Sign in with
Sign up | Sign in
Your question

Q6600 / Xigmatek S1283 / MSI Neo2 Fr temperature problems

Tags:
Last response: in CPUs
Share
May 18, 2008 10:46:30 AM

Hello everyone! This is my first post here, so I certainly hope that this is the right section for this type of post (it seemed like the best choice to me).

I'm having an issue with my cores running warmer than I think they should be. All temperatures taken in core temp 0.98.1. The 'old rig' had the following specs:

Q6600 G0 stepping @ Stock 2.4ghz (266x9) with stock Intel heatsink/cooler
ASROCK 4coredual-SATA motherboard
Sapphire Radeon x1950gt 256mb AGP
2x1GB of Geil PC3200 DDR memory
4 IDE/PATA devices (2 HDDs, 1 cd combo drive, 1 dvd player)
Floppy drive
APEVIA [BLACK] ATXB1KL-BK MID-TOWER 350W ATX CASE (configuration for this setup was 2 80mm *intake* fans at the rear, and an 80mm side panel *intake* fan)

With that combination above, all stock BIOS settings with the exception of changing the AGP aperture to 128MB from 64MB, core temp reported average temperatures on the four cores of ~37-43C at idle, and under load the highest I saw was ~50C.


Fast forward to now, with my new rig:

Same processor, same clock, but with a Xigmatek S1283 heatsink instead of the stock cooler
MSI Neo2-FR motherboard
MSI NX8800GT factory OC w/factory heatpipe/fan combo
2x1GB of Mushkin DDR2-800
All the same IDE/floppy drives
Same case, same fans (see below for fan config)

The first config I had was two 80mm intake fans at the front of the computer, blowing air over the drive bays, the side 80mm fan as an intake, blowing directly down on top of the heatsink, and the heatsink fan sucking air across the fins of the heatsink, and blowing into the intake fan of the PSU. With this setup, idle temps are ~45C at idle. I thought 'that can't possibly be right,' so I took it back apart and changed the heatsink direction to it's current position, which is perpendicular to the PSU, blowing air across the fins and out the exhaust holes on the back of the case. With this new config, idle temps did *not* change. I'm at wits end here: I was expecting a large drop in temperature, not a gain! As I sit here typing this message and listening to music in iTunes, core temp is reporting at 46C core 0, 45C core 1, 40C core 2, and 45C core 3.

More relevant info:
1. Arctic Silver 5 is the TIM being used in both setups with the S1283.
2. With the first setup I used the '2 lines' method of applying the TIM. The second attempt I decided to use the 'latex glove' approach and spread out the TIM first, to get a better mating. Both times the little gaps between the pipes were filled with TIM.
3. Heatsink is *definitely* attached properly and securely. Those plastic push pins can be a real pain.
4. Memory is set to 4-5-4-11 2T @ 2.1v, all other voltages are default.

Any help or suggestions would be *greatly* appreciated! I'm all out of ideas. I bought this combo with the intent of running 3.2GHZ (400*8) at a 1:1 with the memory, but seeing such high temps has me worried that I won't be able to.

Oh, and if necessary I can take pictures of the inside of the case.
a b à CPUs
May 18, 2008 11:08:57 AM



Read the other posts relevant to the quad core rigs and temp problems by using the search engine above ... it will answer everything you need.

All of the resident gurus have left heaps of tips before.

If it doesn't then post again and we will help out ... in earnest.

Good luck.
May 19, 2008 5:30:27 AM

Thanks for the replies.

So, after reading through the temperature guide sticky (again), I went about the process of calibrating speedfan. I made all the necessary changes to the bios, except for a couple options: I did not have PECI, Enhanced C1 Control, or CPU Internal Thermal Control options. I set the Speedstep to enabled. When I launched windows and killed the majority of running processes (verified that CPU use was <2%: it was 0), and then launched CPU-Z and Speedfan. After 10 minutes of idling, I checked the temps (see below). One thing I noticed is that speedstep didn't change any settings in the 10 minutes. All CPU multipliers were still at x9, and vcore was steady at 1.302.

