Tom's Hardware Member System Gallery

Page 175 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Jesus, that's a fracking work of art Jim. Very, very nice. It almost makes me want to build a custom loop now, even though I have not ventured yet to have much faith in water cooling. The aesthetic alone almost makes it work it.
 
Here is the new build, yes I know the HSF is a little off (Didn't notice until later), but that will be fixed once I put in the I7 8086K. :D


DSC_6084.JPG



And running:

DSC_6082.JPG
 
I just shoehorned a Thermalright ARO Ryzen cooler onto a m-itx board in a mid-tower case. I had to remove the rear exhaust fan for it to fit. It was worth it though, as it dropped CPU temps by 10*c over the CM Hyper 212 Evo.

view
https://drive.google.com/file/d/1qvCTFIXLLF1BVCehi9IzoQLR7Odslb9I/view?usp=drivesdk

Specs:
CPU - Ryzen 5 2600x
Mobo - Asrock Fatal1ty ab350m
RAM - gskill Flare X 3200 cl14
GPU - EVGA GTX 1080 SC
PSU - Corsair rm850x
Case - NZXT 400i
 

CRO5513Y

Expert
Ambassador
Now that my RGB RAM has finally arrived, i'm happy with the overall aesthetic (and performance) of my build for the time being! :D

VolhUqY.jpg

Yi2UnsJ.jpg

3W0dEnq.jpg

KX1P2tC.jpg


Specs:

i7-6700K @ 4.6 GHz @ 1.359V (edit) = 17-23c Idle (Can't believe how low it gets, but can verify), usually 25-35c while Gaming, max i've seen myself is 37c in Star Citizen which hits 60-75% utilisation, max of 61c in Prime95 for 15 minutes.
Thermaltake Floe Riing RGB 360mm AIO
16GB Viper Elite RGB DDR4 @ 3600 MHz C16
GTX 1070 Ti Gigabyte G1 Gaming @ 2050 MHz Core / 4500 MHz Memory OC @ 120% Power Limit
ASRock Fatality Z170 Gaming K4
Thermaltake Toughpower Grand RGB 650w
Coolermaster Mastercase H500P (2x CM Masterfan 200mm RGB front intake, 1x CM Masterfan 140mm RGB rear exhaust, 3x TT Riing Premium 120mm RGB top exhaust on the 360mm Radiator)
240GB Galax M.2 NVMe + 256GB Samsung 840 Pro + 3TB WD Red Pro 7200RPM HDD
 

Karadjgne

Titan
Ambassador
Wow, you live in an igloo? @ 72°F is normal for most AC users, that's @ 22°C. Cpu temps, case temps etc are normally @ 6-12°C higher than ambient at idle. So normal idle will be @ 30-32°C on most cpus. You can't mechanically cool a cpu below ambient temps, it takes a chemical process such as LN2 or peltier to accomplish sub-ambient temps. If your cpu sits idle at @ 20°C, minus case temps, puts ambient at @ 10°C or 50°F. Dude, that's freekin cold!
 

Karadjgne

Titan
Ambassador
Not trying to knock the build, it does look good, and it's always nice to see builds not using latest and greatest hardware, just my thought process runs more towards if 1 temp is suspect, then all temps by default become suspect, so your max of 61° under p95 could really be a max of 81°C, gaming temps should normally be in the 50's etc. That may or may not mean changing up some fan curves, locations, OC etc. I once saw a post with similar results, an fx6300 sitting at a clean 18°, several software tests agreed. End result was a sensor error for sure, something op could live with. He just added 20 to everything and called it a day.
 
You can't really know what AMD temperatures are anyhow. Even AMD states that their thermal monitoring process is not intended to provide exacting specifications. Distance to TJmax thermal margin uses an entirely different process and the only surety is whether you are within a specific range and distance from "something bad is going to happen" or are within a tolerable range.

My 6700k manages to score better temps than about 75% of the comparable systems I've seen with that CPU, and still hit about 71°C under full stress at 4.5Ghz. And I have a pretty strong cooling configuration with six Noctua fans, the exhaust being a 2000rpm NF-A14 and a Noctua HSF. I've never seen any AIO get better temps than this cooler, on this CPU, only custom loops.
 

