RTX 3050 backplate is very hot
Friends, a month ago I bought new, with warranty MSI GeForce RTX3050 Gaming X 8G through Yandex Market. Came across with Hynix memory. Got frustrated, everyone remembers those failures on the 3060ti. Calculated in different stress test, max core 62-64*C, hotspot up to 92-95*C, no sensors on memory. Vascrlate, near PCI-e key, very hot at this time, fans spinning at 30-35%. Also Aerocool Eco 500 power supply starts humming. There it is 456 watt on 12 volt line. Yes, the card is connected to the 8pin adapter w/o 2molex-8pin. The first 2 or 3 days while it was gone, it worked on 6 pins at all. I am using a different card for now. System: V660M-E, DDR4 2x8 2400MHz, i3-12100f, airy case, ATX size, power supply on top. Does anyone have something like this, or am I "lucky" as always?
This big difference between average temp and hotspot usually indicates a poorly serviced card.
Post a direct link to the product where you bought it, what kind of store even sent you to see.
If the cooling fails, it's like this forever. The card is under warranty, you can't open it up. So there's an underwalt option
In MSI Afterburner, increase the RPM of the cooler, I have built a direct relation, temperature 60 - RPM 60, temperature 70 - RPM 70.
Pavel Rally, that's the thing - at these temperatures, the fan speeds stay the same, no matter how much I adjust Afterburner settings.
Are you trying to set the RPM through the Nevidia driver? Even though I have AMD, but once there was an episode when I couldn't change the RPM through MSI Afterburner, so I had to use the driver.
Or you can replace the fans by looking around on eBay somewhere.Loken, good time. Here's the linkhttp://market.yandex.ru/product--videokarta-msi-geforce-rtx-3050-gaming-x-8g/1498188508?sku=101556125820&resale_goods=resale_new&cpa=1 Today (hh 2 weeks of inactivity) I ran Furmark 1920x1080 FHD again for 10min. Core temp is the same and hotspot-75*C. Pisses me off that the metal backplate is hot badly, and if it wasn't there at all, the fate of the 3060Ti. Thought msi is not Zhizhabit, they won't screw it up.
Did you get a new one or a markdown one? Maybe they just sent you some repaired card under the guise of a new one, which was badly serviced, that's why a certain area of the chip is fried.
What you call "core temperature" is the average temperature from all the sensors in the chip. The hotspot is the hottest sensor. 75 for the hotspot is normal, and 85 too. In general, the difference of 10-20 degrees between average and hotspot for nvidia cards is in the normal range. But 65/95 is not really normal. Either the stub's air flow is too weak, increase its rpm, or the thermal paste is unevenly applied, or crooked pressure on the cooler is not common, but crappy service guys can do that.
The backplate in your card is thermal gasket, it's making contact with the printed circuit board, and specifically, with the parts where the memory chips and the graphics chip are, so it's normal that it can get hot
Two more questions for you - 1) you know that you have a piece of crap instead of a power supply?
2) - what kind of case do you have, how many and which fans are installed in it for blowing?
Hi Denis Kyokushin. Sundervoltil at once, to the limit. 0,75. Frequency 1990. Here's the test went, on the 2nd minute the hotspot rises above 70*C, fans turned on, at 30%, then T gradually increases, and the fans are not spinning more than 35%.
Denis, the card is new, change the cartons and lose the warranty. But by Nvidea driver I have not tried, is it possible? Send me some guide, if it's not difficult, I'll learn.
Loken, I don't have a temp meter, only a tactile one. It starts to get hot at the key, as I understand from your thermogram, there. Then spreads all over the back plate, and already tolerable. In terms of performance I like the card, but I'm afraid it will break. I started mansplaining for Hynix. Sellers all put up a template spec from Nvidea ads, without specifying DDR6-DDR6x. I also thought about the lack of power. The adapter wires are cold, the connector itself is tight. Card needs 130W, in GPU-Z at full load: PCI-e Pow 29,7-32,5W, 8pin Pow 75,0-80,7W sum 113,2 W , i.e. the margin is small. it's just no one writes about this problem for 3050, And Hyunix is still an ambush for me.
Forget the Hunix problem. The problems were only with specific models of memory chips and specifically the 3060ti. The 2000 cards also had problems with memory at the start, but already with Microns.
If I were you I'd better change your power supply to something more sensible. Your PSU is just a piece of junk that belongs in some office or garage PC with 15 years old hardware at the most.
Your PSU is just a piece of junk that belongs in some office or garage PC with 15 years old hardware.
I couldn't find such a guide.On AMD there is such a username in the driver, so I thought it should be the same for nevidia:
And AMD Sapphire RX6500XT driver from 20.04.23 has no such feature, or I couldn't find it. Well it's another card from Asus P8 B75-M LE, Xeon i3-1240v2, MEM DDR3 2x8Gb 1600MHz. I even touched it, it's quiet and cool, PCI-e 4.0, my mother supports maximum 3.0, I don't take into account losing a couple of fps. I can not understand why in the game, when changing locations, autosave noticeably darkens the image. As a monitor Toshiba LSD TV 1920x1080 60Gz HDMI port using. With Nvidea 3050 this does not. What do you think?