Stock i7 930 w/ Megahalems: cannot get below 42C IDLE. Opinions?

[m0r7if3r]

Did you say they were quiet? ...haha

gt's are about as good as quiet gets as far as i'm concerned

I will definately do reserach on these Scythe GT fans, but have you ever heard Nexus (YL) 1000RPM or Scythe D S-flex models? I guess I can head over the silentpcreview for that

you can't compare a gt1850 to a low speed yate, that's just not an even comparison, it's almost double the rpm. Compare a gt 1150 to it, and for the sflexes, once again you have a low rpm model. I am very familiar with both fans, the sflex is comparable to the gt but imo the gt is better. That said, pick the high speed models and undervolt, that way you have the headroom if you need it, you could do it with a resistor mod if you don't want to use a fan controller.

EDIT: Antec TriCool performance on the Megahalems: Prime95 for 1 hr w/ tricool on high were 70/68/71/66. ambient was probably 1-2C higher than the original attempt)

that seems too high, are you still at stock clocks? if so, i'd say you have 1 of 2 problems (heck, could be both) either you have a bad mount or you have horrendously bad case airflow leading to high case temps.

I am surprised to see no major change in temperature between a Antec Tri-Cool on High 2000RPM @ rated 79 CFM, vs. a Nexus (YL) 1000RPM rated at only 37 CFM. To make it completely fair (ambient temp is the only variable I am not sure about), I will run the test once more with the Nexus.

I'm not entirely surprised, it can only get as low as ambient, the tricool also isn't a very good heatsink fan, its a fine case fan, but it doesn't have enough static pressure to compete with the yate.

EDIT2: I tested the Nexus again on Prime95, didn't wait for an hour, but back to back against the TriCool on high, the numbers are very close 71/69/72/67.

For one degree of temperature rise, I would gladly stick with the Nexus!!!

Can anyone help explain the results?
79 CFM vs. 37 CFM. --- > 1 degree core temperature difference under full Prime95 load?

bad case airflow or a bad mount, you're clearly plateauing. Post a speedfan graph, how quickly does it return to idle temps after prime?

Well said. I may give the 1850RPM Gentle Typhoon a try, but on my previous post, I showed only a one degree improvement from the TriCool vs. the Nexus. That was very unexpected.

1850 gt will blow your yate out of the water, but it might be a bit louder. have you looked at the fan review I posted? The yate loon low is your nexus and the gt1850 is the one that I recommend

I ran an experiment to compare the effectiveness of increasing the airflow across the CPU heatsink to increasing the airflow through the case. My case has a 3 speed controller for five 140mm fans and a PWM fan on the CPU sink. Increasing the speed of the CPU fan from 1100 to 2200 RPM only made a 1C difference at idle, (case fans on low), but increasing the speed of the case fans, (with CPU fan at 1100), made a 4C difference. I'm surmising that, at least at idle, the airflow across the sink is sufficient so that the limiting factor is the mass and fin area of the sink. I'd imagine that under load CPU fan speed would affect more of a difference.

this is a case study...i don't know how much it applies, but i would be willing to bet that you have a similar problem to this guy in that you lack case airflow

 

[CgS Drone]

Thanks m0r7if3r, I wrote up a response three times and then read it before posting and didn't post. I want to respond but every time I wrote it up it it sounded negative and I didn't want it to sound like I was trying to mean it that way.

Simple break down as I see it:
Silent comes at the expense of heat
Small case size is making it difficult for the system to rid itself of heat
CPU fan doesn't have the static pressure it needs

If the test was run with the tri speed both having the side of the case on and off then I would say the tri speed does not have the static pressure needed to work as a cpu fan but then it was never designed to work that way, I had him test it in hopes that it would perform better for him like this but apparently it did not. Sorry my bad.

