Did I get a bad i7-4770k

[nickolb]

I have a stock cooler on a 4770k plugged into a MSI Mpower Max motherboard.

Without overclocking when I run prime95 for 1 min the temps are in the upper 90s and Realtemp reports thermal status LOG for 3 of the 4 cores. I believe this means that the CPU clock has been throttled to keep the chip from overheating. A couple of the cores reached 100C maximum. Real Temp and CPU-Z report that the CPU Clock is at 3.7Ghz. CPU-Z reports core voltage at 1.102V

When I enable OC GENIE on the motherboard through the bios it is automatically overclocked to 4.0Ghz and is unstable. Windows fails at launch, causing a Bluescreen failure.

My stock cooler is installed correctly, I have checked it to ensure proper installation.

My questions is do I need to simply upgrade my cooler or did I get a bad chip?

Idle conditions without overclocking anything:
Core Temp: 28C 27C 28C 29C
Clock speed: 800.01 Mhz


**Additional Information**

I overclocked the RAM to its clock rating of 2400Mhz using XMP profile.
The core voltage is now 0.719V and using prime95 it hits 100C almost instantly.

If you need any additional information please post and I will reply as soon as possible.

 

[sosofm]

4770k are mostly the worst CPU's because are heating like hell.This is how INTEL please they customers. Some of 4770k can be overclocked and some not because you can make fry chips on them.Buy an aftermarket cooler Cooler Master Evo is the best performance/ price.

 

[13thmonkey]

don't OC on the stock cooler...

 

[lp231]

Check and make sure the cpu fan is installed properly. Looks like the CPU push pins are locked, but the HSF is never touching the CPU.

 

[nickolb]

don't OC on the stock cooler...

It is my experience that without overclocking anything I am unable to get the advertised 3.9Ghz turbo boost that should be automatic on the chip and my chip appears to be overheating without overclocking. I dont know if this is normal or something to be concerned with.

Those temperatures seem awfully high to me under 100% load. Should the processor be throttling itself because of temperature when it is not overclocked?

 

[ingtar33]

it's not a bad cpu in the sense i expect you mean it. a number of haswell cpus overheat on the stock heatsink in turbo mode alone. Its intel's crappy thermal paste they use inside the heatspreader thats the issue. If you get a real aftermarket heatsink on it, you'll probably be alright up to 4.1 or 4.2ghz... that seems to be where the worst haswell overclockers will get to.

There was a lot of news when Haswell launched about the horrible heat issues with those chips... and on other sites it seems most of the chips coming out of SE Asia have the worst issues with this (while the ones made in the Caribbean seem to overclock the best); Intel's basic response has been "overclocking will void your warranty, and Intel does not even guarantee the max turbo speed".

You can try an exchange with the merchant you bought the chip from, though it will cost you some money i'm sure... intel however will tell you the chip works as intended so its doubtful they'll be any help.

This is why overclocking is called a lottery. every chip is different. yours is among the worst it seems. But it's not broken per say.

 

[nickolb]

it's not a bad cpu in the sense i expect you mean it. a number of haswell cpus overheat on the stock heatsink in turbo mode alone. Its intel's crappy thermal paste they use inside the heatspreader thats the issue. If you get a real aftermarket heatsink on it, you'll probably be alright up to 4.1 or 4.2ghz... that seems to be where the worst haswell overclockers will get to.

There was a lot of news when Haswell launched about the horrible heat issues with those chips... and on other sites it seems most of the chips coming out of SE Asia have the worst issues with this (while the ones made in the Caribbean seem to overclock the best); Intel's basic response has been "overclocking will void your warranty, and Intel does not even guarantee the max turbo speed".

You can try an exchange with the merchant you bought the chip from, though it will cost you some money i'm sure... intel however will tell you the chip works as intended so its doubtful they'll be any help.

This is why overclocking is called a lottery. every chip is different. yours is among the worst it seems. But it's not broken per say.

-I can return to my vendor without a problem. The cost will be minimal. Should I do this?
-My processor was made in Costa Rica
-Should I consider Delidding my processor and applying a good thermal paste?
-Should I just replace the stock cooler?
-Should I delid and replace stock cooler?

I have watched a couple videos demonstrating how to remove the heat spreader on the 4770k processors and feel like this is something I could do successfully. If I were to attempt this I would use the cutting of the adhesive method opposed to the vice method.

 

[13thmonkey]

you are OCing on the stock cooler, with ivb it had to cope with 65W, now it is coping with 77W with hsw. if you have poor case airflow then your cooler will be less effective. Get a better air cooler before you delid, or anything else, it's cheap and effective.

