More powerful fan on an H7?

[aisle9]

I'm happy with the speeds my stable overclocked 4790K is getting--4.6GHz at 1.24v--but even with an H7, my temps are getting up there. Peak temps during OCCT pinged 88C on a couple of very brief occasions during the 2h test, with peak temps during RealBench stress testing hitting 84-86C for a few seconds on a few occasions. Both programs had typical temps of 78-80C under load, with Intel XTU barely breaking 80C (if it did at all). Because load temps during gaming tend to peak around 65C, I'm ok-ish with those stress test temps, as I know I'll probably never hit them again.

That said, would I see any cooling gain by replacing the stock Cryorig QF120 heatsink fan with something a little more powerful, like the Noctua NF-F12 PWM or another 120mm high static pressure fan, or is the heat dissipation more limited by the heatsink than by the fan?

I'd buy a second QF120 and strap it onto the back end of the heatsink, but I've seen multiple places (including here) that a second fan will buy you anywhere from half a degree to a couple of degrees, and won't be worth the cost. Is that a good assessment?

Edit: my case has a max heatsink height of 155mm. If I could have put a DH-14 or DH-15 in there, trust me, I would have.

[Guest]

What are the temps like without the OC?

 

[Deli]

I think you will see a few degree improvement at most. Unless you are willing to crank up the fan speed and produce a lot of noise. Noctua fan cannot do magic.

[aisle9]

27 minutes ago, Alexokan said:

What are the temps like without the OC?

Right around 78-80C peak at stock.

13 minutes ago, Deli said:

I think you will see a few degree improvement at most. Unless you are willing to crank up the fan speed and produce a lot of noise. Noctua fan cannot do magic.

I'm not looking for magic. I'm looking for 85C peaks during OCCT

[stconquest]

4 minutes ago, aisle9 said:

Right around 78-80C peak at stock.

I'm not looking for magic. I'm looking for 85C peaks during OCCT

What is your cooling solution in it's entirety?  (case & case fans)

[Guest]

Sounds like it's mounted wrong, etc.

Those temps seem way too high for that circumstance. 

[aisle9]

Just now, stconquest said:

What is your cooling solution in it's entirety?  (case & case fans)

Thermaltake Core V31, H7 with stock fan. The case came with two fans, mounted low in front and high in back. I added an Aerocool DS 120mm to the front and two b-Blaster 140mm fans to the top of the case, which has great ventilation in front, in back and on top (and dust screens, added bonus).

The forward b-Blaster and front stock case fan are set to turn off entirely below 55C. The rear b-Blaster, which is directly above the CPU and heatsink, the front DS and the rear stock fan stay on at all times, 60% below 65C cranking up to 100% at 75C. DC fan control limits me on what I can do to ensure good ventilation while keeping the case quiet.

[Deli]

1 hour ago, aisle9 said:

Right around 78-80C peak at stock.

I'm not looking for magic. I'm looking for 85C peaks during OCCT

Ask yourself. After spending $xx for two NF-F12 s(push-pull) and see your CPU load temp drops by 4C. Do you think it's worth it? If your answer is yes, then go ahead.

P.S. I use Noctua fans, but not the standard brown version. Can't stand the color. I've PPC and Redux. They're great.

[aisle9]

1 minute ago, Alexokan said:

Sounds like it's mounted wrong, etc.

Those temps seem way too high for that circumstance. 

The thought had crossed my mind. Is it possible the thermal paste just hasn't baked in yet? Or maybe it's possible that I just got a bad tube of MX-4 and should try again with AS-5? I have heard of that happening.

I did a thin, 1mm wide line of MX-4 vertically, along the center of the chip to cover the cores, extending slightly longer than a grain of rice. I followed the directions on the cooler to the letter, so I can't see that installation being an issue. I'll remove and reseat though, if it seems like that could be a problem.

[stconquest]

7 minutes ago, aisle9 said:

Thermaltake Core V31, H7 with stock fan. The case came with two fans, mounted low in front and high in back. I added an Aerocool DS 120mm to the front and two b-Blaster 140mm fans to the top of the case, which has great ventilation in front, in back and on top (and dust screens, added bonus).

The forward b-Blaster and front stock case fan are set to turn off entirely below 55C. The rear b-Blaster, which is directly above the CPU and heatsink, the front DS and the rear stock fan stay on at all times, 60% below 65C cranking up to 100% at 75C. DC fan control limits me on what I can do to ensure good ventilation while keeping the case quiet.

Well you have good airflow, so it is really just the heat sink.  As the man above me said:  Do you think it is worth it?  (money).

