Queek

My educated guess is that there was a minor imperfection in the plating of the fitting that went. As soon as the underlying metal is exposed, you have a cell that will react preferentially with the tiny bit of non-plated metal. Electrochemical cells will react with the easiest anode-cathode pairing.citation needed; IIRC it was the biggest electrical potential gap wins While you may think you know all the materials in you loop (pretty sure you missed nickel because those look like nickel-plated-something fittings), there's an underlying assumption that anything plated will remain so (anyone who's done liquid cooling can tell you this is never the case). Also, "compatible metals" is a misnomer; as soon as you have two different metals (no matter what they are) connected by a salt bridge and an electrical connection, you have an electrochemical cell. Water has a very annoying habit of dissolving metals it comes in contact with, especially if there are no ions of that metal present.see ionic strength It may take a while, but "pure water" (this is actually very hard to achieve) will not remain pure water if it comes in contact with anything but a chemically inert material (e.g. plastics also a whole other argument). In addition, if you don't have an inert environment (you don't, I'm talking about N2 or Ar gas) you will have dissolved oxygen in your water which will contribute to oxidation giving more ions in your water giving new and exciting pathways for corrosion. Put all of these things together and it's effectively impossible to prevent an electrochemical cell from forming in a water cooling loop. Others have mentioned it, but I'll echo it. Check the rest of your loop (I'd start with your pump and radiators). You still haven't found your opposite anode/cathode. Somewhere in your loop there is the partner to your woes, and if you don't find and replace that, more leaks or failures are inevitable. To soapbox for a moment, I'm not sure what all the fuss is about preventing corrosion. You're never going to win. Container ships are fitted with a sacrificial anode (aka Galvanic anode), the sole purpose of which is to corrode faster than the ship itself. These sacrificial anodes are replaced when consumed and it keeps the boat afloat and the $2.2+ trillion per year international shipping business running. If an industry that large still hasn't solved the problem (or deemed it worth solving), us mere mortals cobbling together cooling loops sure aren't going to. Accept your rusty fate and perform required regular maintenance. (Reader please note: I am also a fool and do not practice what I preach. Forgive me father for I have sinned, it has been 2 years since my last computer loop flush.) Just don't be surprised when chemistry happens, because it will. TL;DR hypothesis: the nickel plating on one fitting had an imperfection which was worn/dissolved away and failed, electrochemical cell resulted, downhill quickly from there. Source: PhD chemist, but electrochem was a very, very long time ago and that isn't my specialization. I'm going off what I remember with a few sanity checks on assumptions. Also, I couldn't let this go without comment. This statement is inaccurate. By definition pure water is pH 7, completely neutral. The follow up sentences also have flawed logic.

My apologies for the late reply, I don't check this forum as often as I did in the past. Ethanol is just fine, there's a blurb on ethanol in my original post, but it seems that the spoiler nesting was ruined in the transfer to the new forum layout. I'll fix that.

@DominicNikon that was a might fast bump, it hasn't even been an hour. My guess is that your voltage is not set to manual mode. If you stress devil's canyon chips with anything other than manual mode, you get huge voltage pulls. Only stress test when you are using manual voltage mode.

I'm using a corsair h100 for it. Idle temperatures are around 30 and load is 60

That will probably work, but the chloroxylenol may stay as a residue after cleaning. It's not ideal, but it will probably work and won't damage your components. Be careful with it if you have pets, that stuff is really toxic to many animals.

Ring = cache = uncore For some newer processors, cache and core are separately overclockable.

I don't have time to do a teamviewer session to help you, but I expect this guide and this video will tell you everything you need to know. The video is featuring the PRO version of your motherboard, so the UEFI BIOS should be very similar. If you can't find something, look in your manual for its location. It's really not that hard, I'm sure you'll be fine. If you run into specific problems, post again on the LTT forums and we can help you out.

What is your processor usage and are the multipliers fluctuating?

Great! Keep in mind that software voltage monitoring is notoriously inaccurate, so the fact that the voltages are changing is more telling than their actual values.

The default C-states should work fine. Change Intel Speedstep to Enabled from Auto. Then boot into Windows and see if your ratio and voltage are behaving in an adaptive manner.

Have a look at the guide BigDay posted above (section 9.3). You'll find those options in the Advanced tab in your UEFI under CPU Power Management Configuration

Do you see the voltage fluctuating at all, or does it remain at a constant level? If it doesn't fluctuate, adaptive voltage probably isn't configure correctly. I'm having difficulty following what you have or have not done. Looking at your manual, it seems like the setup for your board should be very similar to the guide posted above. Have you enable the power saving features of the board? If not, your computer won't step the voltage or the multiplier.

There is some inaccuracy in the temperature readings (temperatures lower than ambient are impossible with radiators), but don't worry about it. I'm glad that worked! If you run into instability at lower speeds, troubleshoot with this thread.

Intel speedstep is what adjusts the multiplier dynamically. Enable it. don't bump, you only posted this three hours ago.

That's interesting. I thought it was a standard Z97 feature. So where it says "CPU Core Voltage" there is no adaptive option?