Question about Haswell 5930k Temps

[Proxy]

I have a Corsair H110i GT AIO cooler, and replaced the stock fans with Noctua NF-P14s redux 1500 PWMs.

Right now the temps are only slightly better than with the stock Corsair fans.

My idle temps are:

Hottest core: 37c
Coolest core: 33c

Under full load:

Hottest core: 83c
Coolest core: 80c

Average (day-to-day use, VMWare running, Chrome, video's, gaming, sometimes all at the same time):

Hottest core: 41c
Coolest core: 43c

So my question is, given the setup, do those temps seems reasonable/normal? What I was most concerned about was the load temps inparticular. I had tested with Prime 95 (yes I am well aware of the Prime 95 with Haswell/Haswell-E dangers). This is all on an i7 5930k with stock clocks.

Thanks guys!

 

[Xinmosni]

I don't see any reason why these would be out of the ordinary.

AIO should be considered equivalent to really good air cooling, nothing more. If you're going to replace fans on an AIO setup, you should do so for acoustics not temps IMO.

For noticeably better temps you'll have transition to a true WC loop.

 

[Proxy]

Thanks for the reply. The system is definately a lot more quiet with the Noctua fans.

 

[ssnyder28]

To me it sounds like your load temps are quite high for stock clocks. The H110i is a top tier cpu cooler and would expect lower temps than that even with a decent overclock.

Did you try re-seating the block?

 

[GotNoRice]

Those temps are worse than what I get with my 5820k using an H100i GTX with 2 stock SP120L fans in push and two Scythe Slipstream fans in pull. I'm usually <30C at idle, and even after cooking with with Prime95 for a while it's rare for any core to break 75C. That is at 4.5Ghz, using ~1.35V.

 

[Proxy]

To me it sounds like your load temps are quite high for stock clocks. The H110i is a top tier cpu cooler and would expect lower temps than that even with a decent overclock.

Did you try re-seating the block?

Yes, I did that when I moved everything to a new case last night, reapplied the Noctua TIM, AND replaced the stock fans with Noctua fans.

Those temps are worse than what I get with my 5820k using an H100i GTX with 2 stock SP120L fans in push and two Scythe Slipstream fans in pull. I'm usually <30C at idle, and even after cooking with with Prime95 for a while it's rare for any core to break 75C. That is at 4.5Ghz, using ~1.35V.

That's what I thought. With this cooler I figured my temps should be better. Maybe the cooler is defective? I have it all hooked up correctly with Noctua fans, and awesome case air flow. I'll get some pics of my rig up later on, but I don't see what the issue is other than something is wrong with the cooling unit.

 

[GotNoRice]

For sure those temps are high, but i'm still curious, have you tried using CorsairLink and seeing if it makes a difference at all shifting the pump, and potentially the fans also, into performance mode? I should have specified above, but those temps that I quoted were with both the push fans and the pump set to performance mode and pull fans connected to mobo with an aggressive smart curve. It still shouldn't explain your >80C at stock either way but i'm curious if it makes a difference at all.

 

[Proxy]

For sure those temps are high, but i'm still curious, have you tried using CorsairLink and seeing if it makes a difference at all shifting the pump, and potentially the fans also, into performance mode? I should have specified above, but those temps that I quoted were with both the push fans and the pump set to performance mode and pull fans connected to mobo with an aggressive smart curve. It still shouldn't explain your >80C at stock either way but i'm curious if it makes a difference at all.

So after looking up the documentation I don't have the cooler hooked up right. I've been building PCs for 10 years and made a rookie mistake.

I have the fans connected directly to the MB CPU fan headers (mine has two, for coolers with two fans). The pump is plugged into the SATA power (that part is correct), but I never put the corsair link cable in. I also need to hook the fans into the pump fan connectors and use the little bypass that plugs into the MB's main CPU fan header. I'm going to do all of that and see if that makes any difference at all. I hope it does.


Edit: I got the Corsair link cable connected, and using the Default modes it idles in the low 30's and loads in the low 70's, with only one of my cores hitting 81c. Seems to be better. I like that it's detecting the correct profiles now, and might keep it on the Default profile because I like how quite it is, and if the CPU does come under a load it can keep up a bit better now.

 

[ssnyder28]

So after looking up the documentation I don't have the cooler hooked up right. I've been building PCs for 10 years and made a rookie mistake.

I have the fans connected directly to the MB CPU fan headers (mine has two, for coolers with two fans). The pump is plugged into the SATA power (that part is correct), but I never put the corsair link cable in. I also need to hook the fans into the pump fan connectors and use the little bypass that plugs into the MB's main CPU fan header. I'm going to do all of that and see if that makes any difference at all. I hope it does.


Edit: I got the Corsair link cable connected, and using the Default modes it idles in the low 30's and loads in the low 70's, with only one of my cores hitting 81c. Seems to be better. I like that it's detecting the correct profiles now, and might keep it on the Default profile because I like how quite it is, and if the CPU does come under a load it can keep up a bit better now.

I'd still be concerned if I were you, 81C even on just one core isn't good at stock clocks.