Ambient Temp: ~26*C
Tcase Min after 10 minutes: 47*C

One other thing I've noticed is that my Tcase and Tjunction (core 0) are never separated by more than 1*C at idle (haven't done any load testing yet). I know this isn't supposed to be, right?

Using the list in the sticky to determine Ambient to Tcase Delta, my formula looked like this: (X[6] + Y[1])/2 = 4*C, where X is a Q6600, and Y is the S1283 heatsink. So, my Ambient [26*C] + Z [4*C] = 30*C.

Now, if I'm understanding correctly, tcase = Ambient + Z. But in my case, that would mean 47*C = 30*C, which would require an adjustment of -17*! That can't be right. I'd greatly appreciate any help in this issue.

Also, as per Shadow's suggestion, I downloaded RealTemp. At this moment (I just finished playing Unreal Tournament III for about 15 minutes) it reads: 46, 44, 38, 40. Speedfan is reporting 51, 49, 42, 45, with a Tcase of 50. There are no offsets applied to speedfan, and no calibration to realtemp.

If you need any more/different information from me, just let me know. Thank you!

EDIT: Just did a quick load test with Prime95. Using the small FFTs option, core 0 spiked to 75*C in about 20 seconds, according to realtemp. I shut it off when it got that high and realtemp showed it drop back down to 50*C in about 2 seconds.
a b à CPUs
May 20, 2008 6:07:40 PM

MrDomino, the calibrations are not valid unless the Test Setup is followed to the letter. Since your motherboard's BIOS is not properly configured for "Auto" settings, then manually set Vcore 1.10, Multiplier 6 and FSB 266, so that your Q6600 is running at 1.6Ghz with all fans at 100% RPM.

CPU temperature (Tcase) is BIOS calibrated, and there are dozens of processor specific tables coded in BIOS. If BIOS is not correctly coded by the programmers for your MSI motherboard and processor combination, then Tcase will not be accurate. I've seen several consecutive BIOS releases from Asus indicate 15c below ambient to 20c above Core temperature (Tjunction) when nothing else was changed, so the indication you have is not a rare occurance. Keep in mind that BIOS does not effect Core temperatures.

Also, remember that BIOS calibrations (Tcase), as well as Intel factory DTS calibrations (Tjunction) are "canned" values, which can't account for the thermal variables of your rig's hardware configuration, and won't be as accurate as possible until SpeedFan calibrations have been completed. Regardless, if your Xigmatek S1283 is indeed properly attached, and your ambient is 26c, then even with 1.302 Vcore at 2.4Ghz and all fans at 100% RPM, then there's no way your Core temperatures in SpeedFan or Real Temp should reach 75c with a high-end cooler.

This may not be what you want to hear, but since an improperly seated CPU cooler is the leading cause of abnormally high temperatures, my best recommendations are as follows:

(1) Remove your motherboard to be absultely certan that all four cooler pins are inserted completely through the board.

(2) Use the Test Setup as shown in the Guide, but modified with the manual settings described above. Re-test your rig, then run the calibrations and post your results.

Comp :sol: 
May 20, 2008 9:40:39 PM

Thank you! That's a huge help. I'll try re-mounting it once more. I recently read that it's better to apply the TIM to the processor, rather than the heatsink itself, so perhaps that's causing an issue.

I'll post back here in the next day or so and let you know how it went.

Also (and this is most likely related) what would a standard difference between idle and prime95 small ffts load be? When I run it, my core 0 spikes to ~75C within 2 seconds, and then steadily climbs until it stops at ~80C 15 seconds later. Despite this, I keep reading about people with the same cooler and processor only gaining 10C during full load.
May 21, 2008 5:37:59 PM

Okay, a quick update:

I reinstalled the heatsink today, making sure that all push pins went completely through. I used the single line method to apply the AS5 direcly on the IHS. I started up the computer, loaded up realtemp (still uncalibrated) and noticed that I was now idling at 40*C on core 0, as opposed to 47-52*C before. I fired up prime95 small FFTs and my load temps took much longer to reach the higher temps. It stabilized at ~75*C, still too high. So, I put some pressure on one side of the heatsink and the temps rose a bit, so I shifted pressure to the other side. Sure enough, temps started to fall (to about 68*, but were still dropping). When I let the pressure off, temps raised back to 75*.

