Bitcoin Forum

Are you a newbie who just start his or her crypto mining journey using graphic cards and you aren't sure which GPU temp is the safest and which one is the most dangerous for your GPU all you need is this screenshot I found online, reason why I'm sharing is because it's accurate, other members should testify

https://i.imgur.com/lHpRyvB.jpg

According to this if my core is 46C and memory hits 110C it's perfectly safe LOL

lol perfectly safe

Accurate? It's about as accurate as a long term weather forecast. It's only one person's opinion of
a general guideline.  As far as the advice goes it's not bad but it's no magic pill.

GPU memory temperature has become a very hot issue, pun intended. Oversimplification doesn't
help confused newbies. It's better for a newbie to get unconfused by getting informed.

That is a useless chart. GPUs temps range depending on what GPU it is. Back in the Litecoin mining days I had a bunch of reference GPUs they were all Radeon 6990 all the way to R9 290. Since the reference cooler was always loud AMD designed the operating temperature to be something like 80-90C. And I’ve mined for maybe 5 years straight with those GPUs and never had any issues.

Some of the later GPUs require a lower temp. However you need to understand that in each bios there is a GPU temp kill and when it reaches a certain temp to say a bad fan it will send a signal to the motherboard and the entire rig will power off. It’s been like this for decades with GPUs.

I accept your claims but why did some GPU died when they mine on high temp for a while then? AMD 6990 are designed to operate on high temp but not all GPU are designed this way, I've seen GPU lose its chip connection due to high temp many times, they do work if you can resit the chip with heater or workstation iron, I don't think that chart is useless but thanks anyway

I don't think high temps are what causes the GPUs to form cracks under the solder joints. I think that high temps can cause other components such as capacitors or mosfets to fail however not the GPU chip. I think the reason why the solder cracks is due to the huge heatsink sitting on it and the handling. Too much weight and causes the solder to crack.

That or when doing a thermal repaste job and you pull too hard and cause it to crack. And then with enough heat cycles it will just stop working all of a sudden. Because each time you start the rig, the heat expands and then contracts making the cracks bigger and bigger.

One of my AMD smoked on me and I found out it's high temp, it's been long since I monitor my rig since it's running very well for weeks with no problem, after changing the mosfet that went up in flame the GPU started working again that was how I knew that the GPU was running on high temp and I had to change thermal paste and the pads

I think these days that chart is mostly irrelevant. It is listing Core temps and most GPU algorithms these days are memory hard and memory temps really depends on memory type. the best source for safe temps is to read the datasheet for whatever memory manufacture / type your GPU is utilizing. But 90% of the time when a GPU fails its normally a capacitor or mosfet and is an easy fix. When im looking to buy broken GPUs the listings are almost always the lower end brands like Zotac because they use cheap capacitors and mosfets to hit there price target. The cheaper parts just don't hold up against the 24/7 usage.

Not really, I think there are nearly the same number of core and mem intensive algos.

I had 2 GPU´s running over months with 85°C, but they still run fine :D but most of my gpu´s running between 40°C and 60°C

I easily get worried when ever my GPU runs on 60+ °C because I believe high temp isn't very friendly to GPUs but if the card is AMD for example I won't mind seeing them run at 70°C I think AMD cards can handle higher temp than Nvidia, makes me feel like that's the way they are designed or something

And should we first care about core temp or vram temp ? For cards like 3060 Ti / 3070 we generally don't known the vram temp, we only have the core, so in this case how do you translate the core temp to have an idea of the vram temp, something like vram temps = core temp + 30or40°c ?

I started knowing about vram temp since the first day I bought my first 3070ti I believe it's very important to watch your vram temp if you are using a high end GPU like 3070ti/3080/3090 you don't want vram to die on you

I think that you need to study this issue in more detail, for each model of video card that you want to buy.And such tables, as people in Russia say: "show the average temperature of all patients in the hospital." Hot 3080 or 3090 graphics cards will have different values when compared to 3060 and 3070.