4770k high temps with H100i

tektechnician · Sep 24, 2014

I have various temp readings within my computer about my 4770k at 4.5Ghz at 1.275v and I do not know which temp I should be trusting. I have a H100i cooling my 4770k in a push pull config, I have a ASUS Z97-C motherboard running my system. So, Ai Suite 3 reports that my cpu when loaded maxes out at 50c and runs around 30-40c when idling. Corsair link as well as Realtemp and every other software reports that my water temp is at 35-40c idle for 2 hours powered on, and when maxed out with Prime95, (my cpu, not the water temps),would hit ~83c with a fresh application of NT-H1 and ~93c on a 4-6 week application, which one should i trust?

JackNaylorPE · Sep 24, 2014

AiSuite does not report CPU Core temps...that is the package temp. Also 1.275 is at the very upper limit for even a 2700 rpm AIO .....

I'd recommend that you stop using AIDA, Prime 95 or anything else like that..... not really recommended for Haswell CPUs if you looking for Real World results.

Download RoG Real Bench, HWiNFO64

http://rog.asus.com/forum/showthread.php?43233-Realbench-v2-Discussion-Thread-Download-Links
http://www.hwinfo.com/download.php

Usage of testing and monitoring programs:

When you open Real Bench, move both windows to left side of screen. Open HWiNFO64, run "sensors only", you will get a pop up asking whether to disable reading the Asus EC chip, click "Disable this sensor". Move the HWiNFO64 window to upper right hand corner of screen. Stretch bottom of window to full screen height. Make the following changes (optional but oit makes life easier):

-Right Click on "System" right at the top, select hide.
-In the next section, hide the last 4 lines starting "Core CPU Thermal Throttling" (if you watch temps, this is useless)
-Skip over the next section and Hide the section after that (section includes CPU Package thru DRAM Power)
-Now the whole reasons we did that was so you could see everything you wanna see at same time. You should be able to see Vcore 0, 1 and 2 at -the bottom of the window. If not hide a few more lines. Save and Quit will save your edits.

Open Real Bench, select Benchmark Tab Check only the last box. Open HWiNFO64, run "sensors only" as described above. Start Real Bench and don't touch mouse till finished. Then try checking all 4 boxes and run again NOTE: During the 3rd test Open CL will send AVX instructions to CPU; pay close attention to Vcores as they will spike as described above. If passes and you want to go further ,,,,

Switch to the Stress Test Tab and select the amount of RAM you have in your system and 2 hours..... (Note: If you plan on raising cache and / or RAM after a run, I will usually save the two hours and skip this step until I have Multiplier / Cache and RAM speed at my targets.

Record the Maximum core temps and voltages.

Davil · Sep 25, 2014

I use prime95 to test clocks all the time, it stresses to 100% and that's all you really are looking for it to do, if you can run that way stable for an hour or so you shouldn't have problems. The 83c reading is almost definitely the correct one. The reason you're getting really high temps is due to the cheap thermal paste under the IHS. The only way to fix it is to de-lid the chip and replace it with better stuff. I have included a video below of how to do it, but it's not for the faint of heart.

https://www.youtube.com/watch?v=sqfpgUK9Ldo&list=UU0_swmlWWBAJofDN1nOUcjA

Sadly even on a custom water cooling loop temps on 3rd and 4th generation i5's and i7's produce a lot of heat that is hard to get rid of unless you de-lid.

tektechnician · Sep 25, 2014

JackNaylorPE :

AiSuite does not report CPU Core temps...that is the package temp. Also 1.275 is at the very upper limit for even a 2700 rpm AIO .....

I'd recommend that you stop using AIDA, Prime 95 or anything else like that..... not really recommended for Haswell CPUs if you looking for Real World results.

Download RoG Real Bench, HWiNFO64

http://rog.asus.com/forum/showthread.php?43233-Realbench-v2-Discussion-Thread-Download-Links
http://www.hwinfo.com/download.php

Usage of testing and monitoring programs:

When you open Real Bench, move both windows to left side of screen. Open HWiNFO64, run "sensors only", you will get a pop up asking whether to disable reading the Asus EC chip, click "Disable this sensor". Move the HWiNFO64 window to upper right hand corner of screen. Stretch bottom of window to full screen height. Make the following changes (optional but oit makes life easier):

-Right Click on "System" right at the top, select hide.
-In the next section, hide the last 4 lines starting "Core CPU Thermal Throttling" (if you watch temps, this is useless)
-Skip over the next section and Hide the section after that (section includes CPU Package thru DRAM Power)
-Now the whole reasons we did that was so you could see everything you wanna see at same time. You should be able to see Vcore 0, 1 and 2 at -the bottom of the window. If not hide a few more lines. Save and Quit will save your edits.

Open Real Bench, select Benchmark Tab Check only the last box. Open HWiNFO64, run "sensors only" as described above. Start Real Bench and don't touch mouse till finished. Then try checking all 4 boxes and run again NOTE: During the 3rd test Open CL will send AVX instructions to CPU; pay close attention to Vcores as they will spike as described above. If passes and you want to go further ,,,,

Switch to the Stress Test Tab and select the amount of RAM you have in your system and 2 hours..... (Note: If you plan on raising cache and / or RAM after a run, I will usually save the two hours and skip this step until I have Multiplier / Cache and RAM speed at my targets.