 

[Proxy]

I'd still be concerned if I were you, 81C even on just one core isn't good at stock clocks.

Well maybe the unit is defective then. I really am out of options now. I'm going to reapply the TIM one more time just to be sure. If nothing happens then I am going to have to get a new cooler.

Edit: I reapplied the Noctua TIM, and I did another idle test and stress test. Whether or not the pump is in Quite Mode or Performance Mode my highest core is ~33c with the lowest sitting around 29c. Under a Prime 95 load (Small FFT for max power/heat) my highest core was 81c with the lowest around 74c - this was with the pump in performance mode, with it in quite mode the lowest temps didn't really change but the hottest core jumped from 81c to 84c. I just don't know maybe it's the cooling unit, maybe it's the CPU? I have no idea at this point.

 

[Agenesis]

Try the 27.9 build instead. It doesn't use AVX2 and runs a lot cooler.

But your results looks about right if you're testing with the latest prime builds. Not even custom loops can handle prime95/aida64 fpu tests once overclocked.

 

[Proxy]

Try the 27.9 build instead. It doesn't use AVX2 and runs a lot cooler.

But your results looks about right if you're testing with the latest prime builds. Not even custom loops can handle prime95/aida64 fpu tests once overclocked.

Yeah I was using the latest build I will try that one and see what I get. Thank you!

 

[addictionsoLA]

isn't it advised to NOT use prime95 with haswell-e? I swear when I was looking up overclocking my 5930k I saw warning after warning about it. That being said I have a 5930k with h110, and hit a max of 73 at 4500mhz at 1.3 vcore. (this is using RoG real bench)

 

[cageymaru]

You don't have the Corsair pump running on a fan header that is variably regulated according to the current temperature by the motherboard do you? I remember botching that with the H50 some years ago. What I remember having to do was to go into the bios and set that fan header to run at full speed 100% all of the time.

I don't own one of the newer pumps so they may have changed it.

 

[Proxy]

Try the 27.9 build instead. It doesn't use AVX2 and runs a lot cooler.

But your results looks about right if you're testing with the latest prime builds. Not even custom loops can handle prime95/aida64 fpu tests once overclocked.

I couldn't find the 27.9 build for some reason, but I did fine the 27.7. I ran that and at fullk load my CPU was not sitting on the low 70's across the board, which is a LOT better that what I was seeing before. The newest build of p95 really kicks the crap out of these things. I'm assuming low 70's is good. Idle temps haven't changed mainly because at idle the fans kick in to silent mode. Which is fine, that's what I wanted, silent desktop operation, and low noise on gaming. During 'normal' day-to-day use the system still sits around the mid to upper 30's with the fans adjusting themselves.

isn't it advised to NOT use prime95 with haswell-e? I swear when I was looking up overclocking my 5930k I saw warning after warning about it. That being said I have a 5930k with h110, and hit a max of 73 at 4500mhz at 1.3 vcore. (this is using RoG real bench)

Yes, I even stated that in my OP, but with the other benching software there was a bunch of extra fluff I didn't want to deal with so I just ran a quick pass with p95. I like to live on the edge.

You don't have the Corsair pump running on a fan header that is variably regulated according to the current temperature by the motherboard do you? I remember botching that with the H50 some years ago. What I remember having to do was to go into the bios and set that fan header to run at full speed 100% all of the time.

I don't own one of the newer pumps so they may have changed it.

The H110i GT is different from the H50 and even the H110i GTX in that the CPU head from the pump only has a single ground cable that connects to the CPU header. According to the guide I found on it, all it does is report fan speed to the BIOS, but does not control pump speed so you don't get the F1 errors.

 

[sblantipodi]

Well maybe the unit is defective then. I really am out of options now. I'm going to reapply the TIM one more time just to be sure. If nothing happens then I am going to have to get a new cooler.

Edit: I reapplied the Noctua TIM, and I did another idle test and stress test. Whether or not the pump is in Quite Mode or Performance Mode my highest core is ~33c with the lowest sitting around 29c. Under a Prime 95 load (Small FFT for max power/heat) my highest core was 81c with the lowest around 74c - this was with the pump in performance mode, with it in quite mode the lowest temps didn't really change but the hottest core jumped from 81c to 84c. I just don't know maybe it's the cooling unit, maybe it's the CPU? I have no idea at this point.

Are we talking about stock frequency and stock voltage?
Use a more realistic benchmark like Asus realbench

 

[Proxy]

Are we talking about stock frequency and stock voltage?
Use a more realistic benchmark like Asus realbench

From my post above. This seems like a more reasonable result, considering that I have the fans in silent mode now.

I couldn't find the 27.9 build for some reason, but I did fine the 27.7. I ran that and at fullk load my CPU was not sitting on the low 70's across the board, which is a LOT better that what I was seeing before. The newest build of p95 really kicks the crap out of these things. I'm assuming low 70's is good. Idle temps haven't changed mainly because at idle the fans kick in to silent mode. Which is fine, that's what I wanted, silent desktop operation, and low noise on gaming. During 'normal' day-to-day use the system still sits around the mid to upper 30's with the fans adjusting themselves.