Heat issues YET AGAIN with the 4770K

Toggle sidebar Toggle sidebar
Andrew Buck

Andrew Buck

Honorable
Jan 28, 2014
2,252
0
12,460
I have had this processor for about 2 months now as well as my H105. Right now, I found pretty much my maximum stable at 4.15 GHz @ 1.23 V. The only issue is that my CPU gets to 86-87 C instantly on FPU test. I don't understand why. I have reseated the CPU and put on Arctic Silver 7 and made sure it is TIGHT. I still get up there. In my house, it is about 24 C and the radiator is actually outside of the case due to incompatibility. The case was modded to fit it on top:

2jfa9aq.jpg


Dust has been cleaned out and everything. I know what I am doing in terms of OCing and my voltage is completely set at 1.23 Volts. I have reset BIOS to default a few times now. Latest BIOS is installed. The pump is working. I idle in the mid 30s (high for idling on water, I am the same at stock 3.5) and load at around 87 C. At stock, I load at around 75 C. Last time I checked, this was extremely high. From benchmarks of this cooler that I have come across, this cooler doesn't let the chip hit 75 C at 4.5 @ 1.3 V. I hit 75 at stock at 1.15 V. This doesn't make sense to me. Airflow is fine, no dust, good thermal paste, radiator gets a little bit warm (not much) on load with fans at 100%, I just don't get it. This is the first CPU the rad was used with. I want to know why I am idling so high and loading so high. Thank you, Andrew.
 
Sort by date Sort by votes
Probably a full third of that radiator, at least, is blocked by the top of the case. Are you using one fan or two. Are the fans set to exhaust or intake through the radiator? My recommendation, regardless of which direction you have the fans facing, would be to mod the case to remove all the sheet metal that is blocking the fins on the radiator and use two fans. This might require a bit of work but it will certainly be more efficient. I could be wrong about the blockage but it looks like a significant amount of fin area is blocked from direct air flow in the picture.
 
Upvote 0 Downvote
Since you have it outside the case anyhow, you might even want to just configure the radiator to stand up divorced from the case and use two fans. You might need extensions for the power cables to the fans. You could also then use an additional case fan to exhaust from the vent the radiator is currently blocking. That would provide a superior solution because you would then be using the cool outside air to cool the radiator plus you would not be venting the heated air from the radiator INTO the case, and, you would have an additional exhaust fan which can provide as much as a 5-10c internal case temp.
 
Upvote 0 Downvote


I agree with Darkbreeze. I would also add that a push/pull configuration with fans and the radiator will give your cooling the best efficiency.
 
Upvote 0 Downvote


The fans are exhausting air, but the temp inside the case is 31 C on my HDD, but it is actually around 26 C in the case. The case metal underneath has been completely removed except for the low density wire mesh, so it isn't being blocked. The radiator cannot stand up outside the case without a lot of structural modding, and the tubes may not even be long enough for that. 2 fans are running. There is not an airflow issue in the case, as there are 4 fans in the case that are running and nothing is blocking them aside from the radiator's 2. The temp is always perfectly fine. I have 2 exhaust fans and 2 intake fans active. The benchmarks of the cooler online also do not include a modification to the orientation of the radiator, so that wouldn't make sense.
 
Upvote 0 Downvote
Can you take a pic of the radiator from the top looking down into the radiator or from inside looking up into it. It really looks like the ends on each side have about an inch to an inch and a half that are blocked. I realize that doesn't seem like much, but if it's true it could contribute to a lack of efficiency. Also, I'd completely remove as much of the honeycomb as possible. Vent structures, depending on the type, can decrease airflow by as much as 50% as shown here:

http://www.pugetsystems.com/labs/articles/Effects-of-Grill-Patterns-on-Fan-Performance-Noise-107/
 
Upvote 0 Downvote
Yeah, it's very hard to see anything from the pics. I guess my whole point is just to make sure the hole in the top of the case is as large as the fin area on the radiator. If possible, remove as much of the honeycomb grill as you can without compromising being able to mount your hardware.

You might also try to raise the radiator away from the case itself a bit with some rubber washers and then seal around the gap between the radiator and the case, kind of like a shroud. You could use silicone or weatherstripping, or get a shroud. They sell them and they do help. You might also try adding two more fans of the same speed or faster, the outer ones pulling don't have to be high static pressure, as the ones pushing.

Also, it's beneficial if the ones pushing are of exact type and speed. If the radiator fans are on PWM, you might change them to 3 pin connections.

