Does this temp behavior sound right?

Soldato
Joined
26 May 2009
Posts
21,611
Hi, im currently water cooling an i7 3820 which is under a EK Supremacy nickel plexi with some AS MX-4 in between, its fed by a D5 with new EK top (3/8 tubing). At stock the idle temps are a bit higher than I was expecting coming from a Xeon 1245V2 (an i7 3770 for heat purposes), I was expecting higher temps as SB-E runs hotter than IB but im getting around over 10c higher at idle.

Anyway under Prime at stock clocks I can hit nearly 70c which is basically where my 3.8ghz i3 540 system with a Corsair H50 sits in the same test. A full blown water cooling system should be giving better results on a stock 3820 no? As to the temp behavior when a stress test starts the temps will spike up then drop slightly and rise stable until they reach their staying point, when the test ends they drop off nearly 30c instantly then slowly go back to idle.

While the test is taking place heat output isn't anything massive, the tubes are not hot to the touch and neither are all the rads. I was hoping to OC this system but if it can hit nearly 70c in Prime's highest stress test at 3.7ghz (turbo) im worried what it would do at say 4.5ghz.

*EDIT*

It just seems when the stressing kicks off that the temp rises much faster than it used too on my IB, that spiked slightly and gradually raised the temp until it leveled off but the SB-E is raising it much faster and higher, like the heat is coming out faster than can be taken away, but as it drops off straight away when the test ends that can't be right.
 
Last edited:
Soldato
Joined
25 Jun 2011
Posts
16,799
Location
Aberdeen
How much radiator space do you have? What speed is the D5 at? also what fans at what speed?

On my 3820 with a solid 24/7 4.75ghz clock (no speedstep/power saving features enabled) I idle around 2-4c above ambient and IBT stress temps hit around 70c, gaming stress usually sits around 55c.

Also you did remember to take the sticker off the bottom of the block before fitting didn't you? (didn't want that to sound like I was insulting your intelligence, but its easily forgotten about:)).
 
Soldato
Joined
30 Nov 2011
Posts
11,073
I have a 3930k at 4.5ghz on a Supreme LTX and I hit mid 60's
I would definitely say you have an issue somewhere that needs resolving

poor mount, no/low flow... something is definitely up
what are your water temps?
 
Soldato
Joined
26 May 2009
Posts
21,611
Rad space is equivalent to about 5x120 but I don't think its an issue of the loop not being able to dump enough heat to air as the rads/res's/tubes would feel hot if that was the case wouldn't they?. The D5 is on setting two but ive tested it on all settings with prime running and it doesn't make a massive difference.

Yeah I took the sticker off :p this is basically the same setup I had with the IB CPU all ive done is change to a Corsair 500R case, added a XSPC EX240 in the top and swapped the Z68/IB for a SB-E 3820 and an EVGA X79. I was expecting the addition of the EX240 to counter the IB to SB-E heat increase as the CPU's have basically the same stock clocks/turbos.
 
Soldato
Joined
26 May 2009
Posts
21,611
I have a 3930k at 4.5ghz on a Supreme LTX and I hit mid 60's
I would definitely say you have an issue somewhere that needs resolving

poor mount, no/low flow... something is definitely up
what are your water temps?

I don't think flow is the issue as when I was pumping out the air while moving the case/etc I forgot to speed up the pump and even on setting 2 it was doing an ok job so its moving the water at a decent speed. Water temps... not a clue I don't have a temp sensor, but the water before the pump (which then goes straight to the CPU block is neutral to the touch. By neutral I mean if I ordered a cold drink and this arrived I would be upset, but at the same time if my coffee went this cold I would chuck it and make a fresh cup, I suck at analogies lol.

Edit: It feels like a drink left out in room temperature would, ambient like.
 
Soldato
Joined
30 Nov 2011
Posts
11,073
I would suggest re-mounting the supremacy, as it sounds like you just aren't getting the heat away from the chip fast enough

you've got plenty of rad space and you'd notice 10-20C over ambient if that was the issue
 
Soldato
Joined
26 May 2009
Posts
21,611
/FACEPALM

Took off the Supremacy and it did look to have way too much on it, which is odd as I only used a small line like people say too for SB-E, I even looked at a youtube vid showing how different paste shaped (dot/line/X/spread/etc) react under the HS lol.

Anyway as a quick test I wiped off the paste on the HS then put it back on (as the was enough on the cpu for a test and I didn't have time to reapply paste and stuff) and it seems to have dropped the idle by ~10c, will test with prime and see how she handles.
 
Soldato
Joined
25 Jun 2011
Posts
16,799
Location
Aberdeen
Glad you sorted it, I put a small rice grain size of paste on and let the block do the rest :)

Its also possible to fudge up the paste spread if you don't clamp the block down evenly.
 
Top Bottom