Sign in with
Sign up | Sign in
Your question

4770k temps bad after wipe?

Tags:
  • CPUs
  • Processors
Last response: in CPUs
Share
a b à CPUs
July 10, 2014 7:17:11 AM

Alright, to start, I am not home, so I cannot troubleshoot myself. Just want some of what you guys think. This is my first Intel desktop chip (besides some processor like 8 years ago in an HP Pavilion), so I am new to these temps. I do know a good bit about processors, but I am stumped by my situation.

I first had an Athlon X4 760K, then an FX-8320. I finally got the money to get a 4770k. I installed all drivers and uninstalled the old ones (Should have wiped then and there) and then started getting errors on booting (black screen with mouse after logging in) and tried to fix it. I got a good overclock going (4.2 GHz @ 1.25 V) and my temps were fine, only touching 70 C after a few hours of Small FFTs in Prime95. I eventually had to wipe my drive and reinstalled Windows. Got it all setup again, but only during Prime95 with the same OC, immediately went to 90 C+ after a second. At 4 GHz @ 1.2 V, I still hit 80-90 C after a few seconds. I dropped to stock, went past 80 C and almost touched 90 C. That is of course unacceptable for me because I have an H105. I idle at 30-35 C, so the pump and rad are working fine. At gaming at 4 GHz @ 1.2 V, I still only hit 50 C. I just figure I am probably stable because I play long gaming sessions at a time on Battlefield 3/4 and some other games and never freeze or crash. My RAM is at 2133 MHz now, but was at stock 1600 MHz when I had tested it with the wipe before, so the frequency is not the issue. The waterblock is on there perfectly fine and tightly last I checked, and the pump is fine. It happens also when I go to Floating Point Math benchmarks in PassMark and in AIDA64's FPU stability test. Any ideas? The batch is L402B412 (looked everywhere online, cannot find somebody else except for some other language forum) if anybody has information about that or has a processor from that batch.

More about : 4770k temps bad wipe

a b à CPUs
July 10, 2014 8:26:30 AM

You will see a difference if using a newer version of Prime95 with Haswell CPUs. I can't remember which version it happens, but my best guess is you had an older version of Prime95 when you were getting 70°C, then after your wipe and reinstall you went with the latest version of Prime95 and now see the higher temps. If you google a bit you can find more information. Again, I'm just guessing here.
m
0
l
a b à CPUs
July 10, 2014 8:34:50 AM

rwpritchett said:
You will see a difference if using a newer version of Prime95 with Haswell CPUs. I can't remember which version it happens, but my best guess is you had an older version of Prime95 when you were getting 70°C, then after your wipe and reinstall you went with the latest version of Prime95 and now see the higher temps. If you google a bit you can find more information. Again, I'm just guessing here.


In AIDA64, though, when I stressed the FPU before, the temps were much lower. I think I did try IBT and OCCT, not sure, but I will try on the 15th when I get home. I don't remember OCCT getting that hot, and IBT froze even at my seemingly stable 4 GHz the first time, but not the second.
m
0
l
Related resources
a b à CPUs
July 10, 2014 9:08:14 AM

Live check your vcore when you run prime95. For me, it would suddenly draw 1.56v vcore instead of the static 1.13 I set. Only in prime 28, though. In version 26.6 or lower temps are just same as with occt or aida64. If you run over 70 degrees on those, something might be wrong.

Edit: Just read that you got same thing with aidas fpu test. However, monitor your voltage when your temeratures spike. Make sure it doesn't go too much over the 1.2(5)v you set.
m
0
l
a b à CPUs
July 10, 2014 9:09:48 AM

DubbleClick said:
Live check your vcore when you run prime95. For me, it would suddenly draw 1.56v vcore instead of the static 1.13 I set. Only in prime 28, though. In version 26.6 or lower temps are just same as with occt or aida64. If you run over 70 degrees on those, something might be wrong.