Side note:  You have negative air pressure with those 140mm in the top.  Even though you have filters on vents, dust gets pulled in through every crack.  Can you jury rig one of those 140mm in the bottom as an intake (if there is a vent in the bottom next to the PSU vent).  It will just keep your PC a little cleaner.

[aisle9]

Just now, stconquest said:

Side note:  You have negative air pressure with those 140mm in the top.  Even though you have filters on vents, dust gets pulled in through every crack.  Can you jury rig one of those 140mm in the bottom as an intake.  It will just keep your PC a little cleaner.

My case is sitting right on the carpet, and there's really not a lot I can do about that right now. If reseating the heatsink solves it, I might just set one of those 140mm fans to stay off except at extreme (70C+) temps.

[Guest]

I mean, just for a reference the quick checks on temps with that processor with that cooler, it should idle around 30 and get to a MAX of ~75c. You're quite a bit off from that, so unless you have a dead spot of air flow in your case or overall confined and uncirculated air, you should be getting better temperatures. I would check different thermal paste and remount it. 

[aisle9]

Well that was weird. When I seated my cooler the first time, the MX-4 made like the Red Sea and parted down the middle. I'm not kidding--it parted down the middle and 80% of the paste went away from the cores, with only about 20% left on them.

I reluctantly cleaned, re-pasted with MX-4 again using the pea method instead of the line, and reseated. I'm sitting around 30C idle right now, and will be firing up an OCCT test after getting an idle baseline. If it still runs high, I'll have to clean everything and repaste again, this time with Arctic Silver 5, which I know works.

[aisle9]

*sigh*

Does MX-4 have a burn-in time like AS5 does, or is it pretty much full performance right away? I turned off the OCCT test at 22m when peaks started to ping north of 85C.

*Edit: I had to bump voltage to 1.245. OCCT crashed on the first try

[aisle9]

3 hours ago, Alexokan said:

I mean, just for a reference the quick checks on temps with that processor with that cooler, it should idle around 30 and get to a MAX of ~75c. You're quite a bit off from that, so unless you have a dead spot of air flow in your case or overall confined and uncirculated air, you should be getting better temperatures. I would check different thermal paste and remount it. 

Yeah, I might just have a bad syringe of MX-4. After a couple of heat up, cool down, power off cycles, it's still pinging 85C after 15 minutes of OCCT, just as bad as before. Would you suggest giving it time to burn in, or just cut my losses and reseat with AS5 or the CP7 paste included with the H7?

[Guest]

12 hours ago, aisle9 said:

Yeah, I might just have a bad syringe of MX-4. After a couple of heat up, cool down, power off cycles, it's still pinging 85C after 15 minutes of OCCT, just as bad as before. Would you suggest giving it time to burn in, or just cut my losses and reseat with AS5 or the CP7 paste included with the H7?

I have no experience with bad thermal paste, but I would try all available options. 

Those temps are still a bit high. I'd be more curious to see what the temps were without the overclock under load - as that is more documented than a specific overclock with that specific cooler. 

I would also do a circular blob rather than the line since it split the way it did. 

The glory of troubleshooting  

[Deli]

13 hours ago, aisle9 said:

Yeah, I might just have a bad syringe of MX-4. After a couple of heat up, cool down, power off cycles, it's still pinging 85C after 15 minutes of OCCT, just as bad as before. Would you suggest giving it time to burn in, or just cut my losses and reseat with AS5 or the CP7 paste included with the H7?

I used MX-4 before. Don't remember there's a noticeable turn-in period. Among the well reputed TIMs, the difference is only 1-2C at most. Your CPU is running 10C above normal. It could be more than a tube of bad TIM. But it doesn't hurt to try AS5. At least it helps to eliminate one suspect.

[aisle9]

2 hours ago, Alexokan said:

I have no experience with bad thermal paste, but I would try all available options. 

Those temps are still a bit high. I'd be more curious to see what the temps were without the overclock under load - as that is more documented than a specific overclock with that specific cooler. 

I would also do a circular blob rather than the line since it split the way it did. 

The glory of troubleshooting  

I swapped it with AS5 last night. The MX-4 had, once again, spread everywhere except over the cores when the heatsink was seated. I didn't have time to do much stress testing last night, but even at 4.6GHz and 1.245v, there was some definite improvement over the MX-4...just not as much as I'd have liked. I'm hammering it with OCCT at stock speeds as we speak, while I'm in another room working...and 15 minutes in, it hasn't gone higher than 60C, with the graph suggesting that it's typically staying around 55C.

