How far can a 3900x go on a custom loop?

[jtmoseley]

I have all the parts for my custom loop. I will build it tomorrow, but I'm curious about how high the overclock can go on a custom loop compared to the stock cooler. Any experiences to share?

[SavageNeo]

depends. on air i can get 4ghz on all core, on 360mm aio i can get 4.3ghz on all core. if you can get 4.5ghz stable you could make it work. even higher if it is stable. temps are not the problem in ryzen overclocking really. ryzen just doesnt really give much overcloking headroom

[jtmoseley]

Just now, SavageNeo said:

depends. on air i can get 4ghz on all core, on 360mm aio i can get 4.3ghz on all core. if you can get 4.5ghz stable you could make it work. even higher if it is stable. temps are not the problem in ryzen overclocking really. ryzen just doesnt really give much overcloking headroom

Alright so 4.4 or 4.5 is where I should go for

[SavageNeo]

yes. you can try higher but i doupt that it will be stabile any higher than 4.5ghz

[Letgomyleghoe]

38 minutes ago, SavageNeo said:

even higher if it is stable. temps are not the problem in ryzen overclocking really. ryzen just doesnt really give much overcloking headroom

It's ryzen voltage is so finicky.

[Letgomyleghoe]

38 minutes ago, jtmoseley said:

Alright so 4.4 or 4.5 is where I should go for

If you get 4.4 or 4.5 I would be absolutely amazed.

[Letgomyleghoe]

45 minutes ago, SavageNeo said:

get 4.3ghz on all core

This is extremely good for the 3700x your using I'm assuming.

 

If you get 4.5 GHz that runs a cinebench stable I will give you a million dollars and drop dead, it's hard to break 4.2 GHz all core on the 3900x.

Edited May 25, 2020 by scuff gang

[Jumballi]

51 minutes ago, SavageNeo said:

depends. on air i can get 4ghz on all core, on 360mm aio i can get 4.3ghz on all core. if you can get 4.5ghz stable you could make it work. even higher if it is stable. temps are not the problem in ryzen overclocking really. ryzen just doesnt really give much overcloking headroom

But if his custom loop only has 360mm or less of rad, then it'll be the exact same amount of cooling you have or even worse.

[SavageNeo]

35 minutes ago, scuff gang said:

This is extremely good for the 3700x your using I'm assuming.

 

If you get 4.5 GHz that runs a cinebench stable I will give you a million dollars and drop dead, it's hard to break 4.2 GHz all core on the 3900x.

i can run my 3700x at 4.325ghz stabile and when i play i use 4.35ghz (not stable in benchmarks btu stable in games)

[jtmoseley]

9 hours ago, Jumballi said:

But if his custom loop only has 360mm or less of rad, then it'll be the exact same amount of cooling you have or even worse.

Incorrect. A custom loop will always be better than an AIO of equivalent rad size. There is way more water to heat up than in an AIO. I'm willing to bet that my 240mm AIO with a d5 and a large reservoir will be better than a standard 360mm AIO with a shitty pump and no reservoir.

 

Also, AIOs use oil-like stuff. Less conductive than water.

[Lurick]

13 minutes ago, jtmoseley said:

Incorrect. A custom loop will always be better than an AIO of equivalent rad size. There is way more water to heat up than in an AIO. I'm willing to bet that my 240mm AIO with a d5 and a large reservoir will be better than a standard 360mm AIO with a shitty pump and no reservoir.

 

Also, AIOs use oil-like stuff. Less conductive than water.

No, I'm sorry but that's just wrong. While a res might add more water causing the loop to take longer for the liquid to reach heat saturation it won't cool any better. A res will NOT improve cooler performance in any meaningful way and to say that a custom loop is magically better than an AIO is also incorrect. Sure you can use thicker radiators which have more surface area to dissipate heat with, all things being equal (size of rad and pump speed) they will perform the same. AIOs also generally use ethylene glycol which might feel slippery or they'll use distilled water with additives like corrosion inhibitors and biocides. In custom loops, depending on who makes the coolant you buy, might have Propylene glycol or they use ethylene glycol or you'll just distilled water with biocides like in an AIO.

[nick name]

My 3900X gets to 4400MHz on CCD 1 and 4250MHz on CCD at 1.26V with second highest LLC setting.  Trying to go for higher speeds doesn't work out well after those speeds as I have to add more voltage which adds more heat which means I have to add more voltage.  That's on a 360 AIO.  

 

And those speeds and voltage are Folding stable.  

[jtmoseley]

1 hour ago, Lurick said:

No, I'm sorry but that's just wrong. While a res might add more water causing the loop to take longer for the liquid to reach heat saturation it won't cool any better. A res will NOT improve cooler performance in any meaningful way and to say that a custom loop is magically better than an AIO is also incorrect. Sure you can use thicker radiators which have more surface area to dissipate heat with, all things being equal (size of rad and pump speed) they will perform the same. AIOs also generally use ethylene glycol which might feel slippery or they'll use distilled water with additives like corrosion inhibitors and biocides. In custom loops, depending on who makes the coolant you buy, might have Propylene glycol or they use ethylene glycol or you'll just distilled water with biocides like in an AIO.

The longer the coolant takes to heat up, the better the performance is. The CPU will stay cool for longer.

 

Heat Dissipation is less or the same, a radiator is a radiator, but the reservoir does add meaningful performance. 

 

As does the D5. Which pumps way faster than what you get with an AIO.

 

[Lurick]

5 minutes ago, jtmoseley said:

The longer the coolant takes to heat up, the better the performance is. The CPU will stay cool for longer.

 

Heat Dissipation is less or the same, a radiator is a radiator, but the reservoir does add meaningful performance. 

 

As does the D5. Which pumps way faster than what you get with an AIO.

 

Well then you and I will disagree with numerous water cooling sites proving you wrong on the res portion. It's as wrong as saying loop order matters in any meaningful way.

 

https://www.overclockers.com/water-cooling-reservoir-theory-and-construction-guide/

 

https://koolance.com/how-to-build-a-water-cooled-pc

[jtmoseley]

7 minutes ago, Lurick said:

Well then you and I will disagree with numerous water cooling sites proving you wrong on the res portion. It's as wrong as saying loop order matters in any meaningful way.

 

https://www.overclockers.com/water-cooling-reservoir-theory-and-construction-guide/

 

https://koolance.com/how-to-build-a-water-cooled-pc

It's literally basic science. You have more liquid to heat up, hence it stays cooler for longer and cools the cpu down better.

 

You increase thermal mass so it takes longer to reach equilibrium. The water stays cooler for longer.

[Lurick]

4 minutes ago, jtmoseley said:

It's literally basic science. You have more liquid to heat up, hence it stays cooler for longer and cools the cpu down better.

 

You increase thermal mass so it takes longer to reach equilibrium. The water stays cooler for longer.

Once you reach equilibrium your argument is completely invalidated, you either have enough rad space to keep the loop cool or you don't. You might get a minute or two on a basic loop of "extra cooling" and that's it. There are DOZENS of youtube videos that prove res size doesn't make any meaningful difference but whatever, I'm out.

[Jumballi]

1 hour ago, jtmoseley said:

It's literally basic science. You have more liquid to heat up, hence it stays cooler for longer and cools the cpu down better.

 

You increase thermal mass so it takes longer to reach equilibrium. The water stays cooler for longer.

Yeah basic science, it can't cool it down more than the temp the water is, which is at best, the ambient temp of the room, having a large reservoir doesn't change that. Having more radiator space will help the water get closer to ambient, but there is a point of diminishing returns, so much so that it's rarely worth adding more radiator space.