My VCore stays at 1.2 V while stress testing, verified by using a voltage meter as well as CPU-Z.
m
0
l
a b à CPUs
July 10, 2014 9:17:53 AM

CPU-Z won't change it's displayed voltage, so that isn't a good indicator. Voltage meter should be pretty accurate, though. ;) 
But as your cooler seems to be properly working (30 in idle and 50 while playing games is what I get too), I can't really imagine what the issue is, justifiying a 30-40C temp rise in stress tests over games. Maybe check voltages again with like HWiNFO64?
m
0
l
a b à CPUs
July 10, 2014 9:20:42 AM

DubbleClick said:
CPU-Z won't change it's displayed voltage, so that isn't a good indicator. Voltage meter should be pretty accurate, though. ;) 
But as your cooler seems to be properly working (30 in idle and 50 while playing games is what I get too), I can't really imagine what the issue is, justifiying a 30-40C temp rise in stress tests over games. Maybe check voltages again with like HWiNFO64?


That is why I am stumped. I am thinking about returning it to MicroCenter, but it has been 31 days now and I will not be home until the 15th, so I do not think they will accept a return. I don't understand the issue. When I get home, I will check HWiNFO64. I use SpeedFan, the actual temp reader in AIDA, HWMonitor, RealTemp, and they all show the same.
m
0
l
a b à CPUs
July 10, 2014 9:22:34 AM

If you're brave, maybe consider a delid? It's very easy. I lowered my temps by about 15°C.
m
0
l
a b à CPUs
July 10, 2014 9:33:46 AM

Oh, when even aida and others show same voltage and temps, that should be alright. I personally wouldn't consider delidding the cpu and removing warranty when you still have a chance of a rma. I'd try to call your microcenter right now and tell them you'd like to RMA. Even if they first say no, stay at it. Might feel a bit uncomfortable but afterall it increases chances of getting help.

The only strange thing is that temps were fine before you changed OS. Your bios doesn't feature a stress test, huh?
m
0
l
a b à CPUs
July 10, 2014 9:34:01 AM

rwpritchett said:
If you're brave, maybe consider a delid? It's very easy. I lowered my temps by about 15°C.


I have thought about that for a long while, but I REALLY cannot afford to loose a $340 processor if I inflict damage or it dies within warranty, but voided.
m
0
l
a b à CPUs
July 10, 2014 9:36:30 AM

I wouldn't delid until I decide to heavily OC it. Before that, temps _should_ definitely be fine so delidding wouldn't help.
m
0
l
a b à CPUs
July 10, 2014 9:39:13 AM

DubbleClick said:
I wouldn't delid until I decide to heavily OC it. Before that, temps _should_ definitely be fine so delidding wouldn't help.


The thing is, I can't OC far. I got a chip that can't go past 4.3 GHz without getting WHEA_UNCORRECTABLE_ERROR. I fixed that on 4.5 GHz by going to 1.4 V and 4.6 at 1.46 V. That was just for benchmarking and I VERY closely monitored temps and only got to 70 C in 3DMark during Physics tests.
m
0
l
a b à CPUs
July 10, 2014 9:40:30 AM

Andrew Buck said:
DubbleClick said:
I wouldn't delid until I decide to heavily OC it. Before that, temps _should_ definitely be fine so delidding wouldn't help.


The thing is, I can't OC far. I got a chip that can't go past 4.3 GHz without getting WHEA_UNCORRECTABLE_ERROR. I fixed that on 4.5 GHz by going to 1.4 V and 4.6 at 1.46 V. That was just for benchmarking and I VERY closely monitored temps and only got to 70 C in 3DMark during Physics tests.


Then that's definitely a reason to RMA. The cpu MUST reach 4.4ghz stable.

Ugh. Just hope that wasn't the reason for my 0x124 (WHEA_UNCORRECTABLE_ERROR) 's. RMA'ed motherboard and ram by now and hope the issue was caused by either of those. Well, worst case scenario I'd have to RMA the cpu too. :<
What's your VID, if I may ask? (4x4.0ghz, voltage setting auto, ram on base clock, EIST off, doing a test with prime95 26.6 and then measuring voltage with coretemp)
m
0
l
a b à CPUs
July 10, 2014 9:42:50 AM

DubbleClick said:
Andrew Buck said:
DubbleClick said:
I wouldn't delid until I decide to heavily OC it. Before that, temps _should_ definitely be fine so delidding wouldn't help.


The thing is, I can't OC far. I got a chip that can't go past 4.3 GHz without getting WHEA_UNCORRECTABLE_ERROR. I fixed that on 4.5 GHz by going to 1.4 V and 4.6 at 1.46 V. That was just for benchmarking and I VERY closely monitored temps and only got to 70 C in 3DMark during Physics tests.


