5GigsOrBust

That sounds like fun, you make me want to build a system with older hardware and mess around with it. I was considering it, but I definitely don't have the space for the mora3. The monsta 400 already kind of hangs off my desk.

Well its a 10 core running 5238Mhz all core at 1.385v LLC6. Runs at 1.4v idle. Not sure if that qualifies as a monster overclock, but that's definitely the limit of this silicon.

Haha yes I like a bit of rgb, but I prefer the look of noctuas. I went with a bench setup so that I wouldn't be limited in my component selection by case size. The 400mm rad is made by Alphacool and I bought it from Modmymods. It took like over 2 weeks to be delivered, but it was cheaper than a 360x85mm rad. I have a small fan that I run over the ram for testing, I removed it for the pictures. I can definitely respect your motivation to hit 5ghz, that was my goal when I started overclocking. From your picture though, looks like you have most of the parts you need to complete your loop.

So this is my first custom loop build. I wanted to go custom to decrease noise and to give myself the ability to watercool the board vrm since it came with a waterblock, and the GPU in the future. CPU itself can draw 300W under load so to cool everything in one loop, I need quite a bit of cooling power. I started looking into the surface area of different size radiators and determined that a 400mm rad is the probably the sweet spot. The 400mm rad has more surface area than a 560 rad, and I only have to buy half the amount of fans. My main concern was the fact that 200mm fans usually have low static pressure, but I found a solution to that. For Science! gave me a link for a high static pressure version of the noctua nfa-20 fans that are not listed for sale on noctua's website, so big thanks for that. Finally got all the parts in and built. After testing, I found the temps to be within only few degrees of my previous setup, but that is with the vrm included in the loop, so I'm fairly certain the temps should be lower without that. At idle, the coolant temp sits right around ambient. Also the max fan rpm has gone down from 3000 with the previous setup, to 1200 so the noise difference is incredible. I went with soft tubing & barbs for now because the system is on a test bench, so who cares. The cable management is terrible, I just have everything set up for testing. Not sure if I could fit this into a case, but for now I like it sitting next to the bench. The size of this thing sitting on the desk is hilarious to me. Wasn't able to find many custom loop 400mm builds online so I wanted to share this.

Nice if I had the space for that setup, I would definitely do something like that with 200mm fans

Thank you, I had no idea there was an HS version of these fans. I ordered them within minutes after reading your post. The static pressure of these fans was a big concern for me, so this HS version should be perfect.

I doubt it! Have you seen this rad? Its not a standard size, its an 85mm thickboy. If I go with push-pull, that's an extra 60mm of fans

Building my first custom loop and obviously I have to go with the largest possible fan size, because go big or go home. Anyways, I just found out that 400 mm radiators are a thing so I just picked up a Nexos 400mm Monsta rad and 2x Noctua NFA-20s. Might need 2 more fans for push-pull, but we'll see after testing. The interesting part is, I wasn't able to find a single video on youtube of anyone doing this, or any photos of a 400mm custom loop on google. I'm sure its been done before, but it's definitely never been properly captured & documented on the interwebs. So, that's why we're here today. I can see you getting ready to assault this plan with criticism. But don't worry, and before you pick up that keyboard, I have researched this and I do understand that the static pressure of 200mm fans is crap, and I know I'll never be able to fit this setup in a case..etc. I'm only doing this because I like the idea of having a comically oversized rad & fans that will be absolutely silent. All that said, I'm curious to hear opinions on how the internet thinks this will turn out. Links below for the items if you want to get an idea of the size of this rad & fans. https://www.aquatuning.us/water-cooling/radiators/radiators-active/180200mm/27838/alphacool-nexxxos-monsta-full-copper-400mm-radiator?c=6566 https://www.amazon.com/Noctua-NF-A20-PWM-Premium-200x30mm/dp/B071SLFBNY

I want the Noctua 3070 because its ridiculously huge.

Thank you, this is exactly what I need.

I haven't tried taking it out and putting it back, but I already installed windows and ran some stress testing at stock settings.

It pulls close to 300W at 1.3v for just the CPU, not sure how much more power it wants. Already close to melting

Yeah I need to do more testing to find out. My main question is how to get the voltage values to show up under the prediction in bios.