2 hours ago, Deli said:

I used MX-4 before. Don't remember there's a noticeable turn-in period. Among the well reputed TIMs, the difference is only 1-2C at most. Your CPU is running 10C above normal. It could be more than a tube of bad TIM. But it doesn't hurt to try AS5. At least it helps to eliminate one suspect.

I've got a Vietnam chip, and that was by request. I'd read in a few places that Vietnam chips tend to OC better at lower voltages, but they tend to be some of the hotter-running 4790K's. Thus far, that's proving to be the case. AS5 has a 200-hour burn-in time, so I'm not expecting miracles on day one, but I will know after running stress tests at stock speed whether this chip is ok or needs to be taken back to Micro Center (40 miles away...hope not) and exchanged.

Side note, and possibly a relevant one: since overclocking, Core Temp does not work on my PC. It freezes the PC forcing a hard reset, or gives me an incompatible driver BSOD early in stress tests. The OpenCL portion of the RealBench stress test also kills my GPU driver within minutes since upgrading to Nvidia's 361.75. Are those just signs of general instability manifesting itself, even though the PC passes stress tests just fine, or might there be something deeper there related to the heat issue?

[Guest]

So it's going up 15-20c with that OC?

The 55-60c at stock seems like a really good temperature and some of the best from what I saw when browsing for a the same CPU/cooler. 

[Deli]

6 hours ago, Alexokan said:

So it's going up 15-20c with that OC?

The 55-60c at stock seems like a really good temperature and some of the best from what I saw when browsing for a the same CPU/cooler. 

I have a 4790K no OC with a Noctua NH-D9L, with constant fan speed of 950rpm(no spin up when the temp increases). AIDA 64 4 cores average max at 58C with room temp at 19C.

Don't know how the NH-D9L(92mm fan) compares to Cryorig H7(120mm fan). But I run it at very low speed for quiet reason. Also I don't see the need to increase fan speed as it is cool even under max load.

It used to be in my main rig with Corsair H105 AIO. When OC to 4.6GHz(like OP) at 1.23v. 4 cores average max temp is 62-63C, at room temperature of 22C.

[aisle9]

1 hour ago, Alexokan said:

So it's going up 15-20c with that OC?

The 55-60c at stock seems like a really good temperature and some of the best from what I saw when browsing for a the same CPU/cooler. 

Yeah, that's what I'm thinking too. Could easily be a bad tube of MX-4 that doesn't spread properly. It's almost done with the one-hour OCCT torture test, and temps have not risen above 60C on any core, or on the package itself. Typical loads are around 55-58C. It's been my experience from OC'ing a G3258 and this chip that your temps typically get about as high as they're going to during the first hour. They might go up slightly the longer the test runs, but if you're looking for high temps, you're going to get diminishing returns the longer you go.

I will not be testing with Prime95 at all because I don't enjoy roasting marshmallows on my PC, but I might give FurMark's CPU burner a go again. That said, OCCT was still consistently the hottest stress test I used, getting warmer than the burner by several degrees previously.

1 hour ago, Deli said:

I have a 4790K no OC with a Noctua NH-D9L, with constant fan speed of 950rpm(no spin up when the temp increases). AIDA 64 4 cores average max at 58C with room temp at 19C.

Don't know how the NH-D9L(92mm fan) compares to Cryprig H7(120mm fan). But I run it at very low speed for quiet reason. Also I don't see the need to increase fan speed as it is cool even under max load.

It used to be in my main rig with Corsair H105 AIO. When OC to 4.6GHz(like OP) at 1.23v. 4 cores average max temp is 62-63C, at room temperature of 22C.

Your voltage is a little lower than mine at 4.6GHz, and your ambient is lower on the NH-D9L (mine is 22-25C in that room). I started this processor out with the SilenX Effizio EFZ-92HA3 that I had used to cool my Pentium while waiting for the H7 to get here, and what I quickly learned is that the Effizio sucks for anything higher than a Pentium. Even at stock speeds, it was only 5-7C cooler than the stock heatsink would have been.

AIDA64 will be my next stress test. I'll give it an hour or so to see what it comes up with, but if history is any indicator, it should be lower than OCCT by 4-5C. I do have the CPU fan set to ramp up to 100% at 70C, with my other fans hitting 100% at 75C. At the temps OCCT pinged on this last run, my lower front intake and forward top exhaust wouldn't even have turned on.

If AIDA64 and XTU temps come back lower than 60C, I might try slowly easing up an overclock as the day goes on and seeing what my temps look like. This is one of those days that I love working from home lol

 

*Edit: Just went in to start AIDA64. OCCT had been done for several minutes, and the PC was idle...at 25C. As low as 23C.