Then that's definitely a reason to RMA. The cpu MUST reach 4.4ghz stable.


The warranty doesn't cover overclocking (I am pretty sure). I would only be able to for that on stock. The MicroCenter return policy only allows 15-day return coverage. It has been 31 and will be 36 by the time I could return it.
m
0
l
a b à CPUs
July 10, 2014 9:45:31 AM

Oh, nevermind. Thought we were speaking of an i7 4790k. Pretty much every i7 thread right now is about that one haha. Humm, running out of ideas.
m
0
l
a b à CPUs
July 10, 2014 9:47:35 AM

DubbleClick said:
Oh, nevermind. Thought we were speaking of an i7 4790k. Pretty much every i7 thread right now is about that one haha. Humm, running out of ideas.


I know, seriously. I answered 5 4790K threads in one day.
m
0
l
a b à CPUs
July 10, 2014 9:52:01 AM

Even one of mine, iirc, haha. Try to reinstall windows again then? If temps were fine the other time, maybe that fixes it? Otherwise, in worst case scenario you just have to live with that, I guess. When temps stay at under 60 degrees in usual working/gaming circumstances that shouldn't be the issue. You can always run a program that notifies you when temps are going over a certain value in background.
m
0
l
July 10, 2014 9:53:19 AM

why don't you set your memory to 1333mhz and 1.5v in the bios to see the difference.

i know for a fact that changing mhz on your memory will increase cpu temperature when stress testing.

make sure your setting vcore and not vvrin or any other cpu voltage setting.

ive seen using a particular version of p95 will automatically increase your vcore voltages even if its fixed in the bios.

m
0
l
a b à CPUs
July 10, 2014 9:54:41 AM

DubbleClick said:
Even one of mine, iirc, haha. Try to reinstall windows again then? If temps were fine the other time, maybe that fixes it? Otherwise, in worst case scenario you just have to live with that, I guess. When temps stay at under 60 degrees in usual working/gaming circumstances that shouldn't be the issue. You can always run a program that notifies you when temps are going over a certain value in background.


Wiping again wouldn't fix it, because all I have now is the drivers and some games, as well as benchmarks and temperature monitors. I always have MSI Afterburner open for my GPUs and all CPU cores. Their usages and temps are all there.
m
0
l
a b à CPUs
July 10, 2014 9:58:44 AM

Andrew Buck said:
DubbleClick said:
Even one of mine, iirc, haha. Try to reinstall windows again then? If temps were fine the other time, maybe that fixes it? Otherwise, in worst case scenario you just have to live with that, I guess. When temps stay at under 60 degrees in usual working/gaming circumstances that shouldn't be the issue. You can always run a program that notifies you when temps are going over a certain value in background.


Wiping again wouldn't fix it, because all I have now is the drivers and some games, as well as benchmarks and temperature monitors. I always have MSI Afterburner open for my GPUs and all CPU cores. Their usages and temps are all there.


I've had pretty amazing experiences with windows installations. Once my pc would only register an action each 10 seconds. Fresh install, no problem. When setting your RAM to 2133mhz, did you increase any voltages? Usually system agent and/or VTT (CPU I/O) need to be adjusticed on oc'ing ram.

Edit: And yeah, like pluke suggested try running it at 1600mhz again? That shouldn't really have impact on the cpu temps, though.
m
0
l
a b à CPUs
July 10, 2014 10:05:09 AM

DubbleClick said:
Andrew Buck said:
DubbleClick said:
Even one of mine, iirc, haha. Try to reinstall windows again then? If temps were fine the other time, maybe that fixes it? Otherwise, in worst case scenario you just have to live with that, I guess. When temps stay at under 60 degrees in usual working/gaming circumstances that shouldn't be the issue. You can always run a program that notifies you when temps are going over a certain value in background.


Wiping again wouldn't fix it, because all I have now is the drivers and some games, as well as benchmarks and temperature monitors. I always have MSI Afterburner open for my GPUs and all CPU cores. Their usages and temps are all there.


I've had pretty amazing experiences with windows installations. Once my pc would only register an action each 10 seconds. Fresh install, no problem. When setting your RAM to 2133mhz, did you increase any voltages? Usually system agent and/or VTT (CPU I/O) need to be adjusticed on oc'ing ram.