As Jevery pointed out the cooler manufacturer says to use a higher CFM fan and I believe the GT will provide this cfm and the static that is needed to work as a cooler fan, with the addition of being quiet enough for fireflame. I recommended this one since it is the only fan that I know of that will give all of what is needed. My Noctua fans can put out the air but its at the expense of noise something I am willing to live with but I knew this going into it and I am pushing my AMD system hard so I have to have it this way.

Fireflame I wish I could give you a simple answer but there are just to many things that can be causing this and having you test with what you currently have is to try to narrow it down.

Also when doing the load test you can do a 15 min. test this will usually give you the temps within a couple of C of running and extended test and save you some time.

 

[m0r7if3r]

If you want to eliminate a variable, unmount the cooler, pull the board out, set it up outside the case, then boot it, see how it does in just open air...it's more extreme than the case side off method, but if it performs to spec like that then it will eliminate the base finish/processor contact and processor batch from the equation, if not, then it eliminates the case, case airflow, and environmental variables from the equation

edit: @op, read this http://skinneelabs.com/prolimatech_megashadow.html

 

[fireflame]

that said, your temps aren't too bad for the fan you have, i wouldn't worry personally. What do your load temps look like?

that seems too high, are you still at stock clocks? if so, i'd say you have 1 of 2 problems (heck, could be both) either you have a bad mount or you have horrendously bad case airflow leading to high case temps.

I'm not entirely surprised, it can only get as low as ambient, the tricool also isn't a very good heatsink fan, its a fine case fan, but it doesn't have enough static pressure to compete with the yate.

this is a case study...i don't know how much it applies, but i would be willing to bet that you have a similar problem to this guy in that you lack case airflow

I'm surprised to see you think it's high temperature. Originally you said it was ok, so I thought I was ok, and got the feeling from other responses that I was ok so I started doing some OC (1st attempt) using the guide on this forum. I think so far I'm hitting 77/77/78/74 at 1.3xx vcore, 1.4 VTT, 190 bclock on prime95 blend for 5 min.

Anyway, I've mounted the heatsink 3 times now, and didn't yield any significant difference in idle or prime, did you see my data in my OP? Do you think its too high for stock clock?

I may have case airflow that isn't sufficent for OC, and I've noticed my GPU temp never got cooler than 42-44C when ambient was around 21-22C.
You probably already understand my setup, the P183 case is a very simple design. I pretty much have only 2 extraction fans mounted. One 7200RPM HDD, one SSD. A CP850 PSU which is completely isolated at the bottom section.

I browsed the link you posted. Basically, the megahalems works best with higher speed fans, which is the first review set of data I've seen supporting this.

 

[m0r7if3r]

I'm surprised to see you think it's high temperature. Originally you said it was ok, so I thought I was ok, and got the feeling from other responses that I was ok so I started doing some OC (1st attempt) using the guide on this forum. I think so far I'm hitting 77/77/78/74 at 1.3xx vcore, 1.4 VTT, 190 bclock on prime95 blend for 5 min.

Anyway, I've mounted the heatsink 3 times now, and didn't yield any significant difference in idle or prime, did you see my data in my OP? Do you think its too high for stock clock?

I may have case airflow that isn't sufficent for OC, and I've noticed my GPU temp never got cooler than 42-44C when ambient was around 21-22C.
You probably already understand my setup, the P183 case is a very simple design. I pretty much have only 2 extraction fans mounted. One 7200RPM HDD, one SSD. A CP850 PSU which is completely isolated at the bottom section.

I browsed the link you posted. Basically, the megahalems works best with higher speed fans, which is the first review set of data I've seen supporting this.

initially I thought it was ok, but after getting more familiar with the situation I feel like it's a bit high...that's about the numbers I'd expect in a crowded mid tower...that said, only 2 exhaust fans might be hurting you, could you try one on intake? Basically what changed is that you're doing everything right, so the normally simple changes we recommend didn't apply here, and you were still at too high of temps.