CRO5513Y

Expert
Ambassador
Not sure what to say really. It's what everything tells me (Afterburner, CPU-Z, Hwinfo, ASROck motherboard utility, etc). I've got some screenshots around somewhere including with Prime95 running. If Techy is lurking around, he's seen them all. As for ambient, not sure. My room is pretty cool but nothing freezing (I should mention it is the middle of winter here at the moment). Pretty strong airflow in the build and the cooler is definitely overkill for my 6700K. Although i agree my idle temps seem a little abnormal hitting below 22c, i'm not too sceptical of my max temps looking at reviews of my cooler on hotter CPUs than my own scoring similar.
8301_28_thermaltake-floe-riing-360-tt-premium-cpu-cooler-review.png

8301_30_thermaltake-floe-riing-360-tt-premium-cpu-cooler-review.png

Link to above > https://www.tweaktown.com/reviews/8301/thermaltake-floe-riing-360-tt-premium-cpu-cooler-review/index6.html.

Tom's has a review too, but they didn't test it as extensively > https://www.tomshardware.com/reviews/thermaltake-floe-riing-rgb-360-tt-premium-edition,5315-2.html.
 
I see why your temps are so low. Your core voltage is only set to 1.3v. I seriously doubt it's stable at that voltage. Just because you're not seeing errors does not mean the system is stable.

Run Realbench on the stress test for 8 hours.
Run Prime95 version 26.6 (No other versions) on the Blend test for 8 hours.

If it passes both of those at that voltage, with no worker stoppages, then you have the greatest 6700k that ever came off the production line.

You will want to keep HWinfo open the entire time so you can visually see if a worker gets dropped. Otherwise, it will continue testing using the other workers and you may not even know it. When a worker stops or errors out, HWinfo will show zero or very low usage of that core and it will not resume work during the remainder of the test.
 

CRO5513Y

Expert
Ambassador


Apologies, it's at 1.359v, not sure why i wrote straight up 1.3v in the original post. Guess after a while i stopped being specific as i rarely talk to other Overclockers...

It's been running at 4.6 GHz @ 1.359v for almost a year now. As for testing... Passed Prime95 26.6 - Small FFT Test for 15 minutes (Max recorded temp was hit here) & Blend test for 3 hours (i have no interest in running longer). Passed Aida64 - 30 minutes. Outside of that initial stress testing, have run Cinebench various times throughout the past year, same with 3DMark, CPU-Z validation bench, UserBenchmark and Unigine Valley/Heaven/Superposition all come to mind from recent runs since i got my new RAM. Honestly if that + a year of running frequently across some pretty heavy games and system benchmarks with no freezing, crashing, BSODs or other issues affected by unstable overclocks isn't enough assurance, it definitely is for me.

But anyway i didn't come here to discuss overclocking/stress testing, i do most of that on Tom's Discord. Just wanted to post my build now that i've finished upgrading it for a while. I'm not doubting slightly 'suspicious' temps but i have serious doubts a lightly overclocked 6700K is going to go very high under a well-rated 360mm AIO + Gelid GC Extreme thermal compound in a system with fairly strong airflow.

EDIT: Here is my idle CPU temps in RealTemp right now.
2iL6b0g.png
 
The instability problem I'm talking about isn't about errors, blue screens or crashing. It's about micro-errors that minor instabilities introduce into your OS file system and non-OS file data like files for games, pictures, movies (Especially if they've been encoded to a different format), documents and backup data.

Obviously it's your system and if you're happy with it, then great. But if or when you start seeing unexplainable issues with programs, files or the operating system, you'll understand why full stability is a major concern. And if you are running the memory at any speed faster than the default 2133mhz, regardless of whether the CPU is overclocked or not, it is ALWAYS a good idea to do at least three passes of Memtest86 AND an 8 hour run of Prime Blend mode.

My current 6700k system with OC memory (And anything over 2400mhz is considered OC on DDR4) was fine for 7.5 hours in Blend, and errored out. I had to up voltage by .020v on memory and raise both Trcp and Trd back up by one point to pass 8 hours. If I had tested for less time, I would not have caught that and I'd have thought everything was fine when it wasn't. A stable system should be able to run any stability check practically indefinitely but certainly longer than 8 hours. 8 hours is just a commonly accepted timeframe suggested that it's unlikely and that in testing it's rarely been seen to still error if it lasted that long. Your system though, I'm good with whatever you choose to do.