Jump to content
Corsair Community

H80i, 4790k and Maximus VII Hero temp jumps


Todde

Recommended Posts

Hi all!

 

I'm a bit concerned about the temperatures I get with my new board (Asus Maximus VII Hero), CPU (4790K) and H80i.

 

I use default settings in bios, I have not even checked if there is any OC default but I asume it isn't.

 

I have the stock fans on as pull-push and radiator in the front of the box (after some modification of the box). The fans are connected to the H80i with the y-cable that came with it. The H80i to a USB-port.

 

I fiddled around a bit with Corsair Link software and the latest setting I have is: pump at max ie. rpm division by one so around 4.4k rpm I think. Fans with custom curve controlled by H80i temp. 36 degree celcius will put the fans at max speed. 34 at 1600. 32 at 1200. I don't remember all points in my head right now, but I think it's sufficient for now.

 

First assemble the bracket were a bit loose. I have re-seated the cooler with new paste and extra rubber washes on the bracket on the backside. The bracket seemed very firm and the tower screws were tight to the mainboard. The cooler feels very firm as well. I used artic silver 5 when I re-seated and I didn't use much. Maybe the size of a green pea or a wee bit less.

 

What happens is (all temps (Celcius) etc is taken from Corsair Link software) Motherboard temp seemed to max at 32 degree in my last test:

 

1. Computer and windows started, Cpu idle temp: around 33 degrees. H80i at 28 degrees.

 

2. I open internet explorer and browse to lineage2.com and there is a picture and an animation (right now it is) this Causes CPU temp to go 50-55 degrees and load is around 15-17%. H80i temp goes to 30-31.

 

3. I fire up Prime95 with a torture blend mix test (I think it's called), 100% load. Cpu jumps to 55 and pretty quick up to 60-61 and stays there for maybe 2-3 minutes. In this time H80i temp goes up towards 36 degrees.

 

4. With Prime95 still running and cpu is at around 60-61 the H80i temp passes 36 degrees and when this happens the cpu temp climbs quickly to 79-80 degrees and stays there for perhaps 2 minutes. H80i seems to top at 36.8 degrees and fans are running at max speed under these about 2 minutes. Then suddenly cpu temp starts to go down under about 30 sek from 80 to 60-61 again. The H80is temp goes down also to around 35.x and fans around 1500-1600 hundred.

 

5. I keep prime95 run a bit more and after about 2 min or something the H80i temp passes 36 again, cpu climbs to 80. H80i continues to 36.8 with fans at maximum.

 

The temp jumps repeatedly up and down from 60 to 80 and back to 60 as described above. Maybe I must just have higher fan speed to prevent the H80i to raise above 36 degrees. Already at 1400+ rpm they are making a lot of noise for me.

 

Why does the temp jump up and down like this and is it normal? Can I prevent it somehow?

 

I wonder if the temps above at idle, at such little load as 15-17% and with Prime95 100% load are normal? If not what could be the problem?

 

I also wonder if I should replace the fans on the H80i but if so with what, that gives enough air pressure, moves enough air and is more quiet?

 

I plan to change the case fans (two of them, front and back) also and would appreciate recommendations on quiet stable ones with pwm so the motherboard can adjust the speed.

 

Hope for help and input!

 

/Todde

Link to comment
Share on other sites

Have you manually set your CPU voltage? Prime95 starts using AVX instructions after a few minutes which bump the voltage up 0.1v which send the temperatures up. This caused me to panic (I've got the Hero VII and 4790k too, one of the guys on the forum told me about AVX) - try stressing the CPU with something that doesn't use AVX such as Aida64 with FPU turned off or Cinebench and see if you get more consistent temps during the run.

 

I must admit my CPU temps fluctuate quite a bit as load changes. If you've left any overclocking options alone the CPU voltage will change quite a bit as load goes up and down which no doubt increases the temperatures.

 

Temperatures seem high for Internet explorer use though! - will be interesting to see what anyone else makes of this (you're not running Kaspersky are you?). If I get a chance I'll see what happens on my machine tonight. My work PC (a 4770k) on the lineage2 page doesn't seem to cause too much temperature increase although it is a pretty constant 15% load.

Link to comment
Share on other sites

Hi iflo, thanks for your reply!

 

I had everything at default in bios, but after I read your reply I changed cpu core volt to manual and let it be as it was default. I tried with prime95 again and the same happend, maybe it took a little longer to reach 79 now. The VID due to cpuid hwmonitor or cpu-z shows 1.295V as max, and 0.719V as min.

 

I downloaded a trial of AIDA64 and ran a system stability test without fpu and the temp on cpu were around 60-64 and the H80i temp didn't go over 34.8. I ran it for a bit more then 10 minutes, which is by far much longer then Prime95 needs to take it to 79-80 degrees.

 

I tried the lineage2 page again, and it's around 47-48 degrees at about 15% load. I tweaked the fan speed a little so it's at 800 rpm as the lowest which is a little bit more then I had first (600). Maybe this caused the cpu temp to be a bit lower when browsing to the lineage2 page.

 

The cpu core speed due to cpu-z shows 4397.40 all the time I run Prime95 or Aida.

 

Edit: Installed Diablo 3 and played for about half an hour, switched to desktop once in a while and the cpu temp were around 48-52 degrees.

