Discussion in 'Overclocking & Cooling' started by dom99, Jul 16, 2013.
Which could be your problem. If it doesn't feel right it probably isn't.
IBT pushes chips far harder then Prime does. Most of the time temperatures will be about 10c higher in IBT then Prime. And to be stable in prime you kind of have to run CPU for multiple hours. 4+ is usually the standard.
With Haswell chips, especially when overclocked, coolers like the Hyper 212 are just not enough anymore. They did well on Sandy and Ivy, but Haswell is a whole other ball game.
Its getting so hot because of that voltage. Im not exactly sure why it is running at that voltage if you have left everything at stock. Something is going on. Maybe set voltage to fixed mode at like 1.2v and see what happens during the test.
The voltage is the problem if it's at 1.32v vcore. For 4.4GHz core/4.3GHz ring (cache) I need 1.23v for both. Temperature (excluding IBT which is not operable when overclocked) is in the mid-80c range normally on a Thermaltake Water 2.0 Performer push/pull mounted in the front fan area of my case.
You don't need a high capacity cooler for these chips because they dissipate under 100w of heat when properly overclocked which is insignificant compared to AMD FX or LGA 2011 chips. Rather, you need a cooler that can offer the least "thermal resistance" and/or carry the heat away from the source effectively like AIO or custom water cooling solutions do. In other words, keeping the heatsink itself cooler likely matters more than the actual size of the heatsink, as a big heatsink that is just sucking in 30-40c ambient air isn't going to do much for a chip that already has high thermal resistance between the die and the IHS thanks to cheap thermal interface material used in production of the processor.
Yes, that is way to much voltage for 4.0 and why is it running 4.0 , that is a OC .
make sure you don't have TPU switch on on that Asus MB and don't use the auto OC stuff .
If you want to OC small amount just raise multiplier on turbo cores , you shouldn't need to mess with voltage below 4.2 or so .
Ok just went into the bios and set it back to "normal" and ran Intel Burn test which gave a maximum temperature of 64 Degrees so yes as you say it was the huge voltage.
Lesson learnt, never use the Asus optimised performance option in the bios.
I just hope there isn't any lasting damage to the CPU, is it possible? What benchmarks can I run to make sure the performance is not effected?
You barely ran it under the previous conditions. Damage is negligible
Yes but IBT is short and hot. Prime might not be as hot but its still F#$KING HOT. Keep the heat down. Short heat (within reason) is better than LONG exposure. 15 minutes at 98 or 5 hours at 96. Your choice.
don't worry about damage all you did was get it a little hotter than norm .
If you had retail HS it would be getting close to that hot by default with heavy load .
Ahh, Quantum Tunneling, right? I can't imagine that the additional power draw due to leakage would cause temperatures to increase by that much. I think it's a concern in mainly mobile devices where battery life is important.
The temperature only hit 96 degrees for about 10 seconds.
I'm just not happy with the asus bios setting that doesn't give any warnings that it will lead to extreme voltages.
Your board is overclocking the CPU, and on haswell those temps are normal I have a 4770k here and a Hyper 212+ at those clocks and with the voltages I suspect are being pushed those temps are perfectly normal. This is what people need to expect when not running extreme cooling solutions while overclocked.
Just like Ivy Bridge Haswell can be good or bad in terms of temperatures. add in the higher degree of integration of parts into the die and you get more heat generating components under the IHS which is only transfering heat via TIM to the IHS which then has to go through TIM to the reach the cooler itself. Under these conditions these temps are entirely normal. I would not recommend a Hyper 212 cooler for anything more than keeping temperatures down at stock.
Also with voltages at auto in tests like IBT / Prime95 / Aida FPU etc etc the CPU will hit a higher vcore due to that work load which on my chip is about 0.30 volts higher than any other work load. Its enough to cause a rather large jump in temps. leave everything at stock get your memory running at its proper speeds and buy a better cooler if you intend to overclock.
Right but Haswell seems to be more voltage sensitive than Ivy , where Ivy you can go to like 1.35 and in general not to hot where for Haswell seems 1.25 is around sweet spot .
212 evo should be ok for mild OC as long as voltage is kept low .
On evo , I have mine with modified fan profile set in bios, manual mode ,that cools down temp few more deg .
well my Ivy is voltage sensitive 1.25 volts and i hit 92'C with a water 2.0 pro and 2x 3300 RPM deltas so all depends on the CPU, and i can tell you right now even at 1.2v and 4.2ghz the Hyper depending on ambient temps is going to be a close call one i myself wouldnt make. but then what do I know? I am chief broken rig
I was talking in very general statement . every chip is unique .
That is hot for ivy with water , not high on voltage .
I don't have Haswell and was going by other reports and reviews .
Plus JJ from Asus recommended not going over 1.27v I think it was even with AIO water cause of high temps .
I think for Ivy he recommended around 1.35v , those are not to go over unless you got real good cooling or don't care if it runs hot .
yea every chip is certainly unique all depends on how much gunk is used to hold the IHS on lol.
I told ya. Glad you are sorted. If you want to overclock, always do it manually. Do not increase voltage first and don't set to auto either. Set it to the stock voltage manually. If it is unstable, bump up 0.05V. If it becomes stable, decrease in steps until you find the exact voltage required for stability. After you find it, set the voltage slightly above that value to keep a headroom.
Separate names with a comma.