|
Hey, just wondering.
Recently my temps got rather high for what i'm using (stock clock i7 6700k + Hydro H115) - around 75 degrees core temp while gaming, with the water temperature around 42 degrees.
Today i kinda got fed up and smacked my tower, which immediately brought the temps down to 52 degrees, with the water only reaching 36.5 maximum. Now, i do realise that there's a seating problem either way which i have to adress (once i get alcohol to clean heat paste off), i'm just wondering if it was properly seated before, or is properly seated now. It kinda doesn't make sense to me - if it was not properly seated before, why was the water temperature so high if the block wasn't entirely covering the lid - or is it not properly seated now, which makes the core temps nonsensical?
|
United Kingdom20150 Posts
Being badly seated would mean that the water temperature was colder and the CPU hotter
You might have problems with the radiator fans running too slowly or not spinning sometimes
I'm using a silver arrow with an OC'd 6700k @1.38v for reference - a 4c8t 100% load (3x cinebench r15 runs back to back) has core temps peaking about 50c above room temp and that's a lot hotter than random gaming loads on stock clocks/volts
|
That's what i'd think, especially since core temps and block temps are measured independently - the block has its own probe.
I can tell that the fans are spinning properly though, can't be that.
edit, waterbubble maybe? Not sure if it could be basically stuck at the pump somewhere and i "bumped" it into the radiator or something? It's an AIO, so i can't really bleed it.
|
United Kingdom20150 Posts
How is your case airflow set up anyway? A generally efficient setup for prioritizing CPU cooling is to have fans intaking air from outside of the case through the CPU radiator so that you have room-temperature air cooling the CPU instead of case-temperature air which can be a lot hotter sometimes
Am not sure about the temperature change thing, it's hard to judge cooling performance in games because the loads can shift around so much. I like to use some easily replicated loads to put hardware at 100% for periods of time to measure cooling performance
|
Before i bumped the tower, and i know that since i was changing vcore voltage because for some reason ROG boards set them way too high, the cpu already was 37 degrees. With cool water in the block, because it was off for 2 hours prior.
Got push/pull, front to back, 240mm front/140mm bottom pulling, pushing through radiator through the top with two high static pressure fans.
I have MSI afterburner open at all times, the second i bumped (or kinda slapped) the tower temps dropped by almost 10 degrees. In idle.
edit: even ignoring that, 80 degrees on a watercooled stock clock i7 even under prime95 is too much.
|
United Kingdom20150 Posts
My board and many others have a "feature" that would ignore intel stock settings and force full turbo clocks on all cores instead of just 1 core. That made it ask for ~1.42v instead of ~1.25v and increased temps quite dramatically, even hotter than it runs now at 24/7 overclock profile. The value that you manually set it to there is quite important
http://cdn.overclock.net/a/a1/a137470f_power.png Higher power = hotter temps, Prime is notably hotter than most 100% CPU loads but far from the hottest out there. The custom x264 & cinebench r15 are quite representative of real and useful CPU loads.
|
That sounds familiar, vcore went up to as much as 1.410v according to hwinfo - manually set it to 1.250v now. Seen people go lower, down to 1.180v, but that's not stable for me, crashed twice.
Wouldn't explain why the temps changed through a physical bump though, i just can't wrap my head around it. I had very high temps on my cooler/cores, properly smacked the tower, and they went down on both cooler/cores. As in, immediately. Not slowly going down - they literally immediately dropped 10 degrees on the cores, water i didn't check - but i know that the game i'm playing daily, where the water constantly hit 42+ degrees, after the smack (we just finished playing), the highest it reached was 36 ish degrees. Nothing changed, no settings - in fact, today was the first time i ran it non-frame limited because of it.
edit: idles lower now too, 28 degrees rather than 37 of before. Water was around 32 degrees, is now 29 in idling.
|
United Kingdom20150 Posts
Core temps are extremely volatile, can go from 25c to 60c in half a second. Water temps not so much, they should change a lot slower and be more representative of the cooler performance. Sounds like something was screwy there.
|
Yeah, that much i do know - would just like to know what, in case it happens again. Don't really feel comfortable accepting that "the russian method" is now my way of getting low temps, hehe.
And yeah, i know they can jump. But the avg is still around "somewhere", and that somewhere dropped by 9 degrees on avg now.
Only thing i could imagine is that it wasn't properly seated, maybe slightly kinked, so the pump sensor (i don't know where it sits) gave out a false reading on the water temp. The radiator pretty much always feels/felt cold to the touch.
|
2774 Posts
If there was water bubbles in your loop you might've heard noise coming from it. Could've been some type of blockage but I highly doubt that, that doesn't really happen with AIO water coolers. Neither does it sound like there's any faulty temperature sensors.
As far as I'm aware most coolant temperature sensors are situated in the intake before the waterblock by the way (or potentially the block itself), although that doesn't really matter considering the heat properties of water.
The first thing to do is to simply reseat the cooler, hard to draw any other conclusions from this really.
Oh and I'd trust something like HWiNFO more for CPU temperature monitoring simply because Afterburner has had some curious bugs and misleading readings over the years and by how it determines the readings. HWiNFO gives you both the CPU core temperatures and socket temperatures.
|
Fair point, but not an overlay in games.
As soon as i get some alcohol i'll reseat/re-heat paste the cooler, can't take it off because i can't clean it properly currently, have to wait for delivery. Although, having HWI and MSIA running next to each other, they seem to state the same temps - just that MSIA doesn't update as fast. HWI seems to tick multiple times a second, MSIA maybe once.
It's idling now at 25 degrees in 19 degrees room temps, that seems like what i'd expect even if it's "just" an AIO. Certainly not 35 degrees like usually.
Sidenote, is there a way to have a "preset" in HWI? I don't want to collapse half the things constantly just to have everything important in direct view?
|
United Kingdom20150 Posts
You can open the sensors tab in HWI and then double click on a stat to track it. You can also change update rates on both MSI AB and HWI iirc
|
You can rearrange everything about the sensors window in HWI. There's a configuration button for this at the bottom of the window. You could move the entries you are most interested in to the beginning of the list.
|
Cheers, didn't know - opened the case yesterday and checken the thumbscrews for the cooler, they were tight-ish, but had a quarter turn roughly left in them via screwdriver.
I have to admit that i actually don't know if there's a "too tight".
|
|
|
|