my semi-custom loop for 3900x system

I've seen occasional benchmarks on that sort of thing from time to time, but its been awhile since I've looked into it.
 
YOu keep mentioning the fin density but keep glossing completely over superior the tube density in the copper one which would work better IMHO. Slightly lower fin density but more tubes for water to flow = more water flow in contact with air and less air resistance as well.

That's because in the picture is a 3 way 360mm and the 240 is two way. The copper 240's were also 2 way.

The fin density was the same for the 3 way 360mm and two way 240, but to be fair, the ekwb did have 4 extra water paths.


I only bring up the fin density because people seem programmed to shit on aluminum when the headroom on heat transfer capacity in radiators is so large that it is literally arguing over a couple fans having to run a hundred rpms faster. Is 100rpms worth paying 30+ dollars more? To many it is not.
 
Re-ran my temp benchmark i ran before and after setting up the custom loop. This is now without pbo enabled and set to auto and with about 5 days of the loop settling in. the initial test was done the day i made it.

https://openbenchmarking.org/result/1908221-HV-1908170HV01


Basically, no real change in performance with pbo disabled. But my temps now are significantly lower. Pretty sure most of that is disabling pbo, at least on the peak side. But pbo alone wouldn't be able to explain the low frequency temp improvements.

I'll know for sure tonight. I'll re-run it after work with pbo enabled again just to see how settling in has gone.

But for sure, i'm going to be running with PBO disabled as a perm setting. the pro's of it do not outweigh the very real and very harmful cons. Unless it can be used with manual settings in a way that doesn't push voltage so high it produces way more heat than it's worth. I'm not aware of what those settinsg would look like though.
 
And here is a full comparison

https://openbenchmarking.org/result/1908220-HV-1908221HV86

As you can see, the temp drop appears to be totally due to pbo being disabled. Since pbo enabled + auto appears to give no real performance difference, i'm not losing out.

going to run with pbo manual settings 2x scaler and up with 200Mhz top boost and no thermal throttling.

So far, 1x scaler (the default when set to auto) is as described in the bios documentation, effectively not using any pbo. So no effect. Which if other motherboard manufacturers are using the same logic, is probably why nobody sees any difference when set to enabled but all the subsettings set to auto.

2x shows less performance than pbo at 1x or no pbo

still in the process of trying the other scalers.

edit: so i tested via the pbo menu in the amd overclock submenu instead of just at the top level. This menu had pbo option "advanced" which allows you to manually set the wattage, tdc and edc values on top of the normal Mhz setting and scaler manual settings.

Even with those though, I was unable to really impact single core and full all core frequencies over what I can achieve with just plain pbo auto mode. Granted, I had already set the cpu phase to extreme, current limit to 130% and boosted other current limits etc.

I tried a tons of variations


It's hard to tell really if many of the normal current/voltage limit overrides overlap with the behavior of pbo. In any case i'm tired of rebooting and changing settings and seeing nearly no benefit from adding a bunch of extra heat

https://openbenchmarking.org/result/1908234-HV-1908220HV55

I basically settled on 0.1uv nopbo and just pushed the normal power settings for max current for cpu, vrm, etc and load lines etc up a bit. Performance is basically unchanged, but temp is back to reasonable. Pretty much all the pbo settings I changed really did nothing to overall performance, but did drastically impact temperature.

again, this is on an asus pro x570 ws ace
 
Last edited:
Back
Top