Let me preface this by saying that I intended to post this question to Asus forums, but I am physically unable to create an account there due to their completely broken captcha system. Honestly, that deserves its own rant.. Anyways I recently picked up a Maximus xii formula and 10850k for testing. The bios SP prediction gives this CPU a rating of 115. For context, an average chip is around 63, and the world record 10900k was like 117. Soo... either I got an absolute golden sample unicorn chip or this board's prediction is F'd. Now after looking at the vid tables for this particular chip, it does seem to be able to run on pretty low voltages compared to other 10900k & 10850k samples. I haven't done enough testing yet to confirm this. However, what makes me suspicious about the prediction is the fact that there are no voltages provided under the prediction value. Normally, the bios gives expected load voltages in addition to the prediction value. Instead of having voltage values, all of the fields say N/A. This has me wondering if the prediction really is broken on this board. I've cleared bios, and toggled AI tuning on and off in an effort to try to get those voltage values to show up so I can validate the SP prediction. Any guidance would be appreciated. Screenshot below of what I'm looking at. IMG_3286.HEIC

Yeah makes sense, I was wondering if that would be a factor. Idk why I thought it would be easier to find a decent one. Hilarious. Clearly that's the answer

Dumb question, but seriously its not easy to find an ATX test bench. I see cheap crappy-looking ones for under usd $100, and I included a picture of those below for comedic value. I also found something from Praxis that costs over $200. Why would a test bench cost more than a decent full case?! I would go for the cheap one but I'm gonna be running a 420 rad so I need something more sturdy. First time buying a test bench so idk if the market is always this crappy or its a more recent supply chain thing. I see decent-looking test benches in tech videos, so I know they exist somewhere, but I can't find anything online. I'm getting desperate enough to consider something like the Core P3, but its not really a test bench. As stated, I've included a picture below of the kind of thing I'm trying to avoid, any recommendations are appreciated.

You can change the frequency of the ram, and it MIGHT be stable. It depends on if your motherboard is capable of setting stable auto timings at the frequency you select. Generally if you downclock to 2666, it'll probably be stable because that's a low speed and its easy for the board to set stable timings.

Idk, cpu is already maxed out at 5.1 allcore 1.32v core and 1.25 vccsa & vccio Yeah that's what I tried first and it is stable, but the secondaries are loose on auto. I can tighten all secondaries except trfc & trefi.

Before you read on, its not Ryzen so you don't care. I'm kidding. Anyways, I found a 16gb kit of 4000 16-16-16-36 ram on amazon for $100. Obviously B-die, and obviously I had no choice but to buy it. Loaded it in, enabled xmp, aaaaaand............... the stock xmp profile is unstable. Not surprising for cheap B-die. So here we go on yet another memory overclocking adventure. I tried a cl16profile I had saved after OCing my previous 4000 cl19 B-die kit, and it seems to work so far. I've had to loosen trfc & trefi because it won't pass memtest otherwise. Now it runs fairly stable at 16-16-16-34 with decent secondaries, except trfc & trefi. I'm curious about what others have been able to hit with trfc & trefi on B-die. Not sure why my kit won't go low on those two timings. 374 trfc & 30000 trefi are almost kinda somewhat stable, and even over 1.5v I can't tighten them at all. 15600 trefi is stable, but 19667 trefi is not. Makes sense.. Isn't ram overclocking run ?! System specs below

Increasing ram frequency & voltage will not degrade the CPU. VCCSA & VCCIO are the two main memory-related voltages on Intel. Setting those voltages too high can degrade the IMC. Usual advice is to not go over 1.30-1.35v longterm on those. Usual advice for ram voltage depends on the specific memory IC, but on B-die its not recommended to go over about 1.5v for daily usage. I run mine a bit higher because I have fans above the ram sticks and I don't mind replacing them, but that's not recommended for everyone. I would recommend before you overclock your ram, find out what IC you have, and look up common OC settings for that specific IC.

I would consider that perfectly okay for an overclocked 10 core under avx load.

The answer to your question is a definite maybe.

CPUs stock behavior is to enable turbo based on how many cores are active. When idle, it can reach max single core boost, but when all cores are under load ie gaming, the all core boost is lower. You can fix this issue by overclocking.

So I bought a 4000 19-19-19-39 B-die kit to overclock. Couldn't raise frequency no matter what I tried, so I settled on tightening the timings at 4000. After quite a bit of struggle, I've got it down to 15-15-15-32 at 4000mHz with 1.51 volts. Stress tested 400% memtest. Any lower on the dram voltage and I need to raise the cas to 16, which is quite annoying. VCCSA & IO are at 1.28v and I refuse to go any higher on those for daily use. Since its stable, I've decided to leave the ram voltage at 1.51v and run that daily. Every guide I see says up to 1.5v on B die should be okay for daily, so maybe I'll be okay? Hopefully the sticks don't catch fire, but I'll post pics if they do. As stated, wish me luck..

My 9900k can spike to 259 W power draw under avx load, so I would expect you could easily see 300+ with that 10 core