So, it seems like I'm not getting enough pressure on two of the pins to provide maximum cooling. Anyone have any suggestions? I'd like to avoid a bolt-through kit if possible, but it's not completely out of the question.
a c 127 à CPUs
May 21, 2008 6:16:49 PM

My suggestion is the Zalman CPNS9700. It screws through so yea its a bolt through but thats the best bet to get even pressure across the CPU instead of only one side.

It also has a bracket that screws to the backplate/holder.

Othe possibilities would be the Thermalright Ultra120 and I can't remember the others.
May 21, 2008 6:50:07 PM

ya man somethin is up indeed, i have the same mobo, same CPU, same heatsink, and AS as well. I'm runnin crucial ballistix tracer @ 4-4-4-12 @2.2v. I did the strat jumper OC to 3.0, and idle my temps are around 30 C, the hottest temps i've seen on the cpu W/ a hotroom and gaming, was around 45..

edit:
i switched from coretemp to speed fan also just FYI
May 21, 2008 7:11:22 PM

@Jimmysmitty: thanks for the suggestion, but I'm going to stick with this heatsink. Just might get a bolt through kit.

@bravesfan: Wow. I'm still running stock speeds with way higher temps. Are you using the push pins? Mine are locked in place, and the heat sink doesn't really 'move' per se, but it isn't really completely immobile, either. It seems to flex a bit, which is what I did to lower the temps. When I let go, the temps raise again.
a b à CPUs
May 21, 2008 7:20:59 PM

mrdomino, Vcore and Ambient are the most dominant variables which affect temperatures, so remember to consider that you may not be running the same Vcore at the same Ambient as braves00fan7.

Here's a possible fix; remove the heat sinc bracket from the side that doesn't seat properly. Adjust the center and each end of the bracket by carefully bending slightly, so as to increase tension on the heat since, then reinstall and re-test.

Comp :sol: 
May 21, 2008 9:09:46 PM

Good idea, I'll give that a shot. Thanks!
May 21, 2008 9:33:38 PM

ya make sure you dont have the bracket screws tightened ALL the way down, makes for a near impossible install. I just left the screws backd out a thread or 2, they aren't going anywhere. i'm in a CM690 case that has 3 fans in and 3 out duno if its making that big of a difference though.
May 21, 2008 9:35:59 PM

@bravesfan: Ah, so you are using a bolt-through setup. I'm still considering it because...

@comp: Now that I think about it, my only concern about bending the bracket is the added pressure to the plastic pins. They don't look like the strongest things in the world. Have you had any experience with this?
a b à CPUs
May 21, 2008 10:01:38 PM

braves00fan7, if you're familiar with the Xigmatek S1283, or if you read this thread, then you know that we're discussing push pins and not screws.

MrDomino, yes, I've previously used this method to increase bracket tension. As you've anticipated, the concern is to not distort the bracket to the extent that it increases the tension by more than about 50%, which should be adequate for the Xigmatek to function properly without over-stressing the pins.

Also, visually compare the brackets very carefully to one another to determine how well matched they are. Check the processor as well as the Xigmatek for flatness. Use a straight edge from side to side as well as from corner to corner. I suspect that either excessive IHS / HSF contours are present, or a bracket is slightly bent, or perhaps a combination of both.

Comp :sol: 
May 21, 2008 10:15:33 PM

CompuTronix said:
braves00fan7, if you're familiar with the Xigmatek S1283, or if you read this thread, then you know that we're discussing push pins and not screws.

MrDomino, yes, I've previously used this method to increase bracket tension. As you've anticipated, the concern is to not distort the bracket to the extent that it increases the tension by more than about 50%, which should be adequate for the Xigmatek to function properly without over-stressing the pins.

Also, visually compare the brackets very carefully to one another to determine how well matched they are. Check the processor as well as the Xigmatek for flatness. Use a straight edge from side to side as well as from corner to corner. I suspect that either excessive IHS / HSF contours are present, or a bracket is slightly bent, or perhaps a combination of both.

Comp :sol: 

CompuTronix, if YOU were familiar with the XigmaTek S1283 you would know that the brackets mount to the HS with screws/bolts. No where in my post did i reference bolting straight to the mobo.