 

[fireflame]

If you want to eliminate a variable, unmount the cooler, pull the board out, set it up outside the case, then boot it, see how it does in just open air...it's more extreme than the case side off method, but if it performs to spec like that then it will eliminate the base finish/processor contact and processor batch from the equation, if not, then it eliminates the case, case airflow, and environmental variables from the equation
[/url]

Instead of doing that, basically, I removed the side cover off, took a giant box-type room fan and set it right against the side of the case, set it on high for 10 minutes. Result? GPU 42, Cores 42/40/46/40. Currently ambient is 27/28C.

After this test, I replaced the side cover. In 10 min, GPU temp is 49C, Cores are 44/43/48/41.

From this, I can say that better airflow (in proof the room fan experiment) in the case may improve my core temps 1-2C, but I am now getting the feeling that this is the normal operating temperature of the combination between my motherboard, CPU, case, and components and I cannot improve it more than a couple degrees.

here is my setup in the P183 (dual exhaust fans, no intake fans. PSU is completely separated in its own chamber.

 

[CgS Drone]

fireflame the only problem that I see is when you took the side cover off you keep referring to the idle temps, sorry not trying to be rude so don't take this that way. You need to let that go, at this point they don't mean anything. Its the load temps that matter. When you use your computer those idle temps will jump and thats whats important, this is why we load test computers for worse case scenario's.

If you take the side off of the case put the box fan on there and load test the system for 15 minutes (OCCT and Real temp I think its called for the intels) then do a screen grab and show those temps. then do it with the side on. this will tell you if it is air restriction that is causing the temps.

 

[CgS Drone]

Here was my idle temps on a H50 that was giving me problems. When I first installed it the temps where great by the end of the third day I was hitting 95C and having to have my computer throttle. It was fine even with the cooler being messed up at idle it was at load I couldn't even do a screen grab because I had OCCT set to shut off at 95C and it hit it FAST.

 

[fireflame]

fireflame the only problem that I see is when you took the side cover off you keep referring to the idle temps, sorry not trying to be rude so don't take this that way. You need to let that go, at this point they don't mean anything. Its the load temps that matter. When you use your computer those idle temps will jump and thats whats important, this is why we load test computers for worse case scenario's.

If you take the side off of the case put the box fan on there and load test the system for 15 minutes (OCCT and Real temp I think its called for the intels) then do a screen grab and show those temps. then do it with the side on. this will tell you if it is air restriction that is causing the temps.

So I performed the 15 minute Prime95 and monitored temps with Real Temp, back to back:
Ambient around 26-27C
with case side panel on (normal): 70/69/71/67
w/o case side panel, box fan: 68/66/69/64

based on these results, and the unrealistic airflow by the box fan into the case, can i rule out that it's the case flow which is causing my core load temps to be high?

EDIT/ADDITION:
I had a 4th remount attempt, this time with AC5 in their instructed method, and the same as my 3rd attempt with PK-1 TIM.

Ambient was similar, around 27C
Under 15 min load i got 73/72/74/70

that's 3 degrees higher than the PK-1 TIM yesterday. I know AC5 has a break in period, but I expected only a 1-2C change.


---------------

I wanted to add a response from Prolimatech regarding the Megahalems:

Thank you for emailing us.
Here are the answers:
>> 1) What is the best method of application of the thermal compound?
>> (spread, line, center pea). Does it makes a difference?
Please spread the grease evenly both on the base of cooler and cpu surface with card. In that way, we can make sure that the contact between CPU and the base of the cooler is 100% okay.

>> 2) Is it true that your thermal compound is the most compatible with
>> your Megahalems HS? why?
Yes, it is. The particle size of the PK-1 is designed for the base of Prolimatech coolers for better contacts. That will help the thermal conduction more effective.

>> 3) I discovered my megahalems revB has a very slight concave shape
>> (less than .2mm i would guess) using a high precision straight edge.
>> I understand that the i7 (1366) core location fits in line with this
concave shape. Is this true? Can you elaborate?

Yes, it is. The slight concave shape is for the better contact with cpu.

