Improving an already pretty quiet build

Got a shopping cart of parts that you want opinions on? Get advice from members on your planned or existing system (or upgrade).

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Improving an already pretty quiet build

Post by quest_for_silence » Sun Aug 03, 2014 2:30 pm

vishcompany wrote:OK, here is a comparison of the Vcore values. Looks pretty similar to me.

Unfortunately that graph is useless.

Broadly speaking, AFAIK the only parameters which may affect the load CPU temp are: vcore, FSB and clock.
Given that benchmark (Y-cruncher) doesn't show any difference, we can say that FSB and clock didn't change.
Whether vcore didn't change, then the CPU temp readings are false, either the ones after resume, or the ones after cold start. Do you agree?

vishcompany wrote:But maybe I'm looking in the wrong places?
https://app.box.com/s/n62b8afxgb78o289j1od
https://app.box.com/s/znllcv43nhv4ajsppjoa

I can't help, I'm currently unable to watch those links, but I will retry later.

vishcompany
Posts: 96
Joined: Mon Jan 20, 2014 1:46 pm

Re: Improving an already pretty quiet build

Post by vishcompany » Sun Aug 03, 2014 3:13 pm

quest_for_silence wrote: Broadly speaking, AFAIK the only parameters which may affect the load CPU temp are: vcore, FSB and clock.
Given that benchmark (Y-cruncher) doesn't show any difference, we can say that FSB and clock didn't change.
Whether vcore didn't change, then the CPU temp readings are false, either the ones after resume, or the ones after cold start. Do you agree?
Interesting approach, but sounds logical.
quest_for_silence wrote: I can't help, I'm currently unable to watch those links, but I will retry later.
That would be cool. Don't waste too much time, though. I'm happy as it is.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Improving an already pretty quiet build

Post by quest_for_silence » Mon Aug 04, 2014 1:04 am

vishcompany wrote:
quest_for_silence wrote: I can't help, I'm currently unable to watch those links, but I will retry later.
That would be cool. Don't waste too much time, though. I'm happy as it is.
Well, eventually I did it, but I didn't find any useful hints among those logged data: perhaps you may try to ask for opinions to some more focused enthusiasts community, like HardOCP or maybe Overclock.net.

On the other hand, you might try to play with K10Stat, in order to save some more °C (if any).

vishcompany
Posts: 96
Joined: Mon Jan 20, 2014 1:46 pm

Re: Improving an already pretty quiet build

Post by vishcompany » Mon Aug 04, 2014 2:31 am

Thanks Luca,
The assumption of false temp-readings seems to be correct. There are two hints:
a) In resume mode the CPU temp when idling is now lower then ambient temp (20° vs. 27°), which of course is physically impossible, as long as I don't sprinkle the heatsink with water :wink:
b) There always was an offset between CPU socket and CPU package temps. In resume mode this offset is practically zero.

I will have a look at K10Stat, just for the fun of it...

EDIT: K10Stat seems to not work with FX-8120, at least I get a dll-error, when trying to start it.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Improving an already pretty quiet build

Post by quest_for_silence » Mon Aug 04, 2014 5:32 am

vishcompany wrote:EDIT: K10Stat seems to not work with FX-8120, at least I get a dll-error, when trying to start it.


It shouldn't be an hw issue, more probably is a sw one, already seen and perhaps permission-related, but unfortunately I'm not aware of any readily available solution.

vishcompany
Posts: 96
Joined: Mon Jan 20, 2014 1:46 pm

Re: Improving an already pretty quiet build

Post by vishcompany » Mon Aug 04, 2014 7:23 am

quest_for_silence wrote:It shouldn't be an hw issue, more probably is a sw one, already seen and perhaps permission-related, but unfortunately I'm not aware of any readily available solution.
No bother, thanks anyway.
I guess I could try undervolting a bit when gaming.

The next step would be to get those fans drop even lower at idle, but this means getting a mobo with true PWM headers and I'm not willing to invest atm. (Don't want to spend even more in an AM3+ platform. No need to jump ship to intel right now)

vishcompany
Posts: 96
Joined: Mon Jan 20, 2014 1:46 pm

Re: Improving an already pretty quiet build

Post by vishcompany » Mon Sep 15, 2014 2:25 pm

UPDATE:
So it seems I picked one of those Seasonic Platinum PSU, which develop coil whine. Under load it's really loud now.

At first I was considering a Corsair CM series, but thought to first try the bequiet PSU (L8 430W) - which was bought originally for this build - with a different fan. It's stock fan used to be quiet for the first 20-30 minutes only, as soon as the temps were rising it would ramp up and produce whining from the bearings. Still had a NF-S12A around, which now is residing inside this power supply:



Image

I routed the fan wire outside the PSU enclosure and it's now daisy chained to the CPU fan. The CPU fan connector is now driving 3 fans: CPU, back exhaust and PSU.

As I had the unit opened anyway I also put one of the kaze master temp sensors on bigger heat sink inside the PSU.

Did a few experiments and found, that this PSU needs proper active cooling with fresh air. It gets too hot if it's turned with the fan facing inside (like in the photo), thus it is turned upside down now.
It switches off at ca. 70°. If the fan is running on a Low-Noise Adaptor, the PSU can get too hot und heavy load.
It's running now at the same speeds as the back exhaust and stays cool enough even under the worst artificial loads. At gaming it goes up to ca. 60°, idle <40°.

Post Reply