TacticalTactics

Those "feature sets" are some ripple and voltage regulation requirements for the top tier units, which are actually quite generous, and mostly protection features, as well as a few factors which often contribute to PSUs performing outside of ATX spec. Multi vs Single rail, DC-DC vs Group regulation, OCP, OPP, OTP, SCP, all shown in the methodology, in the tier list (In case you're wondering about what those mean, there are links in the tier list under Commonly Mentioned Terms to explanations). Stuff to keep your PSU from frying other components, for example. Not sure what you mean by reliability, if you could elaborate on that, and how you would structure a list or buying guide to rank by "reliability". I'm curious which high ranking unit in particular you could point to and give evidence for it being garbage too. As for hardware failure, not sure where you see anecdotal stories of hardware failure in the post... All rankings are done with objective PSU reviews done by people with years of experience like Jon Gerow, Oklahomawolf, and Aris. And there are citations, if you look close enough: Clickable link, if you'd like to look for yourself: https://drive.google.com/file/d/1PBO1JI_yMZfgupiWdkIi9228xJ0LMYzf/view?usp=sharing

Not a thing afaik. If it does exist, it's probably just a gimmick to make a low quality pump and/or CPU block sell.

I never said it would always deliver 384W... I said if that much power was being delivered to the chip

One 8 pin (Assuming it isn't the high current variant) can deliver 384W. If that much power was being delivered to the chip, it would have to be emitted as heat. No AIO can handle over 384W of heat. Hence why its for LN2 overclocking. Another use for it on some boards is powering PCIe slots, if you have multiple GPUs which only draw power from the slot. 24 pin ATX alone can be insufficient. Other boards would use PCIe power connectors for that.

Just because its on 24/7 doesnt mean you HAVE to go custom water cooling. For overclocking cpu + gpu, you'll want at least 4 fans worth of rad space, as a rule of thumb. More is preferred of course, but there is diminishing returns since its still ambient cooling.

Custom watercooling is definitely a case of "if you have to ask the price, you can't afford it". You shouldn't go in looking to spend a specific amount and no more. It all depends on whether you overclock, hardline or soft tubing, aesthetics choices, and maybe some quality of life choices like a leak tester, drain port, or flow meter. Heatkiller is a much better manufacturer of waterblocks than EK, but finding stock right now could be hard. Also, you don't really need to liquid cool those VRMs, with a monoblock. Monoblocks on X570 are pretty much a waste of money over just CPU blocks because most X570 boards' VRMs just don't need more than the passive cooling with the stock heatsinks.

About a week ago, I noticed short bursts of static in DCS, along with occasional hard crashes (sometimes application freeze, sometimes system restart). It seemed to improve when I had less open in the background (closing chrome/discord). It later occurred in other games and applications. Sometimes mouse inputs would be registered late. I had updated my chipset drivers (I don't remember whether I did so before or after I noticed the artfacting), and then updated motherboard audio drivers, as well as updating steelseries software. I used latencymon to check and surely enough, Highest measured interrupt to process latency was reaching 2000-3000 microseconds. Reported ISR and DPC execution time also reported high values and showed dxgkrnl.sys and nvlddmkm.sys as the highest execution time. So I updated graphics drivers, using clean install and only installing the two essential drivers. USB C, PhysX, and Audio drivers were not installed. That did not help. I tried modifying my ram OC, disabling PBO, and even resetting the ram to 2133Mhz. Nothing did anything except resetting ram to 2133, which seemed to lower the ISR and DPC execution times greatly. But the interrupts to process latency was still 2500 or so. I tried changing the power plan in control panel to ryzen high performance, and applied my ram oc again, along with enabling PBO. It had the same effect as resetting the ram OC. I used malwarebytes to make sure there isn't malware either. Relevant system specs: Ryzen 7 3700X, PBO 300 230 230 x5 Gigabyte X570 Aorus Elite, bios is version F10, the oldest which had AGESA 1.0.0.4B. I haven't bothered to update since then since F11 and F12 reportedly only improve PCIe device and memory compatibility. Patriot Viper Steel 4400C19, tightened down to 3733 14-15-15-30-48-1T 1.47V GDM enabled, tRFC 298 EVGA RTX 2070 XC Gaming, +131Mhz core, +915Mhz Mem EVGA G3 650W Windows 10 Home Arctis 7 headset Corsair K68 Keyboard Logitech M510 mouse Logitech Extreme 3D Pro joystick At this point I'm running out of ideas, so any help is appreciated. If you need any log files or any more information I will try my best to provide it.