Water Cooled: GPUs and CPU gradually warming up
Water Cooled: GPUs and CPU gradually warming up
Temp probe has been received.
Ambient temperature - 28°C
Idle temperature - 29-30°C (CPU and GPUs)
After running Dying Light on ultra settings for 90 minutes:
CPU - 74°C
GPU1 - 61°C
GPU2 - 58°C
Water temperature (reservoir) - 49.9°C
The Noctua fans did help with the GPU temperatures, but the CPU is still behaving strangely, and the water temp of 49.9°C is quite high.
My top res fans have an SP of 1.8, so I might consider upgrading them a bit more, though I’m not sure what else to do right now.
Pump speed is currently set to 2 (out of 5), but increasing it doesn’t seem to help much.
Appreciate all the suggestions so far. Any further ideas?
What are your current RPMs? Are your radiators operating at high or low FPI settings? It’s likely there’s a restriction somewhere in your loop. I previously had a single 480mm top-mounted radiator in my 800D before upgrading. Before the overclock, I was cooling a 375W TDP Radeon 6990 and an i7 3930k with 130W TDP. I used four 38mm fans on the radiator. My CPU temperatures stayed below 60 under normal load or just over 70 during Prime95. The 6990 never exceeded 54°C per GPU core while gaming. I was using a MCP655-B pump that consistently runs at 4500 rpm. My temperatures were acceptable, but the system was quite noisy. I typically ran the radiator fans around 2500rpm or so.
Could you consider setting up your loop to work alongside a temperature probe and fans without overloading the rest of your system? I’m interested in observing what happens with the coolant under no-load conditions.
My initial reaction is in agreement with 1LiquidPC; you lack sufficient radiator cooling area to complete the task.
However, considering what you currently have running, the optimal improvement would be to utilize both radiators as exhausts, adjust your rear exhaust fan to function as an intake, and possibly install higher CFM fans for both rear and front intakes. This would ensure fresh air is supplied into the case, allowing the radiators to expel heat effectively.
In your setup, you require more incoming air than is being expelled, creating a slight positive pressure benefit.
Your image indicates the bottom radiator absorbs heat from the case, which causes temperatures to rise over time due to increased loop load. Consequently, the top radiator ends up with preheated air.
Having multiple GPUs significantly alters the situation, and it seems you are now experiencing these effects firsthand.
I really question whether he lacks sufficient dissipation space because he only needs to manage around 600W with the CPU and both GPUs at full capacity. The heated air seems to be doing little more than its job, and even if it slightly influences temperatures, it wouldn't lead to noticeable heat buildup.
Mr Kagouris:
I really question that he doesn't have sufficient dissipation space because he only needs to manage around 600W with the CPU and both GPUs at full capacity. Or that the warm air is influencing things in any significant way. He has two front intakes drawing in far more cool air than his radiator is expelling, and even if the heated air did affect temperatures, it would do so gradually rather than causing heat buildup.
I disagree!
That's why I shared what I did!
since in practice, two standard 360mm radiators are more than sufficient for a CPU and two GPUs. having three 360mm units improves performance but two will still function.
I'm happy to see similar thoughts about the OP possibly lacking enough radiators. I often notice discussions about using 120mm of radiator per device, but I don't fully support this rigid approach. For instance, comparing an 88W CPU to a 130W one, or considering a 250W TDP GPU as mild for heat management. My reasoning leans toward either insufficient radiator capacity or poor heat dissipation, which is also affecting water temperatures and causing noise from the fans. This was an issue I faced with my previous setup, where I had to manage high noise levels to maintain desired temperatures.
I'm not certain about the origin of the idea that '120mm per component is a good rule of thumb,' but it's not a precise or optimal guideline because many factors influence loop delta. Also, determining the exact size and thickness, flow rate, fan selection, and FPI can be tricky. Many people assume a single 120mm closed loop cooler will handle any heat source, but in practice, especially for high-power setups like overclocked CPUs, it often falls short. The cooling performance is likely limited by the radiators' inability to manage the thermal load effectively.