Record the Maximum core temps and voltages.

it hits on each core, 76, 76, 75, 68 at 1.275 on auto with 1 quick spike bumping them up to 1.357v

tektechnician · Sep 25, 2014

Davil :

It is a shame the TIM is so terrible on the 4770k, I probably should have gotten the 4930k as I had the choice, however I believe it was on preorder due to the fact that it was later "released" a week later.

JackNaylorPE · Sep 25, 2014

Davil :

100% of what ? There are two problems with prime95.

1. Most people are not using the version with AVX instruction sets ... if that's the case, then you are not fully testing the CPU .... I haven't checked in a few months but last I did check, itv was not testing all instruction sets in today's processors.

2. If you are using the AVX version, then that's testing an artificial la\odd that your PC will never see. it's like testing a beam intended to support 1000 pounds buy placing a 3000 pound weight on it. Now if you want to use prime95 because "years back" that was the "go to" tool, I have nothing contrary to advise. But of your goal is to test your CPU under real workd applications, then prime 95 is not the tool.

For one, it presents an AVX load that is unrealistic and would never be duplicated. If on adaptive, as most OC'ers are, your voltages can spike up 0.10 to 0.12 making heat rise drastically. Under real world applications this never happens, AVX instructions may be present but never as a constant hammering. You can do damage to your CPU in such a manner.

Second, you can be P95 stable and then fail under RoG Real Bench..... reason is P95 does 1 thing at a time, stops and then does something else. This does not test the CPU and being subject to a variety of instruction sets at the same time so it can not detect such instabilities.

Third, P95 limits overclocks....Being able to be stable doing that one thing at a time... means the heat and voltage pressed on the CPU is constant and localized to that one instruction subset. The CPU will never see that ina real world scenario..... so you may wind up limiting your OC to say 4.5 because of heat under P95 when you could quite happily run at 4.7 under any combination of very stressful programs you could throw at it.

My 4770k @ the 4.6 profile tops out at 74C at 1.385 volts .... and that's with 46 cache multiplier and 2400 XMP setting....way cooler if i drop cache down to default.

tektechnician · Sep 25, 2014

I think I would probably use Rog Benchmark tool for recording down realworld temps as I am a video editor and a gamer, (yes I know a 4770k is complete overkill for gaming as well as a GTX 780, but its super useful for editing such as CUDA cores). I would use Prime95 for reliability, I don't mind clocking down for the reliability of my computer by a fraction, however, I don't think there is any real world program that stresses a CPU as much as P95.
Also, thanks for helping me out with my question.

Davil · Sep 25, 2014

JackNaylorPE :

100% of the processors available capability... If you open up performance monitor it will tell you what % of your processor's maximum capability is being used.

The AVX instruction sets are primarily only really used in distributed computing programs i.e. folding@home or something similar so really not a concern for me. They can be used for anything but I don't believe they are widely used, but yes prime95 does test that.

It's funny you should mention testing a beam rated for 1000lbs at 3000lbs and also real world, because in the real world they are in fact tested typically to 4x the load they are expected to carry. There is a good reason for that, and it's because you want to be sure it's reliable. Like I said when I run prime95 for 2 hours and have no crashes it doesn't fail while doing anything else. That's the point.

As for voltages in adaptive, yea I believe anything that uses AVX instructions will mess with adaptive voltages, so cool don't see your point.

It's fine to have an opinion but please don't try to make your opinion sound like fact when it's not.

JackNaylorPE · Sep 25, 2014

tektechnician :

Exactly what I'd expect for a 100i at 1.275 volts. That's about the upper limit for a AIO

When you run the Open CL test, you should expect voltaged to bump up 0.1 to 0.12 v as a reult of the AVX instructions

JackNaylorPE · Sep 25, 2014

Davil :

As a practicing engineer for 38 years and owner of an engineering consulting firm, I could spend days talking about materials testing, ..... note that beams that are tested are no longer suitable for service .... Dead loads are multiplied by 1.4 and live loads by 1.7 so that your 1000 pound load used for design purposes is only a fraction or the real load being supplied.

As for my "opinion", Asus, MSI, Gigabyte's all recommend against using prime 95 with Haswell / Ivy Bridge

I use it very carefully, and as Asus and everybody else will tell you, it's very easy to pass a prime 95 test and then be unstable in RoG Real Bench. Start at 4 minute mark.

https://www.youtube.com/watch?v=4mkGQhE1o2w

The old logic of prime95 was that if you could pass P95, you could pass anything that you could throw at it. That has been proven false as P95 stable OC's routinely fail on RoG Real Bench. But Prime95 now presents a load that no combination of programs can get anywhere near .... making it totally useless and limiting..... All you get out of it is being able to say "I passed prime 95". It's now meaningless...... not according to me, according to the Motherboard makers your testing your CPU with.

Davil · Sep 25, 2014

Agree to disagree that is all

Search

4770k high temps with H100i

tektechnician

Reputable

JackNaylorPE

JackNaylorPE

Titan

Davil

Distinguished

tektechnician

Reputable

tektechnician

Reputable

JackNaylorPE

Titan

tektechnician

Reputable

Davil

Distinguished

JackNaylorPE

Titan

JackNaylorPE

Titan

Davil

Distinguished

TRENDING THREADS

Latest posts

Moderators online

Share this page