Poor cooling with megahalems

[Mellman]

So i got my I5 2500K and saw idle temps around 35*C with a minor OC of 3.9Ghz (all i did was up the BCLK to 105) If playing Crysis i'd see load temps well into 65 C. I decided if iw as going to do any overclocking i needed a new cooler so did some reading and decided to spring for the megahalems.

Installed it yesterday and have left it alone for the most part. in any case, my temps INCREASED! i have a single nexus / yate loon fan on it, with plenty of case airflow and my temps are now 39*C idle and still creep to 65*C load.

It feels like the HS is on there tight, however when installing the backplate the screws holding the backplate to the motherboard were somewhat loose, they could wiggle. Figuring the backplate was fairly secure, i'd see how it felt with the whole thing mounted. the HS locked down pretty tight on the CPU, so I thought perhaps its not a problem.

So i guess my question is for you megahalems users, preferably someone who's mounted it on a P8P67 Deluxe...are the screws that connect to the bracket tight? I have it installed in the order below...

Long Thread
Thumbscrew
Short Thread
Plastic Washer
MOTHERBOARD
o-ring
backplate
rivet with nipple going through MOBO holes

do i really need a dual fan configuration for this thing to truely shine? I expected load temps of 65* C if i really started to OC, but not at virtually stock speeds!

Cheers...

 

[Lightflash]

My first question is why would you up the BCLK when you only need to play with the multiplier?

 

[Mellman]

My first question is why would you up the BCLK when you only need to play with the multiplier?

Why not? partially habit from the old way of overclocking...

 

[Lightflash]

The BCLK now affects more than just the processor. Most people I have seen on these boards are able to keep the BCLK at 100.00 and then boost their OC by changing the multi to around 4.4 with minor adjustments to the Vcore.

I just wonder if this is a potential problem that is causing a high temperature under load. May want to see if lowering the BCLK back to stock and then increasing your multi would lower the temps. Also may want to look into how well your TIM spread when installed if it is just a single core and not the entire processor.

 

[zagood]

I haven't installed on that board, but my first question would be, where did you plug in your fan? Motherboard? Is the speed controlled by the BIOS (or UEFI or UEFA or whatever it's called...)

It seems like you're seeing almost exactly the same temps, so here's another question for you...is it quieter than stock?

What TIM did you use?

 

[Mellman]

for grins i went back to stock BCLK and temps are the same...as expected a minor increase shouldn't bump temps up that much.

i'm aware of what upping the BCLK does; i was running 105BCLK and a 44x multi on stock voltages, prime stable but heated up like a mofo...

spread should be OK but i haven't pulled it to look. fan is plugged directly into the 12v molex; not the motherboard, i do have two PWM fans on the way that will be connected to the mobo and mounted for a push pull, i'll see if that makes any difference. I jsut really expected temps to drop with this beast, they did on my scythe ninja back on 775 days.

i used the prolima tech stuff that was included, i'm sure i could use something better and see a 1*C improvement...but this stuff was reviewed fairly well at SPCR.

 

[RussianSensation]

Most likely poor thermal paste application (perhaps applied too much), poor case airflow, poor installation (i.e. remove the heatsink and make sure it left a nice solid square footprint with the CPU heatspreader) or wrong temperature readings (are you using HWMonitor or Asus software?).

My temperatures are lower than yours for both idle and at load for a 45nm i7 860 @ 3.9ghz. This is basically impossible considering your chip is 32nm and lacks hyperthreading (unless your ambient temperatures are in the 28-32*C range).

 

[Mellman]

my "motherboard" temp is at 30*C, i dont know where that sensor is. it's an antec P180B, with a front fan, top fan, and rear exhaust fan plus one on the HS. I clean the dust screens regularly. 30*C does seem high but i dont know where that sensor is.

I'm using Asus' tools to monitor temps, are they innacurate? I will pull the HS tonight or tomorrow and perhaps re-apply.

 

[RussianSensation]

Try HWMonitor Version 1.17.

Unfortunately, your case is very poor for airflow. So you aren't going to get optimal performance. With a better case, I can see 8-10*C reduction for your CPU load temps.

Do you have a powerful discrete GPU in the system as well?

 

[dac7nco]

I get 82c on CPU 0, and 79c on CPU 1, with a 1GHz overclock (3.4GHz) on 32nm quads in Cinebench 11 x64. The Megashadow is just a black-annodized Megahalems, and I've used Arctic Silver-5 as a TIM. The 5620 is the lowest 2-QPI Xeon with HT (2.4GHz, 32nm, 12MB L3), and runs stock @ 2.4GHz. I'm using Gelid 120mm "nanoflux-bearing" fans in the coolers in a push-pull. This is in a Lian-Li PC-P80, with two and a half video cards and 1/2 airflow from the stock case fans.

I don't know if this is a typical OC temp or not, but it's better than my OC temp on my 5520s (45nm 2-QPI 8MB L3, SR-2 - Alpha). This system is a bit warmer than a typical (that I've seen) 980X @ 4.0

Daimon

Edit: The GTX-580s probably don't hurt internal heating too much, and these temps were side-panel-ON.

 

[Wayward]

I have the same problem. I have a 2600k, Asus P8P67 Deluxe, and a Prolimatech Megahalems rev. B with a Scythe PWM fan, in an Antec P182 case. Idle temps are ~39 C not overclocked.

I have reseated the CPU about 4 times now, and even pulled the motherboard to make sure the back plate is sitting nice and flat. I'm using Gelid GC-Extreme thermal compound.

I had this same setup with an AMD 1090t, and the temps were about 10 degrees cooler than with the 2600k.

 

[Mellman]

I have the same problem. I have a 2600k, Asus P8P67 Deluxe, and a Prolimatech Megahalems rev. B with a Scythe PWM fan, in an Antec P182 case. Idle temps are ~39 C not overclocked.

I have reseated the CPU about 4 times now, and even pulled the motherboard to make sure the back plate is sitting nice and flat. I'm using Gelid GC-Extreme thermal compound.

I had this same setup with an AMD 1090t, and the temps were about 10 degrees cooler than with the 2600k.

were you able to get your screws holding the backplate on tight? mine were wiggling around...the HS itself is sturdy and tight since the synch down points are lower than the CPU but ...but was still concerning.

 

[Infrnl]

I get great results with my megahalems, but I am not on SB yet. I only run single fans at low-med rpm on mine. rund WCG 100% load 24/7. Depending on system I get max temps between 50 on 1055T, 60's on W3520 and dual x5670's. all running at around 4ghz.

On the mobos that have play; I use sm washers to get the play out. These have cooled better than the TRUE's I used to run

 

[Wayward]

were you able to get your screws holding the backplate on tight? mine were wiggling around...the HS itself is sturdy and tight since the synch down points are lower than the CPU but ...but was still concerning.

There is a little bit of play like you have observed. But it is very small and I don't think it has any effect. The back plate is level and the heatsink screws down tight.

I'm tempted to install the stock heatsink/fan to see if there is any difference, but I don't really feel like pulling the mobo again