AndersWSP

About 3 years ago I took the wrong call, and ended up helping brother-in-law setup a small network in a concert/standup venue. The gear he supplied: Ubiquity Edge Router PoE Ubiquity NanoStation M5 Took us, (mostly me), a few hours but I rigged the nano station (powered over Ethernet) up as a wireless access point, and the Edge Router as a switch. A few sound mixers, a nice audio pult were rigged with Ethernet. Phones/Tablets were connected over WiFi as remotes. This worked for 3 years. Then the passwords he could definitely remember were lost to time. Was kind of sick, and didn't want to stay for another concert about to start, packed up the difficult gear and went home. Just now, after resetting AP/Router I realize how much of a pain older Ubiquity gear is. I can configure PoE to power the access point. I can configure AP or Router separately but can never ping between them. Can connect phone to the AP, seemingly but it doesn't actually work. I can setup DHCP on the router with a small range reserved for static audio gear. As such, I'm asking for backup. Was it always this difficult and I've simply gotten worse? Worst case, I deliver it back and ask him to hire someone more expensive. Best case, I'll learn something useful.

That's not a good noise. Sounds like your AIO pump is eating itself. If it coincided with the machine having been moved I'm guessing physical, like a bad bearing, caused or accelerated, by moving it around.

Not trying to be too pessimistic here, but it's possible it's not grounded correctly. Touching it would make it ground through you. That it's doing the same thing in several locations and outlets is a little disconcerting. If you can test with a different power supply it will at least tell you if your current PSU is the bad apple.

Verify that both ram sticks are actually working. Consider that the previous owner did this on purpose, maybe some of the ram slots are bad. With 1 for sure working stick of ram, test all the ram slots. I'm not aware of any setting in bios to disable half the ram, unless it's reserving them for an onboard gpu, that you don't have.

80c is still *technically* fine. What part is running hot, cpu, gpu or both? It'll narrow down the troubleshooting steps a little. Also curious to know how the machine has been assembled (thermal paste degrades over time, but your hardware is still relatively new). Also has it been recently moved/transported?

Considering your mobo flashed CPU and RAM i would try, in order: 1. Try just running 1 stick of ram in a slot not currently in use. If you have more than 1 stick of ram, try a few ram sticks in different slots. 2. Re-seat the cpu / inspect for damaged pads while you have it out anyway. If you're sure you plugged in everything, including the small CPU_PWR1 connector, then those 2 steps is everything I'll suggest without having hands on myself. Plugging a normal fan into 'Pump fan1' is fine. It's just a beefier fan plug for pumps (or fans) that draw more power - the machine should still boot with no fans plugged into CPU_FAN1.

*Also, scrub your pc clean of anything it doesn't need to run at startup. Full task bar + Mozilla with 4 tabs open as of writing this and above post.

TL;DR ~15w idle 3600 (according to HWInfo). Mobo: ASRock Phantom Gaming 4. 2x sticks of 3600, cl16 memory where I immediately turned off the RGB. Before you bring out torches and pitchforks yelling at me for having unreasonable numbers, I won the silicon lottery big and my 3600 is a beast. I'm running a static, all core overclock. 4.3 ghz @ 1.2v. (4.275 is p95 small stable, but I never run p95 anyway so 4.3 it is). 4.2ghz runs p95 stable @ 1.1625v which was my setting for the past 6 months. Infinity fabric locked to 1800, to match 3600 cl16 ram kit. The 'real' secret sauce to lower power consumption appears to be setting 'low numbers' into bios, that aren't low enough that it crashes or fails to boot. Specifically anything voltage related. Make sure you know how to clear cmos before embarking on this adventure. While I don't mind sharing exact settings I don't think they'll help unless you run the same motherboard. Can get much lower idle consumption but don't really need to at this point.

Alright, will re-run Ryzen clock tuner. See if that does anything interesting. One more point of order. I am "testing" in various real world scenarios. I'm actually using the computer, twitch streams, youtube, assorted games. Mostly running several of those at once. So far it's been fine. If any one game / software pops up that don't run with any given set of settings I'll deal with it then.

I would also use Prime95 for stability verification - there is a huge difference between them. I have also accepted that I don't care about p95. If everything else runs good just don't start p95.

I'm curious where you're getting the 1.2v limit from?

So, turns out setting Cinebench R20 priority to normal or above is causing the misreading (it also makes Ryzen Master act weird). I have Core Temp open in all Cinebench runs for consistency. The numbers it reports normally match with what's reported by other software. I just want temperature & power draw visible in one location. Just fucking around with the 1.3v oc. Don't think it's fully stable, so will dial back down within a day. Only started tweaking/experimenting because the internet stopped working for a few hours... Turns out all the games installed required internet access.

Been messing with overclocking on and off ever since I got my 3600 (nice cpu by the way), and I recently discovered that it seems "stable" at higher clocks than initial testing suggested, so trying to see if that's actually viable running 4.45ghz instead of the usual plebeian 4.4ghz - doesn't make a real difference, but higher numbers = better. Cinebench was set to high priority in windows 10 task manager - affinity was set to cpu 0. So far CPU 0 seems to be the only one doing this, others report normally. *Edit common trend. So with that in mind I present to you these lovely screen grabs. I wasn't quick enough to screen grab it, but core temp briefly reported a 2401W power draw which honestly seemed a little excessive for a 1 thread workload. Not sure what there's to learn from that. Probably wont matter, but running an ASRock Phantom Gaming 4 570x board, the 3600 and windows 10. Rest of specs shouldn't be relevant for this.

I found a program called 'FanControl' it allows you to set fan curves or fixed values for fans without going to bios. You can do all testing/experimenting from windows - and if you smart replicate whatever you like in BIOS - or be lazy like me. I just keep running FanControl to let it handle it. Doesn't really take up any meaningful resources to run. You can find FanControl here:

Tried cranking the fans (still letting the CPU do it's own thing), slightly colder, but no change in clocks.