 

[m0r7if3r]

Well, the bit about their paste being optimal is dubious at best imo...I'd say check to see where the concavity is failing, because the contact's definitely not great in the base shots you posted. Maybe you lap your cpu, maybe the hs, maybe both. That would be the route I would pursue.

 

[fireflame]

I unmounted my 3rd attempt (i.e. tinting + AC5 vertical line method), as seen on the last photos on my first post. I used a drop of red food coloring on a glass coffee table.

If the Megahalems base isn't precisely machined, I don't know what is. It shows to have a very strategically placed concave (bump) in the exact center, but it is extremely subtle, as I mentioned in my first post.

The CPU doesn't look like a lapping would be worth the risk/effort. The lighter amount of food coloring you see, in a kind of hourglass shape on the last photo, seems to resemble the TIM pattern on the first photo, I probably didn't clean it thoroughly enough with IPA.

Opinions?

 

[CgS Drone]

fireflame there is your article and the other article that I linked to. Now I noticed today that there is another one about people talking about high temps, all of these have been on 930 cpu.

Have you compared your batch number with other batch numbers?

 

[fireflame]

fireflame there is your article and the other article that I linked to. Now I noticed today that there is another one about people talking about high temps, all of these have been on 930 cpu.

Have you compared your batch number with other batch numbers?

nope, I'm brand new to every aspect of this process. Good to learn, i guess!

it's number on the last row of the stamp on the IHS, right?

 

[CgS Drone]

Should be, I just did a fast search and I think they are starting with the #9 but that was the page I looked at.

 

[Conumdrum]

Has anyone mentioned idle temps mean little?

Has fannage, ambient temps, case airflow been discussed?

EDIT, just one fan and WHAT fan on that heatsink?

No intake fans and the filter still installed?

Mora Mora fannage dude.

 

[m0r7if3r]

nope, I'm brand new to every aspect of this process. Good to learn, i guess!

it's number on the last row of the stamp on the IHS, right?

yea, and they're often referred to by just the letter and numbers following it, so search that (ie: golden batch is a187 or 3001a187

 

[fireflame]

Well, since I can't really do anything about my CPU other than know if its a good batch or not, I'm still trying to address my heat issues.

So basically, last night, I made a spontaneous compulsive decision, now I have in my hands, an Gentle Typhoon 1850RPM, Shin Etsu microsi TIM, and an NH-U14.

I'm bascially going to max out the potential of the Megahalems (w/ the GT & new TIM), measure the results. If the load temps aren't to my satisfaction, I'm going to repack the HS and send it back to newegg that is, after I check the performance of the NH-U14.

Results most likely tonight. thanks to anyone sticking around this thread.

 

[CgS Drone]

It will be interesting to see the results of the GT on it. Please post some screen shots of it under 100% load. Thanks.

Also let us know if you think the GT is loud.

 

[Conumdrum]

So basically, last night, I made a spontaneous compulsive decision

SNIP!

Glad it wasn't a spontaneous combustive compulsive decision!

Yea, sometimes you just have to order stuff to make it right. Trust your brain then your heart, then 15% of the Internet (us guys).

Keep us posted.

 

[fireflame]

GentleTyphoon 1850RPM (AP-15) as HSF results (vs. 1st config w/ Nexus RealSilent)

I wasn't consistent in the screencap on the 1st (nexus) run, don't mind the current temp or cpu load, please only look at the high temps.

Ambient was about 23C:

Nexus: 72/71/73/69
GT1850: 69/67/70/65
GT1850(box fan): 67/65/68/63
GT1100: 69/67/70/66
Nexus (2nd try): 71/70/72/68

- It looks like the GT reduced core temps by a nice 3C @ 1850RPM
- Open air (ideal case flow) reduced core temps by another 2C
- What's pleasantly surprising to me is the nearly identical core temperatures between GT@1100RPM & GT@1850RPM
- Also notice the nice reduction in GPU temps while using the GT as the HSF