Belgar

That’s a lot for an all core load.

Stock is up to 1.5V during single core boost. That's normal. 
 
If you're getting BSOD on stock, that's not normal. 

Yep, that's Ryzen. The all-core loads use the lowest voltages. Single core uses the highest.

Welcome to Ryzen. Their idle is ludicrously high, around 1.42-1.45V. This is normal behavior, unless if its not going down when youre putting it under load.

It really depends on your local ambient temps. In my experience, anywhere from 1°C to 15°C over ambient would be reasonable. So if it's 30°C where you are you might see temps between 31°C idle to 45°C under load.
 
Really, what you should do is monitor and get a feel for how your system performs. Find a max RPM that you are comfortable with the noise and see if that is sufficient to keep things cool, and then setup a fan curve that slowly ramps up to that speed as the fluid temp goes up.
 
Personally, I have my custom loop setup such that my fans start to get really audible around 36°C fluid temp and it never really gets over about 38°C. I have AC so my ambient air temps are like 22-26°C.
 
And for my wife's 280mm AIO I have it setup similarly. Her fluid never gets much above 37°C

It's a different heat sink and will have a different effect on the noise, "50% fan" means literally nothing, and the test was designed to compare air coolers, not fans.
 
The best way to directly compare the fans would be to test them against each other running at the same RPM in open air, and on the same heatsink at the same noise level with the same heat load. AKA do science and actually control some variables.
 
The test as described cannot be used to compare the noise levels nor the performance of the fans.
My hypothesis is that a Noctua fan will perform better under a noise normalized test on the same heatsink, or will produce less noise at the same RPM, but I make no such claim since I'm too lazy to see if any such test has been performed. 🤣
 
The Arctic fans are fine and a good value from what I see others say.
The temperature of the liquid in your loop has a direct impact on the peak temps that you will get with your CPU. Every 1°C increase in fluid temp corresponds to 1°C increase in CPU temp. The fluid tends to take a while to heat up, so if you can run your fans a little faster when idle to prevent the fluid from heating up too much you'll see improvements in temps when under load. Ideally you would set up a fan curve based on the fluid temp, and you wouldn't get those sharp ramp ups and downs in fan speed.
 
From my experience its not too hard to setup a fan curve such that your fluid temp when idle is around 1°-3°C above ambient and under load is 7°-10°C above ambient.

My 2C as someone who runs NF-A12x25s on a radiator: they're worth it. I'm on a pretty small radiator setup (240 for both CPU and GPU) and with proper fancurves my fans are barely audible under load.
 
I don't think push/pull is worth the extra $90 tho, three fans even through a radiator will give plenty of case airflow especially if you have proper exhaust fans.

Push Pull is only marginally better than Push or Pull only when running high RPM, but at low RPM Push Pull will out perform Push or Pull only. So if you want silence and low RPM then Push Pull is the way to go to still get good cooling.

I would start with 3. 6 is still double the noise... and if you think of running 6 of  them at half speed.. they would probably still sound like 3 on a normal profile  Just wasted money imo.

Basically yes. The NOCTUA fans are known to provide better air flow with less noise. I'm sure 6 of them running full speed would still be a bit noisy, but you probably won't need full speed with 6 in push pull.

cool, ive done what you have, ive lock it to ccx not to ccd and I hit 7779 on my first cinebench. No idea how that works but still I gained some performance just by doing that, thanks

so its 10% in productivity difference between auto and 4,35 all core but only 5% in gaming

cool, ive done what you have, ive lock it to ccx not to ccd and I hit 7779 on my first cinebench. No idea how that works but still I gained some performance just by doing that, thanks

I have a 3900x on a x370 gaming pro carbon
 
my setings are
 
oc per ccx 
44 44 43.50 43.50
no pbo 
1000w on the ppt 
amd overclocking seting for voltage
1.36v
-100mv offset
llc lvl 3 , lvl 1 if the top
low riple corrent 
 
 
I get around 7800 on cine 20 
Stock does 7300

ok ok im creating a reminder  

Not installed atm but I can do that. thanks a lot, I didnt really thought of that that will give me real life performance difference numbers. Will update tommorow with results if I dont forget  

Yeah maybe. Still though. Do you have any games with built-in benchmarks that we can compare one setting to the other?

oh I use gtx 1080 and Im waiting for upgrade to new generation. And I game on 1440p. So Maybe im more Gpu bound and CPU matters less?

Hi everyone,
 
Ive been playing aroung with my R9 3900X overclock, since autooverlock was too low for what it could be (4GHz at all core load at max) so I decide to do some manual overclock.
And I have a question I cant really find good answer to on the internet and since this forum was helpful to me before I give it a shot.
I got 2OC:
 
1. 4.35 on both chipsets on everycore
2. 4.425 on first one and 4.2 on second one
 
What I do is mostly gaming or heavy multitastking and my questing is what for what I do is the first one or the second one better?
From my understanding second one should be better in games and worse in productivity (it has lower cinebench score), but im not really sure.
 
Thanks in advance
Have a nice day

 
 

well in single thread for 4 thread games the 7600k may get higher minimum fps, but in multithread games with both cpu's overclocked there would be hardly any difference. 

Yes

It depends on how much the 2600K is overclocked. If it's stock, then a 7600K is much faster. If the 2600K is overclocked to 4.5 or 4.6 Ghz, then it's only a little bit faster, and not worth an upgrade, IMO.

In one VR task LOL, wait for the benchmarks don't repeat marketing. Please.

Of course, that's what all the benches are saying!!!

1 1080 would destroy a 980ti. the 1070 beats the Titan X