And I can see grillwork to each side of the radiator, on the case. This is probably allowing some of the air pressure created by the fans to escape past the radiator without passing through the fins which decreases efficiency. Try sealing that up so that all of the air flow from the fans goes through the radiator.
 
Upvote 0 Downvote


Back in June when I got my system, I ran at expected temperatures. I will try to seal the gap. Until then, why do you think my CPU is heating up so much? It baffles me.
 
Upvote 0 Downvote
I can't find an exact spec, but I've seen cpu's heat as much as 10c per second under load with a stock cooler so that's not surprising. What's surprising is that the cooler isn't handling the heat generated by the load. What do you have the pump power connected to? Is it on a 4 pin PWM header, a 3 pin header or on a 4 pin molex straight off the psu?
 
Upvote 0 Downvote


It is connected to SYS_FAN6 (not sure why), which looks like it is a 3-pin connector. It is not a PWM connector, though. It is only a 3 pin connector. My BIOS only goes to SYS_FAN4, so it may be on normal speed like SYS_FAN1-4 and CPU/CPU_OPT. That couldn't really be due to it not being PWM, though. Could it possibly make a difference?

EDIT: Something I noticed: I am idling right now (<5% CPU, going down to 0.5 at times) and it feels a little bit warm. Not too warm, but warm.

Yet another edit: I did some research and found that it can be controlled on 3-pin connectors on newer mobos. Not sure if mine can, but I am assuming it can because it is Z97 and came out only months ago.
 
Upvote 0 Downvote


Plugged into CPU and no difference. I do see that it is running at the specified 1850 RPMs. I tried a quick Power Savings setting and reset to defaults run and I got to 27 on the lowest core and 30 (lowest)-35 on the hottest. Keep in mind this was with SpeedStep bringing it to 800 MHz and 0.75 V. On reviews, they idle lower than this at 4.5 GHz at 1.3 Volts. I still don't understand this.
 
Upvote 0 Downvote
otoluh.png


IDLING. Those 50 C temps are normal for starting up when I get like 15-20% load. Not sure why so high, but whatever.

2lay3d5.png


This is full load. First time in a while with Prime95. Ended up getting to 93 C after about 3 minutes. At 1.23. 4.15 GHz. Don't pay any attention to those PSU voltage readings. It is a misreading and in Gigabyte Software and BIOS it is correct.

Something else I noticed - look at the Powers section. That is extremely high. Not sure how high it actually is, but that TDP is high. Anybody else with a 4770K, wondering if you could Prime95 at my settings and show how high of a TDP you have. Not sure how reliable that is. In perspective, when I had the balanced settings with a BIOS wipe, it was at around 4 Watts idle. With this, it is at 40+. How?
 
Upvote 0 Downvote
Holy... MAJOR discovery. I closed my case and installed the side case intake again. The room has some doors (closed a lot of the time) and a ceiling fan that is mostly on. The AC is always running. I found out that I guess it cannot keep up as well with my CPU. In the room next to this one, it is around 76-77 F. In this one, it is anywhere from 80.5-83. I measured the exhaust temp on the back - a whopping 97 F! Not even sure how. This thing throws out SO much heat! It is 82 F outside right now, might as well open up the window in the morning when it is 65 F. That should hopefully bring the temp down. I think that what is throwing off so much heat - Large Flatscreen TV with a cable box. Xbox One on for the TV for now (that really throws of some heat there), a MacBook Pro used by my brother for gaming (gets pretty hot), lights, the monitor, and this beast of a machine. One word - summer. In the winter, hopefully we can open the windows and 40-50 F air can come in, dropping temps by around 20 C. Either way, though, it should not be getting that hot. About the 97 in the exhaust - the radiator only measures 85... It is exhaust, too. Huh. That would explain the rather high other temps, but still - that is too much even for that temp. Also, within a few minutes, [strike]the case temp dropped to about ambient (83 F exhausting)[/strike]Derp, the temperature monitor's cap was on, still 97. Maybe I should reconsider my PC location...
 