Edit: And yeah, like pluke suggested try running it at 1600mhz again? That shouldn't really have impact on the cpu temps, though.


I will see when I get home. I stated that I also tried running at 1600 MHz and still no luck. Voltage was OCed to 1.65 V and the other voltages like system agent were bumped up a little bit, too.
m
0
l
July 10, 2014 10:05:16 AM

running memory at below overclocked settings will reduce cpu temperature at load. mine has a difference of of 5c on all cores. and to find a stable overclock with good temperatures you need to isolate ram first and not overclock it.

i'd just start from scratch.
easier said than done but if you want the overclock stable and relatively low in temperature you need to cover your bases.



m
0
l
a b à CPUs
July 10, 2014 10:10:10 AM

Hmm, bringing my ram from 1333 to 2400 mhz actually had no impact on my cpu temps. O.o
m
0
l
a b à CPUs
July 10, 2014 10:12:43 AM

DubbleClick said:
Hmm, bringing my ram from 1333 to 2400 mhz actually had no impact on my cpu temps. O.o


Exactly. Mine is the same, whether it is at 1600 MHz or 2133.
m
0
l
July 10, 2014 10:13:22 AM

you'll need to disable the xmp profile for your ram in bios to see that it does reduce temperatures.
m
0
l
a b à CPUs
July 10, 2014 10:15:57 AM

pluke the 2 said:
you'll need to disable the xmp profile for your ram in bios to see that it does reduce temperatures.


I did that. In the beginning, XMP was off anyways.
m
0
l
July 10, 2014 10:26:42 AM

increasing the memory speed puts direct heat load on the cpu core.
when you use higher memory speed, your overclocking the memory controller to do it, therefor, youre going to generate more heat. not sure why you guys are not seeing this but thats just the way it is.


i'd do this...

get

http://www.intel.com/content/www/us/en/motherboards/des...

get

http://www.alcpu.com/CoreTemp/

get

http://www.cpuid.com/softwares/cpu-z.html

put your bios to defaults.

look at your bios version and go online to verify you got the latest.

then read this

http://www.overclockersclub.com/guides/overclock_intel_...
m
0
l
July 10, 2014 10:30:26 AM

i know what you're thinking.. been there done that.

i'd set bios to defaults and fire up the machine and do Pre-Overclock Stability testing and see where your at with your core temperatures. monitor what the vcore is at and dont exceed 90c
m
0
l
a b à CPUs
July 10, 2014 10:31:29 AM

pluke the 2 said:
increasing the memory speed puts direct heat load on the cpu core.
when you use higher memory speed, your overclocking the memory controller to do it, therefor, youre going to generate more heat. not sure why you guys are not seeing this but thats just the way it is.


i'd do this...

get

http://www.intel.com/content/www/us/en/motherboards/des...

get

http://www.alcpu.com/CoreTemp/

get

http://www.cpuid.com/softwares/cpu-z.html

put your bios to defaults.

look at your bios version and go online to verify you got the latest.

then read this

http://www.overclockersclub.com/guides/overclock_intel_...


I am experienced with overclocking Intel now that I have read many guides and worked very long on it, and I have CoreTemp and CPU-Z. Again, not at home, but when I am, I will try the Intel utility. As for the memory, it doesn't make CPU temps any higher as I recall. It is the same issue at 1600 MHz with XMP off and at stock voltages. I have temp issues when resetting to stock. I have tried that many times.
m
0
l
July 10, 2014 10:42:41 AM

reset to stock and fix a manual vcore setting and see how low you can get your vcore voltage while being stable and record your temperatures at load. you may have other issues. maybe going to fast.
m
0
l
July 10, 2014 10:44:48 AM

i got the same coooler and processor as you and i am at 4.6ghz w/ 1.38vcore and 72c at full load for hours. so either youre doing something wrong or you got a bad chip
m
0
l
a b à CPUs
July 10, 2014 10:44:55 AM

pluke the 2 said:
reset to stock and fix a manual vcore setting and see how low you can get your vcore voltage while being stable and record your temperatures at load. you may have other issues. maybe going to fast.