 

[ingtar33]

it's not a bad cpu in the sense i expect you mean it. a number of haswell cpus overheat on the stock heatsink in turbo mode alone. Its intel's crappy thermal paste they use inside the heatspreader thats the issue. If you get a real aftermarket heatsink on it, you'll probably be alright up to 4.1 or 4.2ghz... that seems to be where the worst haswell overclockers will get to.

There was a lot of news when Haswell launched about the horrible heat issues with those chips... and on other sites it seems most of the chips coming out of SE Asia have the worst issues with this (while the ones made in the Caribbean seem to overclock the best); Intel's basic response has been "overclocking will void your warranty, and Intel does not even guarantee the max turbo speed".

You can try an exchange with the merchant you bought the chip from, though it will cost you some money i'm sure... intel however will tell you the chip works as intended so its doubtful they'll be any help.

This is why overclocking is called a lottery. every chip is different. yours is among the worst it seems. But it's not broken per say.

-I can return to my vendor without a problem. The cost will be minimal. Should I do this?
-My processor was made in Costa Rica
-Should I consider Delidding my processor and applying a good thermal paste?
-Should I just replace the stock cooler?
-Should I delid and replace stock cooler?

I have watched a couple videos demonstrating how to remove the heat spreader on the 4770k processors and feel like this is something I could do successfully. If I were to attempt this I would use the cutting of the adhesive method opposed to the vice method.

if you are confident you can do it, you'll likely see a drastic improvement in temps; just know the moment you do it you'll void the warrenty and the merchant won't take the chip back.

I also suggest you fork over $30 for a hyper evo 212

 

[4Ryan6]

I have a stock cooler on a 4770k plugged into a MSI Mpower Max motherboard.

Without overclocking when I run prime95 for 1 min the temps are in the upper 90s and Realtemp reports thermal status LOG for 3 of the 4 cores. I believe this means that the CPU clock has been throttled to keep the chip from overheating. A couple of the cores reached 100C maximum. Real Temp and CPU-Z report that the CPU Clock is at 3.7Ghz. CPU-Z reports core voltage at 1.102V

When I enable OC GENIE on the motherboard through the bios it is automatically overclocked to 4.0Ghz and is unstable. Windows fails at launch, causing a Bluescreen failure.

My stock cooler is installed correctly, I have checked it to ensure proper installation.

My questions is do I need to simply upgrade my cooler or did I get a bad chip?

Idle conditions without overclocking anything:
Core Temp: 28C 27C 28C 29C
Clock speed: 800.01 Mhz


**Additional Information**

I overclocked the RAM to its clock rating of 2400Mhz using XMP profile.
The core voltage is now 0.719V and using prime95 it hits 100C almost instantly.

If you need any additional information please post and I will reply as soon as possible.

It is my experience that without overclocking anything I am unable to get the advertised 3.9Ghz turbo boost that should be automatic on the chip and my chip appears to be overheating without overclocking. I dont know if this is normal or something to be concerned with.

Those temperatures seem awfully high to me under 100% load. Should the processor be throttling itself because of temperature when it is not overclocked?

Turbo Boost is an Intel stock feature you don't have to overclock to use it, I'm not sure how you came to that conclusion?

-I can return to my vendor without a problem. The cost will be minimal. Should I do this?
-My processor was made in Costa Rica
-Should I consider Delidding my processor and applying a good thermal paste?
-Should I just replace the stock cooler?
-Should I delid and replace stock cooler?

I have watched a couple videos demonstrating how to remove the heat spreader on the 4770k processors and feel like this is something I could do successfully. If I were to attempt this I would use the cutting of the adhesive method opposed to the vice method.

First of all delidding the CPU will cost you your CPU warranty, with the temperatures you are experiencing it won't help much anyway!

Never even use auto overclocking features with a stock cooler!

You say you're not overclocking for some of the previous tests but you are overclocking your CPU memory controller running the memory at 2400mhz as the 4770K controller was designed to run at 1333mhz or 1600mhz.

Your CPU may be one that seriously overheats when the memory controller is overclocked, you'll need to get away from the XMP profile and manually set the memory speed lower and the timings and test to see if your temperatures come down.

You could try setting the memory to 1600mhz at timings of 9,9,9,24, with a 2T command rate at 1.50v DDR3 slot voltage and see if it will run those timings manually and retest to see if your temperature drops.

Get an after market CPU cooler, someone previously suggested the Cooler Master EVO, but with the temperatures you are reaching you may need a better cooler than the EVO.