 

I'm not running Kyspersky or any antivirus right now. Usually I run Avira.

 

Would you recommend that I change anything in the bios?

Link to comment
Share on other sites

1.295v is pretty dam high voltage. You should expect those temps with that voltage on that chip. My 4770k could handle more than 1.25v on my h105 and it hit 90° at that voltage. After delid I'm topping out at 90° at 1.4v. 4790k is only about 5-10° better than 4770k as it doesn't use solder between ihs just a better thermal paste. So those temps are totally in line with that chip and that voltage. But who knows if your chip is good you may be able to get 4.6-4.8ghz with that voltage.
Link to comment
Share on other sites

I have set my adaptive voltage on the motherboard to be -0.05v so my voltage peaks at 1.2v rather than 1.25 under normal (non-AVX) use. That helped things a bit, but I'm running at stock clock speeds. I first of all used a fixed voltage underclock to lower it down and benchmark, then once I was comfortable used the adaptive voltage setting.

 

Been looking at the lineage 2 page for a few minutes now an my temp is sitting around 36-40 degrees (on a h100i)

Link to comment
Share on other sites

I have change one thing in bios now, I disabled turbo boost so the cpu maxes at 4 GHz. I have not fiddled with any voltage settings. Due to HWMonitor the max VID were 1.116. Watching that Lineage2 page gives around 15% load and the temp is around 38-40. Also playing d3 is around 40-42 so around 10 degrees lower.

 

Prime95 takes the cpu temp up to 48 instantly. After 1 min the cpu temp is 50 and h80i temp 33.0. After 3 min: cpu 50 and h80i 33,6. After 6 min cpu 59 h89i 34,4. After 9 min cpu 60 h80i at 34,8. After 12 min cpu 51 h80i 33,9. It goes up again towards 60 degrees on cpu and back to 50, it repeats the same cykel as in my first post (probably because it starts to use those AVX instructions), but now between 50 to 60 instead of 60 to 80.

 

I noticed now with turbo boost off that the VID maxes at 1.116 with Prime95 running.

 

I would like to have it running with turbo boost but not at 1.295v, must figure out if I can limit that and to what voltage, maybe 1.25?

Link to comment
Share on other sites

Like I say I've got mine running adaptive voltage so it peaks at 1.2v under load turboing up to the 4.4Ghz. I gave up worrying about Prime95 temps with AVX as nothing I do will ever stress it this much, and programs that use AVX are few and far between (at the moment) and as far as I understand, the ones that do don't hammer it like Prime95.

 

I don't need 200 or 300 extra Mhz so am happy to down-volt a bit to get a more quiet fan profile at stock clock speeds!

Link to comment
Share on other sites

I'm using adaptive now also and -0.05V. I can see cpuid hwmonitor shows that vid was 1.252V as maximum. The lowest I could set extra voltage for turbo mode was 0.001V. I choose auto instead of Sync all cores on CPU Core ratio , not sure if I loose performance by this but it seems to lower the temp.

 

All this keeps it idle at 28-30 degrees and I seems to max at 40-42 when playing Diablo 3.

Link to comment
Share on other sites

I found a strange problem when I use auto instead of sync all core on "cpu core ratio". Sometimes when I fire up internet Explorer and start to type an url the keyboard input freezes for a short while. Then the text I enter pops up. This continues and gets worse and worse and I must restart windows. The mouse Point once totally froze while the keyboard still could handle input. All this problem seems to disappear if I use Sync all core instead of auto on cpu core ratio.
Link to comment
Share on other sites

  • 4 weeks later...

I would be quite happy with the results you are seeing. I'd expect the die temperature to change rapidly with CPU load since the temperature sensors are on the die and so see changes immediately. It will take a little time for the heat to conduct out to the cooler. It is the maximum die temperature I'd watch. Under Prime 95 v285 Small FFT Torture Test my maximum die temperature rises close to 100c at the moment! The cooler temperature remains in the low 30s. I'm going to try to improve the seating of the waterblock now to see if I can get your results!

 

I wonder what maximum temperature should be regarded as acceptable - since Intel are measuring this on the die, surely they will not allow it to rise to dangerous levels?

 

Incidentally, try getting a (free) copy of Piriform Speccy - that shows the temperatures and multipliers for each core separately. This is illuminating and the large multiplier changes with load help understand why the temperature fluctuates so much.

Link to comment
Share on other sites

I found that the waterblock support was quite loose and the pre-applied thermal compound was only wetting about 2/3rds of the processor cap. The support structure was fitted exactly as instructed - but the standoffs do not tighten completely (seems to have been designed for a thicker PCB). I fitted four hard rubber washers between the rear frame and the board so the standoffs tighten against the front of the board. I also fitted small packers about 1mm thick between the water block and the support frame in each corner and used the four washers supplied (but not mentioned in the fitting instructions) between the support frame and the thumbscrews. Now the thumbscrews can be used to tension the waterblock against the chip, but you have to be careful not to over-tighten.

 

This has made the necessary difference, enabling the multipliers to be stable at x44 even during the Prime95 extreme torture test (before it was limiting to x40). But the peak temperature is still alarmingly high in the high 80s occasionally edging over 90c. If this is stable on a soak test, I will be satisfied, but disappointed not to have more headroom for overclocking.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...