW3 Ultra failure rate that has always seemed to exist. *shrug*

Would've given my left everything to have a Kingpin as an option, choosing beggars or something like that. I didn't really have the option, and it was either a Zotac 3090 or the FTW 3 Ultra ... seems like I made a mistake.

kevinc313

degenerate

kevinc313
Yeah I'd set a frame cap.  
 
Undervolt does not prevent OCP, went into OCP (hard power disable, either psu or gpu OCP) at 1V and 2,040mhz on a 3080 Ti the other day.

Yeah I'll probably end up capping at 80-90fps since I don't expect to be hitting 117fps at 4K based on what I've seen watching streamers with 3090's. Figured the undervolt will just help keep power usage within reason since I'm running the 1kW bios. We'll see what's up when I actually get in-game though.
 

Yeah I'd still set a hard power limit around 450-500W.
 
(btw my ocp was psu, instant full system power shut down)

kevinc313

degenerate
True, forgot I could just set the power slider to 50% and effectively limit power to 500W. Going to fire it up here shortly.

We are praying for you.  lol.  Good luck.

liud21
It's not happening at the massive level that's happening to EVGA cards... Reading one or two dead cards from 3 other different AIB is not the same as reading 20 or so plus from EVGA... Yeah, EVGA will get to the bottom of it, just like they did earlier in the year with the RLED of Death right? RMA, and release revisions until people forget about it, until another Amazon game comes along....

gsrcrxsi
I can’t imagine how much money EVGA has lost RMA-ing all these 3090s. Some people RMA 2-3x. Crazy.

"StarCraft 2 is apparently causing overheating issues on some systems, literally frying graphics cards thanks to the lack of a framerate cap in some areas of the game. Fortunately, Blizzard has issued a temporary fix for the problem until a patch can be rolled out via Battle.net.

According to the problem occurs at the main menu, where the lack of a framerate limit allows hardware to max out, potentially failing permanently.

"Certain screens make your hardware work pretty hard," Blizzard said via its .

"Screens that are light on detail may make your system overheat if cooling is overall insufficient. This is because the game has nothing to do so it is primarily just working on drawing the screen very quickly."

Blizzard's current solution is to tell customers to go to Documents\StarCraft II Beta\variables.txt and add the following lines, replacing the final integers with the desired value.

frameratecap=60
frameratecapglue=30

Some fans have voiced outrage that such a big and obvious bug, one potentially capable of destroying entire systems, slipped through to the release version. As Dailytech points out though, Blizzard isn't entirely to blame and even a maxed out GPU will only overheat that quickly if it already has cooling problems. "

Eisai_Kurosawa

Almighty_Moe_FR
Igor's lab suspects it's not shunts or voltage converter, but the fan control chip that dies.  Thus why predominately an EVGA issues as the other AIBs don't use the iCX3 chips given most of the ICs have self monitoring capabilities built-in.

 
Sounds impossible, but might be true.
I find my 3090 once beyond 72C, GPU fans well suddenly running with the loudest noise I ever heard (under BIOS auto control), at the same time, PX1 will lose the RPM signal from GPU fans, GPU power goes up to 500w. then after GPU utilization and temp drops, RPM reading will come back.
 
Seems those fans control by two different systems?
 

Meguas2
 Software shouldn't be able to kill GPUs, so it's not new world.  GPUs should be able to work at 100% load without their OCP tripping or even worse blowing fuses.  This is going to come down to either Nvidia's OCP not being fast enough to catch spikes or some other flaw imho.  But I won't jump to conclusions as people that know a lot more than I do about this subject will do their investigations and we will find out.

gsrcrxsi
I can’t imagine how much money EVGA has lost RMA-ing all these 3090s. Some people RMA 2-3x. Crazy.