@ domino, no i used the pushpins, just dont tighten the brackets all the way down to the hs, is what i meant. The tighter is is the harder it presses on the cpu which isn't needed extremely bad i would say
May 21, 2008 10:22:07 PM

@braves: Ah, wasn't sure which screws you were referring too. I actually think the reason my temps are so high is that I don't have enough CPU/HS contact on one side, so loosening the bolts may not be in my best interest.

@comp: I had thought that might be the case as well. Lapping is definitely out of the question, though. A little too intensive a procedure for me, since I'm still quite fresh in the PC building world, and I don't have the money to replace a processor or heatsink if something goes wrong.
a b à CPUs
May 21, 2008 10:42:46 PM

mrdomino, I can appreciate your situation. Regardless, if you check for flatness on both surfaces, then at least you'll understand more about what the problem is, or isn't.

braves00fan7, I am familiar with the Xigmatek S1283, but I mistook your comments, just as easily as mrdomino did. We often have individuals on these Forums who are eager to help, and have the best of intentions, but are vague and perhaps less than descriptive when offering comments, or just skip to the end of a thread without reading the entire context. I obviously offended you in my effort to clarify the topic, so I apologize, as I meant no disrespect.

Also, if you click on the "Edit this message" icon rather than the "Quick edit" icon, you'll find a checkbox at the bottom for "Delete this reply".

Comp :sol: 
May 22, 2008 2:05:02 AM

bloody hell thats awesome lol never knew that. I'm def. no expert domino but i've read a few places that extreme pressure between the two can have the opposite effect than you would think. I'm not sure though, like i said though, mine are bout a thread or so lose and it mounted a little easier, is still on real secure, and ya kno my temps. I duno man seems like theres something else coming into play.
a c 127 à CPUs
May 22, 2008 4:01:08 AM

chuckm said:
A couple of things, first try this method on the Xigmatek, I've used it twice and it works great;
http://benchmarkreviews.com/index.php?option=com_conten...
And get one of these;
http://www.sidewindercomputers.com/lgbowiscsp.html

I know the Xigmatek will out perform the Zalman anyday.


I am 100% sure that a HDT HSF will out perform a Zalman. I was just suggesting it since the backplate system for it is very easy compared to bolt throughs. In fact it is even easier than the old PGA478 locking mech. But personally I can't wait to see Zalman come out with a HDT version of the CNPS 9700. It already cools nice for not being Quad certified.

braves00fan7 said:
bloody hell thats awesome lol never knew that. I'm def. no expert domino but i've read a few places that extreme pressure between the two can have the opposite effect than you would think. I'm not sure though, like i said though, mine are bout a thread or so lose and it mounted a little easier, is still on real secure, and ya kno my temps. I duno man seems like theres something else coming into play.


Well the problem with yours that I see is that the push pins are not that strong especially on a large HSF like that. So what seems to be happening is lovely gravity and the HSF is pulling down and the push pins are not holding it well enough. For the type of newer tech it is I would expect it to come with a better mounting system but at least it was decently priced.

Well good luck with your quest to get it to cool your CPU better. And if you do go with the bolt through good luck with it too.
May 22, 2008 4:24:43 AM

Thanks for all your replies everyone!

@chuck: That was actually the first article I read about applying TIM. The first time I put the HS on, I used the two lines method, but I think I added too much, as temps were too high. Plus, it glopped a bit: real messy. My best result has been with the AS suggested method of one line across the top of the IHS. Though, I did fill in the cracks of the pipes on the bottom of the HS before tightening it down.

@jimmy: Thanks! I've made a decision:

@everyone: I'm going to get the bolt-through kit that chuck linked to from sidewinder. I was thinking about it today at work, and the hassle of removing the motherboard again is worth it to me. Plus, I'm really uncomfortable with the pressure required to insert these push pins. It makes me cringe every time I need to push down. I'll post back up when I've received the kit, reinstalled everything, and calibrated my temps in speedfan!

@Comp: regarding ambient temperature making a difference, I do believe that for sure. I live in the desert (it was 113* out the other day...) ambient can get a bit warmer than normal.
!