site banner

Friday Fun Thread for December 20, 2024

Be advised: this thread is not for serious in-depth discussion of weighty topics (we have a link for that), this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

2
Jump in the discussion.

No email address required.

Nice, custom loop or AIO? If custom loop what are people using nowadays for water-blocks on their GPU, did EK survive their 'try just not paying their suppliers and employees' experiment?

What's your workload like, mostly gaming? I feel like the sweet spot for AIO cooling of CPUs is something like 30s to 10 minute saturation workloads. Shorter than that and you're not producing enough heat for your cooling system to matter. Longer than that you're saturating the liquid anyway. But for some GPU limited games and some workstation tasks even a "cheap" AIO has way better peak noise normalized performance than even pretty premium air coolers.

It's an AIO named Arctic Liquid Freezer III 360. :) It was supposed to be notoriously difficult to install, but it went fine as soon as I figured out I had the brackets upside down. The smoother, prettier side needed to be facing down to the mb, quite counter-intuitively.

I'm trying to figure out the best activity/cooling curves. I'm using Argus Monitor.

I've got the connector labeled PUMP plugged into the header aio_pump, and I plugged the connector labeled VRM into either cpu_fan1 or cpu_fan2, and the connector labeled FAN into cpu_fan2 or cpu_fan1.

In the program, there's one thing labeled cpufan running at 640 rpm while at 32% speed - I assume this is the vrm. Then there's auxfan0 reporting 1020 rpm while at 32%. And auxfan1 at 1240 rpm(?) while at a fixed 20% - guess this is the pump. I set it at a fixed 20% because it was making a repeating whining or wind blowing noise every second at higher speeds.

My biggest regret with my last build (besides choosing Intel and getting a sure-to-die 13900k which I'm loathe to RMA which would strand me without a desktop for 2+ weeks) was opting for air cooling. I've used Noctua fans for a decade now, they've always done well by me. The first (and last) time I tried an AiO water cooler was in 2013 and it was so terrible I wrote them all off entirely. Pump headers weren't standard on motherboards back then and I remember absolutely hating Corsair's software that was mandatory to keep running 24/7. (I'm not sure if I could've "just" pegged CPU_FAN1 to 100% in the BIOS and things would've worked; probably should've tested it.)

Anyway, despite having a case designed for maximum airflow and a huge Noctua fan, my 13900k will thermal throttle instantly if I run any benchmarks. Though it's fine in daily use and I haven't seen it exceed 80 while gaming, it is the principle of the matter... also the Noctua fan is so large I need to take the whole assembly off if I want to swap out RAM or m2s, which is rather annoying. Apparently AiOs have gotten better lately, so I'll give them another shot for my next build. Or just go whole hog with a custom loop, it doesn't look that hard.

Heh. I wouldn't touch Intel with a bargepole at this point, and I assume you will surely pick AMD next time. The whole "RMA/stranded with no desktop PC" issue is why I'm keeping my old system instead of selling it. Wouldn't get more than five or six hundred for it anyway. Always safer to have a backup option.

Custom loop sounds like fun. I might try that next time around. It seems like if you want a much quieter build overall under gaming load you need to liquid cool the GPU as well.

I wouldn't touch Intel with a bargepole at this point, and I assume you will surely pick AMD next time.

Can you elaborate? Last I checked (year ago maybe), intel still seemed to be the choice for power, with AMD really only preferable for budget / green builds. You make it sound like there’s no contest anymore - was I mistaken before / did something go wrong at intel / did AMD put out new cpus?

I don't have the full picture of what things were like around a year ago, except that for gaming purposes the AMD 7800x3d was definitely ahead of Intel back then. For productivity I don't know. Right now though, AMD is ahead of Intel for all purposes, including raw power/productivity, with better performing and less power hungry cpus. And they have some more cpus coming out in early 2025, which will probably widen the gap. With the current AMD 9800x3d, it's not even a contest for gaming. Something did go wrong at Intel. Their stock is down by around 60% since the start of 2024. They released worse products, and worse yet, there were serious errors in some of them, which SubstantialFrivolity and ThisIsSin have touched on. It looks like there won't be an easy way back up. There's been talk of Intel being bought out and split up etc. I wouldn't go near their stock or their products for the time being.

It's telling that the most popular Intel product is N100, an ultra-low voltage CPU that competes with Raspberry Pi and friends. It looked like Core 12xxx was going to be Intel's return to glory, but they dropped the ball and released two completely shit CPU generations.

Intel has had a problem with chips that damaged themselves because they weren't properly signaling power draw to the motherboard. They put out a microcode update to fix it, but the processors which were damaged can't be fixed. This has damaged Intel's reputation in the eyes of a lot of people who got burned by this bug.

Ah yes, this generation’s “Pentium bug.”

Well, there have been a few of them. There was one in 2011 where the SATA3 ports controlled on the northbridge would cook themselves (though that one was a problem with the chipset, not the processor); then there was Skylake's shit-ton of bugs in 2016 or so (to the point that Apple said 'fuck it' and accelerated their own CPU development). And then are the series of speculative execution attacks that made every older computer 30% slower when the patches were installed (and for some of them, there is no patch, they're just broken forever).

It seems Intel fucks up in a major way every 5 years or so at this point.

a repeating whining or wind blowing noise every second at higher speeds

Hopefully that goes away as the bubbles in the loop make it to the top.

With a 360, I wouldn't expect a load any normal user would reasonably run on a computer that's sitting in the same room as them to fully saturate the liquid in the loop. Maybe a supper long session of Cities: Skylines II, if you have literally Linus Torvalds needs for compiling Linux, or if your room needs an electric space heater so you run a synthetic benchmark/mine to keep warm at night.

I suspect there's something wrong with the fans.

The VRM fan seems to be the culprit for the annoying noise. The noise changes when I regulate its rpm in Argus. Setting it to bios controlled seems to reduce it the most. It's now noise free at idle. That's the most important thing.

The fans pointed at the radiator seem to be a little off-kilter. The graphic on them around the centerpoint moves.

The fans that were delivered with my case were also messed up. They made a much more high frequent irregular noise, due to bad bearings I think. I had to replace them.

Is there an unusual amount of shittiness in the fan production industry right now?

Edit: I swapped the connections. I thought I had VRM on cpu_fan1. It was on 2. Radiator fans was on 1. Now, with vrm on cpu_fan1 and radiator fans on cpu_fan2, the noise is mostly gone, only appearing briefly during ramp-up! Yay! Hardware idiosyncracies. ¯\_(ツ)_/¯