Aurora R11, RTX 3080 memory TJunction temps up to 102°C max

I have seen up to 92°C on mine. Aurora R10, RTX 3080.

For the average, try restarting HWinfo when you are gaming. The average is exactly that, the average taken over time.

If you start with 60°C for 20 minutes followed by 100°C for 10 minutes, your average will not be much higher than 70°C.

Try the average again when you start at 90°C, and I bet it will not drop but increase over time to mid 90°C.

102°C is rather high. I think it starts to throttle at around 95°C.

Per this article it starts to throttle at 110°C.

On what games you see 92°C max? I see especially these high temps (102°C) on Battlefield 4 online, on other games I see lower values.

The TJunction reported temperature is the highest of all reported ones. It means all the other memory chips have a TJunction equal or less than the reported single value.

The reason why it get's up there is because of the memory chip temperature under load, and the VRM temperature of the close by VRM's that add extra heat to the already hot memory chips.

Since the DELL version of the 3080/3090 is shorter than the reference design, it can fit inside the compact Aurora case. Since that case is not know for it's superb air flow, I can see some of these reported card failures in threads on this board be attributed to these high temperatures, leading to permanent damage.

The memory chips are Micron, and Micron stated that 110°C is the max before damage can occur. So if it throttles to late, it will go past 110°C before it has a chance to cool down and can cause permanent damage.

I have resorted to running the TJunction readout on my second monitor on a permanent basis, so I can keep an eye on it at all times.

I have never see it rise above 92°C on my card. This is playing WoW Classic at maxed out settings and unlimited frame rate, running 3Dmark benches etc. At that test setup the card boost to about 2050 Mhz and the power is at 100% as per afterburner. That should technically stress the card to the max.

Normally I play WoW classic maxed out but limited at 60 FPS, and at those settings I reach around 62°C on the TJunction. But that is because the card barely has to work.

Once it does have to perform work, it does go up to 92°C max most of the time. It's possible this varies a bit between game to game because of the way the game uses 3D and because of the various 3d settings in the game that affect required video card load.. I don't have that game to test on mine, so I can't give you a direct comparison.

Some of your numbers are also much different from mine.

Your core voltages are lower, meaning you will draw more current for the same power draw.

Is it running stock, or overclocked? Mine is OC at 135/100.

I also use a custom fan curve and keep my temperature lower under load because of that.

No these values are on stock settings and without custom fan curves...

I can see the Tjunction temps with Hwinfo64 and can put them in the tray while on windows desktop, but I don't know how to put them on an overlay to check while playing. It would be interesting to see them...

With Msi Afterburner I cannot do this... You only see it in the second monitor tray? I have a few more monitors but I cannot fit it in my current desk...

I have overclocked with MSI Afterberner up to +145Mhz (Core) +545Mhz (Memory) and the Tjunction max temps stay up to 102 Degrees Celsius max while gaming. But I don't see alot of improvements just up to 5fps more here and there and I usualy leave it at stock settings except sometimes when benchmarking...

I put the sensor readings from HWInfo on my second monitor, together with the graphs from afterburner.

This way I can monitor everything just by glancing over to that screen.

I have been playing with custom fan curves, and splitting up the 2 fans control wise so each fan has it's own curve. I was able to shave off an additional 2-4 degrees Celsius by having fan #2 run about 500 RPM faster than fan #1.

I am not sure which fan is what, but ramping up fan #2 seems to cool down the VRM section better.

I also noticed ramping up fan #2 results in significantly less noise than ramping up fan #1. I am assuming this may have to do with airflow and the resulting noise. So basically I can run fan #2 much higher than fan #1 without generating any additional noise. That's always a bonus.

I now idle the GPU at 1710 core clock, 31 C GPU, and 54-56 C Tjunction. Power draw is 27%, fan #1 1870 and fan #2 2375. 

Running several 3dmark benches in a row it goes up to around 60 GPU and 90 Tjunction.

While I am still not happy with the 90 on the Tjunction, I think that is about the best I will get this without resorting to more drastic measures like moving everything over into a different case with more direct airflow on the video card.

I wish there was an option to put a side fan in the Aurora case. There are 2 vents on the side of the case, and they would be perfect for extra GPU cooling if you could attach fans to them.. The bottom one is at the perfect location to blow directly onto the video card.

I think it would greatly help cooling down the video card, and would cost maybe $20.

To add that the 2 fans I am talking about above are the 2 fans on the video card, not the case fans.

I downloaded HWinfo yesterday to see what my R11 3090 vram temps were playing DCS / VR and letting it free fly as I observed this wonderful program. I also had MSI/AF running and did not oc the card. I noticed about 10 mins. The VRam temp was around 86c-90c while the gpu temp @70-75c so I don’t think that’s to bad I guess. So how I use my fans both gpu and case is that in AWCC thermals settings I have been using the the offset slider and adjusting the Top fan at 45% and the Front fan at 35% Msi/AF I have been running both gpu fans at 70% manual, no curve no auto but full 70% while running the game. I noticed that my vram temp highest was 82c and never got over 85c. The gpu temps stayed around 70-73c. As I’m new to this kind of troubleshooting I think I helped ease the temps alittle. 

There's also an additional hotspot value you can monitor with GPU-Z. It is not the tjunction, but the GPU hot spot sensor.

Your GPU hotspot might be 10-15 higher than the reported GPU temperature because they are 2 different sensors each measuring a different parameter.

I played around more with the fan controls and am now below 85 on tjunction.

Great advice thank you. I'm also think about adding 2 new fans for the front and I take it that the Corsair ML120 Pro is the most popular ones to great with these Alienware Machines. Also a splitter, will there but any conflicts with the bios adding these new fans from Corsair? 

Thanks !

20 minutes of CyberPunk 2077 and I am at 82 degrees, side case open, with a water-cooled RTX 3080. I will do a longer test tonight, with the side doors closed, and see where that gets to.

Thats not bad really. 

Case closed, 10900KF, 3080 RTX, 30 minutes of Cyberpunk. Maximum GPU Memory Junction Temperature: 86.0 degrees celcius. This is in a custom loop watercooled Alienware Aurora R11 setup, so not the normal air-cooled setup. Same card, just placed into a water-block, in a custom internal loop with 360mm of Radiators inside the case. The heatplate on the back of the GPU gets toasty hot - to the point where I zip-tied my water-tubing away from touching it to make sure I don't melt any of the plastic tubes. But I guess it is sitting about 15 degrees cooler than an air cooled card, so I will take that.

For what it is worth, the 10900KF averaged 74 degrees for this run, with a max of 89 degrees. It is a small case, so heat doesn't dissipate easily!