I have dedicated about 2 weeks to overclocking. I have tried at 3.5 GHz (no turbo) and 1.05 V and got to around 75 C, but that is what I got at 4.2 @ 1.275 before.
m
0
l
July 10, 2014 10:58:10 AM

Andrew Buck said:
pluke the 2 said:
reset to stock and fix a manual vcore setting and see how low you can get your vcore voltage while being stable and record your temperatures at load. you may have other issues. maybe going to fast.


I have dedicated about 2 weeks to overclocking. I have tried at 3.5 GHz (no turbo) and 1.05 V and got to around 75 C, but that is what I got at 4.2 @ 1.275 before.


with this said, i'd remount the heatsink on the cpu. if you do this, take note of thermal paste contact on the cpu.

if you do this, you'll need to remove the old paste from both the cpu and the heatsink. i'd use arctic silver 5 for the paste.

m
0
l
July 10, 2014 10:59:30 AM

plain and simple. 75c is very hot for 1.05 volts. but 4.2 at 1.275 and 75c sounds a bit more normal. do you recall room temp?
m
0
l
a b à CPUs
July 10, 2014 11:04:45 AM

Yeah 75C seems reasonable for 1.27v, not for 1.05 though. As for thermal paste, don't go "cheap" on that. With the same cooler going from some cheap one to mx4 (and applying a bit less) I dropped from 65 degrees max to 47 max. under prime load on my q6600. Which is pretty amazing, considering I got it at 1.37v 3.0ghz (worst chip i could i have got, I guess, even though C0 stepping), lol.
m
0
l
July 10, 2014 11:18:01 AM

these haswell cpus are just teh suck comparing to the ivys and sandys. they run really hot and need a lot of volts. delidding is very very common and popular.
m
0
l
a b à CPUs
July 10, 2014 11:23:06 AM

I have Antec Formula 7 NanoDiamond Thermal Compound, so that is almost the best there. The room temp is around the same at about 23 C.
m
0
l
July 10, 2014 11:40:05 AM

i dont have an answer for you as to why no overclock nets you the same temperature as 1.27volts w/ the overclock. something just sounds like it came loose coincidentally as you formatted/used different versions of prime 95.

like i said before. i've heard not to use p95 for the haswell cpus.

i'd be okay with: 4.2 @ 1.275 at 75c if its stable.

are you not okay with this?
m
0
l
July 10, 2014 11:42:16 AM

can you get back to this over clock/temperatures? 4.2 @ 1.275 at 75c

m
0
l
a b à CPUs
July 10, 2014 11:48:50 AM

pluke the 2 said:
i dont have an answer for you as to why no overclock nets you the same temperature as 1.27volts w/ the overclock. something just sounds like it came loose coincidentally as you formatted/used different versions of prime 95.

like i said before. i've heard not to use p95 for the haswell cpus.

i'd be okay with: 4.2 @ 1.275 at 75c if its stable.

are you not okay with this?


That was my temp before the wipe. Now it goes over 90 C. I will get back to you on an update when I get home.
m
0
l
July 10, 2014 11:57:19 AM

i think something came loose. there is no reason for a format to cause this. but the prime 95 program could alter your voltage and cause your temperature to skyrocket. let usknow
m
0
l
a b à CPUs
August 7, 2014 10:38:14 AM

Well I am really mad right now. Sorry I forgot about this, but I did try tightening the cooler. No improvement. I also got my 850 Pro over the weekend and installed the OS on it again. Ran some tests, not stable at 4.2 no matter if I am at 1.2 Volts or 1.3, get the same damn bluescreen, WHEA_UNCORRECTABLE_ERROR and some CLOCK_WATCHDOG error. At 1.3, though, I touch 85 C, when on some H105 tests like this:



Even at 4.0 @ 1.2 V I hit 75 C (after about 30 seconds) on the hottest cores (1 and 2, 74 on the 3rd) and 68 C on the coolest core. It just doesn't make sense to me. I am completely stable at this clockspeed and voltage, but past there, bluescreens plague me. It has been 2 months since I purchased the CPU, so no returning it now. I also froze yesterday at this clock and voltage, which really makes me mad. Not looking to delid in case something happens and there is no point provided I can't go past 4.0 GHz and have to probably give 1.325 Volts or more to be stable at 4.2.
m
0
l
!