 

*Edit 2: AIDA64 hit its peak at 60C within the last couple minutes of the hour-long test, typically hovering around 55C. Ambient in the room was 23C when I went to finish the test, same as at the start. I'm not even going to bother with additional stress testing at this point. When I take a break later on this afternoon, I'll dial the OC back up and give it ten minutes of OCCT. If the temps remain out of line at that speed/voltage, at least I'll know that it's not the cooler or the paste.

[aisle9]

So after 60 minutes of OCCT at 4.6GHz, 1.245v, I peaked at 86C a couple of times, with load temps most often between 70-75C based upon the OCCT graph. Based upon that, I guess I just have to conclude that I got a hot i7.

I'll probably order a second fan for the H7, since it's only $13 from Cryorig for another QF120. I could just go and get a Corsair SP120 performance edition to slap on there, but I'd prefer my PC not sound like a B-52 on its takeoff roll. I'm also expecting, if history is any indicator, another 2-3C improvement over the next couple of weeks as the AS5 burns in, so I'm optimistic that I can get those peaks down to 80-81C over time. It doesn't leave me much headroom to go higher, but realistically, we're only talking about 5% improvement or so from 4.6 to 4.8, which seems to be about the cap for air-cooled 4790K's, and even less up to 4.7, which seems to require about as much voltage as I'd be willing to pump through in the first place.

Unless you guys are thinking any differently, I think I've just got a chip that runs hot when OC'ed, although using AS5 instead of MX-4 definitely helped, shaving about 5C off of everything.

 

Edit: Against my better judgment, I'm going to swing by Fry's and get a two-pack of Corsair SP120 performance edition fans, then push-pull them on the radiator. Airflow is 14 CFM more than the Cryorig fan, which will be moved to the front of the case to replace the stock case fan. I'm hoping that the PWM on those Corsair fans lets me set them to work at speeds that are quiet enough aside from when they really need to go full speed--70C+ CPU temp. I've also dropped the voltage a hair to 1.242 (reading 1.241) to see if I can buy a degree or two there. Given how hot my CPU runs at overclock voltage, every degree counts at this point, and if I can shave a couple off, I will.

[stconquest]

1 hour ago, aisle9 said:

So after 60 minutes of OCCT at 4.6GHz, 1.245v, I peaked at 86C a couple of times, with load temps most often between 70-75C based upon the OCCT graph. Based upon that, I guess I just have to conclude that I got a hot i7.

I'll probably order a second fan for the H7, since it's only $13 from Cryorig for another QF120. I could just go and get a Corsair SP120 performance edition to slap on there, but I'd prefer my PC not sound like a B-52 on its takeoff roll. I'm also expecting, if history is any indicator, another 2-3C improvement over the next couple of weeks as the AS5 burns in, so I'm optimistic that I can get those peaks down to 80-81C over time. It doesn't leave me much headroom to go higher, but realistically, we're only talking about 5% improvement or so from 4.6 to 4.8, which seems to be about the cap for air-cooled 4790K's, and even less up to 4.7, which seems to require about as much voltage as I'd be willing to pump through in the first place.

Unless you guys are thinking any differently, I think I've just got a chip that runs hot when OC'ed, although using AS5 instead of MX-4 definitely helped, shaving about 5C off of everything.

 

Edit: Against my better judgment, I'm going to swing by Fry's and get a two-pack of Corsair SP120 performance edition fans, then push-pull them on the radiator. Airflow is 14 CFM more than the Cryorig fan, which will be moved to the front of the case to replace the stock case fan. I'm hoping that the PWM on those Corsair fans lets me set them to work at speeds that are quiet enough aside from when they really need to go full speed--70C+ CPU temp. I've also dropped the voltage a hair to 1.242 (reading 1.241) to see if I can buy a degree or two there. Given how hot my CPU runs at overclock voltage, every degree counts at this point, and if I can shave a couple off, I will.

Don't buy anything and drop to 4.5GHz.  If you can get the voltage down at 4.5 GHz, you will effectively do the same thing but it won't cost you a dime. 

[aisle9]

Just now, stconquest said:

Don't buy anything and drop to 4.5GHz.  If you can get the voltage down at 4.5 GHz, you will effectively do the same thing but it won't cost you a dime. 

But...but...MOAR POWER!

I'll give it a shot. I want to say I was stable at 4.5 at 1.21v, which is essentially stock voltage.

[stconquest]

1 minute ago, aisle9 said:

But...but...MOAR POWER!

I'll give it a shot. I want to say I was stable at 4.5 at 1.21v, which is essentially stock voltage.

That won't work with me.  =D

I still run my i5 3570K at stock because I have yet to find a need for speed.  I am ready to OC, just don't need to.  3.3GHz + boost baby!