Upvote 0 Downvote


Wrong. I'm talking about the four pin molex, not the 4 pin PWM cpu header. You know, like the type of plug that you used to use to plug in to the back of drives like DVD burner. It looks like this and delivers full time 12v which is what that pump needs. I agree the room ambient temperature matters but you should still be able to keep it cool. So, try this and if it doesn't change things, you can always change it back. Connect your fans on the radiator to the three pin CPU_fan if there is one. If not, use a three pin case fan header. Doesn't matter which one's, just whichever ones you can reach. Preferably with a Y splitter so they are both on the same voltage level and so spin at the same speed. Hook the pump up to 12V molex coming straight from the power supply, like this:

http://www.dtekcustoms.com/ProductImages/products/addons/3-4%20pin.jpg

The small end will connect to the pump lead and the large end will connect to one of the molex connectors coming from the power supply. It sounds to me like the pump is only getting 7v from the board and is not going full speed. Either that or the fan configuration on the radiator needs to be switched to the fans on top as intake. I've always been against that arrangement but after reading several (Many actually) test results comparing both arrangements recently, I'm beginning to change sides on this. The idea with that is that cooler air from the outside is better for cooling the radiator and thus the cpu and that the amount of heat coming off the radiator into the case is negligible because the case exchanges air at the rate of more than once per second. Of course, you need to make sure you've got sufficient exhaust fans. Before all that, try the pump on full speed though and make sure to seal up those open areas around the radiator.
 
Upvote 0 Downvote
How much thermal paste are you using on the cpu and heatsink. If you're using more than a rice sized amount, you're using too much and should repaste. Too much paste can cause a significant temperature problem. So can too little. A rice sized dab is good. A pea sized dab is too much.
 
Upvote 0 Downvote


Not sure that I have a 4-pin Molex to 3-pin/4-pin fan converter. I will look, though. As for fan switching, I can't do that. [strike]The fans outside of the case would not really work and could be a little bit dangerous (we have cats, we don't want them to lose tails).[/strike]Just though about that, I could just flip over the fans -_- long day. For thermal compound, the other night, I put a little smaller than pea sized. The temperatures are the same as they were with the stock thermal compound on the H105. Isn't 1850 RPMs a good level for the pump or should it be more?

Edit: I DO have some 4-pin molex to 3-pin fan power. One thing I noticed - on both, there is a red and black cord and no other pin, so it is two pin. This is obviously positive and negative. Would that affect much? Also, the SP120L fans are on the Y splitter. Switching out the pump and cords.

RESULTS:

Nothing improved. My temperature increases by, on average, 0.538 C per 1% of load. That translates to hitting 94 on full load after around 4 minutes.
 
Upvote 0 Downvote
No, that should be fine. The third wire is for voltage control or monitoring rpm. Flipping the fans over isn't a good solution because apparently, and this goes against everything I've known in 25 years of working on automotive cooling systems, they cool far better when pushing than when pulling. I've seen the results to support this and while I don't understand it I can't argue with the fact that it's true. If you move the fans outside on top of the radiator you can get fan grill covers like these: http://www.newegg.com/Product/Product.aspx?Item=N82E16811998059 to keep the cat's safe.
 
Upvote 0 Downvote
I don't know then. Switch it all back to how you had it except the pump power, that should stay on the Molex. How many intake fans and exhaust fans do you have, not counting the radiator fans? Maybe the cooler is bad or you need to repaste using the rice grain sized method.
 
Upvote 0 Downvote


I have 2 intake in the front and 1 on the side. In the back, I have the LED fan as exhaust. I then have the 2 rad fans as exhaust. I corrected the temperatures I am getting to 8 C (I am around 27-28 C in here right now, in the fall and towards the winter, I get to 18-20 C and that is more of what I am getting. Correcting by 5 C for 22 C is still a bit high, though. My worry is the high load temp. Is there a good way to test the cooler or the CPU? I may be able to return it to Intel, because surely with the stock fan I would exceed max temps at stock settings. Not sure how I feel about 4790k's because I have helped solve tons of heat issues for them that resulted in RMAing the CPU.
 
Upvote 0 Downvote
Sheet metal and silicone.

1/16" plastic sheet material and silicon.

A strip of heavy duty garage door seal, the kind that looks like this would work:

weatherstrip.jpg


Two pieces of wood, one on each side shoved up against the radiator would be better than nothing, at least temporarily.

Any of the above solutions, among surely many others, would work. Duct tape? Heh. Ok, that one's a little bit redneck, but I imagine it would work.

By the way, you can get that weatherstrip at Lowe's or Home depot, or probably any hardware store. Even some plexiglass might not be bad. Just make sure whatever you use you seal the edges that contact the case using 100% silicone or even better, some Lexel. That's the best caulking you'll ever use. Better than silicone and also paintable, whereas silicone is not.

http://www.lowes.com/pd_455441-13518-13010_0__?productId=4714985&Ntt=caulking&pl=1&currentURL=%3FNtt%3Dcaulking%26page%3D3&facetInfo=
 
Upvote 0 Downvote


I will take this into consideration, thank you.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom