4790K high temps Custom Loop
4790K high temps Custom Loop
Initially, very little load would occur during window startup, yet operations continued. I configured the core temperature to start with Windows, and it reported the highest temperature above 60 minutes without any activity except launching Windows. I also noted these readings while using Windows Media Player Classic and Chrome.
As a general guideline for radiators, I follow a rule of thumb of 120mm per heat source plus one extra for three sources, which brings the total to around 480 degrees. I tend to choose larger radiators, so I selected the 420 model, which claims it can handle over 1400W of heat—far exceeding the 900+ you mentioned.
I also consider high 40s to low 50s temperatures on a pair of 980Tis that are water-cooled and modified for safe operation. This applies after an hour of running Windows, when the coolant reached its operating temperature. This suggests the surrounding air was sufficient to keep the CPU cool, while the GPU temperatures remained relatively stable around 50 degrees.
With a good pump, the coolant absorbs only a small amount of heat from the block. Although it does absorb some heat, if the processor is running at about 65 degrees Celsius, the coolant should remain much cooler. This is supported by the fact that GPUs stayed steady in the low 50s. I was essentially ruling out issues with the cooling system.
I understand that this leads to some wear, so increasing the vcore makes sense. Have you thought about it? If you're okay with it, it will significantly improve your CPU performance and allow you to run at higher speeds with cooler temperatures.
Jake Anderson says he understands the situation and agrees that more vcore could be beneficial. He hasn't done it yet but would consider it if comfortable. He plans to upgrade to a 5775C and prefers lower wattage processors for better performance at lower temperatures. He mentions past experiences with delidding, such as fixing Athlon XPs into MPUs before multicore chips were common, and recalls the challenges of working with single-core chips even with hyperthreading.
I know what it's like to feel uneasy. I spent about three weeks after its release thinking a lot and was really anxious, but in the end I didn’t regret it. There was a quick drop of 20 cents or more in temperatures once the system stabilized. That said, if you can upgrade, maybe the best choice is to lower the CPU a bit to cut down on heat. Even at stock specs, this processor remains a solid option.
Jake Anderson explains his experience after the release. He felt nervous initially but never regretted it. Temperatures dropped quickly and stayed low afterward. He suggests reducing the CPU slightly as a temporary solution until an upgraded model arrives. He notes that the stock processor still works well and is reliable. He mentions his hands shake when handling the device, which affects his comfort, especially during precise tasks.
Yeah, the mine reached around 90c full load before my delid, even though I was on air. After delid, it rarely dropped below 70 with 1.25vcore while playing.
When you run tests at full load with a demanding program, do you have manual, auto, adaptive or offset settings for your vcore? Or does this occur during regular computer and gaming use (not under stress)?
Have you recently cleaned out your system? Excessive dust in the radiator and vents can significantly impact your thermal performance. Your ambient temperature is a bit high too, which might also influence your headroom—does it usually stay that way?
Jake Anderson shared his experience about Mine performance under load. He mentioned hitting around 90c before his delid, though he was on air. He noted that with a stressing program, he checks if his vcore settings include manual, auto, adaptive or offset options. He also asked if this issue occurs during regular computer and gaming use without stressing. He inquired about cleaning his system recently, as dust buildup in the radiator and vents can impact thermal performance. Regarding his ambient temperature, he said it tends to be higher, especially when the system is on auto, and that high temperatures appear during manual settings. He referenced HWinfo data showing his chip pulling around 1.36 VCore, while another source indicated a base of 1.8 V for the whole chip. He explained that lowering the setting further enough to improve cooling without causing boot issues required resetting the BIOS. He emphasized keeping his system clean and frequently adjusting settings, noting the dust from his cat in a 3800 sq ft home affects intake filters but doesn’t significantly impact internal temps. He mentioned planning to discuss this with EK tomorrow to see if their new mounting system will work with the updated design.
I managed a short-term workaround by setting the manual vcore to 1.23 just by chance. It starts Windows and I suspect the issue was related to Prime95, not my hardware. I also have AIDA64, and with CPU stress testing only, temperatures stayed in the low 60s (high 56s). I ran a short FFT test for maximum heat, it spiked to 85°C before restarting. More research suggests Prime95 may not give reliable stress results on Haswell chips due to a new instruction set. I think my last stress test might have been before that change was made. I don’t recall updating Prime95, though I might have. Full disclosure: my medications affect how I perceive time.
Maybe I should let it run for a few days and play some games to check stability. If it holds, I can gradually lower the manual temperatures and see what limits I can reach.
I’m not sure if this fixes the problem, but at least I’m no longer panicking.
Uncertain about closing this thread right now, but I’d like to express my gratitude to everyone who contributed. Sorry if I came across as condescending sometimes; it’s not my intention. No harm intended. Please share more ideas if you have any, I’ll consider them.
Jossrik explained how they typically assess radiator performance, noting that a rule of thumb is about 120mm per heat source plus one extra for better efficiency. They mentioned using a radiator rated at 420 watts to handle over 1400W loads, which aligns with their expectations. They also discussed acceptable temperatures for high-end components like a 980Tis under water BIOS, especially after extended use, and questioned whether the cooling system was functioning properly. The discussion covered factors such as processor heat output, ambient conditions, and the impact of software applications on temperature readings.