 

[lp231]

Haswell doesn't use a solder as they did with Ivy Bridge, but uses TIM under its IHS. Even though stock fan is crappy at OC, but at stock clocks, the HSF should keep the CPU's temp at a acceptable range even when running Prime 95. To have the CPU temp raised into the 90s around a minute of using Prime 95 at stock clocks, then there must be something wrong with your setup. Either the HSF is not installed correctly, something is wrong with your bios settings or some OC software is messing with the CPU.
Uninstall any OC software you have installed
Reset bios back to defaults
Remove the board and look at the HSF from the side to make sure it's actually installed correctly. Sometimes when installing the HSF from after the board is in the case, will make it seem the HSF is correctly installed, but there might be a millimeter gap which caused it to have high temps.
for OC get a EVO or Noctua DH14 or a Corsair 100i or H110.

 

[nickolb]

Turbo Boost is an Intel stock feature you don't have to overclock to use it, I'm not sure how you came to that conclusion?

I am using Realtemp 3.7 and CPU-Z 1.67.1 x64, both of which report the current clock speed of the cores in the CPU. As the load to the cores increases the core speed increases as well. At idle or under no load my cores are at 800.01 MHz. When they are under 100% load, using prime95, the core speed gets boosted to 3700.46 MHz. Thereby I assuming that my processor never reaches the Turbo Boost maximum for this processor of 3900.00 MHz. Again this is under no overclocking of any type.

FYI: The RAM speed is set to 1333 MHz

Hope that clears it up for you.

You say you're not overclocking for some of the previous tests but you are overclocking your CPU memory controller running the memory at 2400mhz as the 4770K controller was designed to run at 1333mhz or 1600mhz.

There are three different conditions that I reported information on in the original post.

• 
  Condition 1: CPU overclocked to 4.0 GHz; no RAM overclocking--> resulted in system unstability
  Condition 2: No overclocking at all --> resulted in temperature throttling of the cores and excessive core temperatures.
  Condition 3: Overclocked ram to 2400 MHz; no CPU overclocking --> resulted in cores reaching excessive temperatures very quickly and very low core voltage

I suppose I should have been more clear when I was explaining things. Hope this clears it up for ya.

------------------------------------------------------------------------------------------------------------------------------

So everyone please understand that my system is throttling down the clock speed because of heat issues in the cores when NO overclocking of any type is enabled. This is happening with a correctly installed stock HSF. I have triple-checked the installation of the HSF and it is making contact with the heat spreader on the CPU.

Anyone have any clue to what could be causing this temperature problem?
Can anyone confirm that 98-99C is too high for a processor to be operating at under 100% load?
Can anyone tell me what normal operating temperatures are, using a stock cooler, under 100% load?

 

[Nate The Great]

Looks like the main problem is probably your Haswell chip is a bit faulty... And trying to overclock on the stock cooler didn't help very much at all with the faulty hardware, I'd get it RMA'd if possible and ASAP. And yes 98-99C is way too high for 100% load, well it's not too high but I know at those temps you'd start to boil your water-cooling if you had any, so that's obviously too high. 98C is just too high under any amount of load imo, i'd stick to 90 and lower otherwise you risk the destruction of your cpu and mobo at 100C-ish. But this is just my opinion and what I think a safe-zone is.

 

[Nate The Great]

double post.

 

[unclewebb]

It's well known that Intel's latest Haswell technology runs extremely hot. As CPUs get smaller and smaller, there is less surface area to dissipate the heat through the heat sink. I guess Intel does not consider running your CPU fully loaded with Prime95 to be a typical work load. It becomes obvious during this test that the Intel OEM cooling solution is barely adequate.

The target market for the K series CPUs are enthusiasts that intend to overclock. For maximum performance, enthusiasts usually buy aftermarket air and water coolers. To keep costs down, Intel spent as little as possible on the included OEM heatsink and fan and it shows.

The LOG word in the RealTemp - Thermal Status area confirms that the core has reached or exceeded the thermal throttling temperature. Even if a core reaches the thermal throttling temperature for a millisecond, this flag will be triggered within the CPU. RealTemp doesn't even need to be running at the time. As soon as you start RealTemp, it checks to see if there has been any throttling since you booted up. Monitoring software typically samples these temperature sensors only once per second. If throttling is very brief, the maximum recorded temperature might be a degree or two less than the actual maximum temperature.

The 39 multiplier is the maximum multiplier but this is only available when a single core of your CPU is active. As more cores become active, the maximum multiplier decreases. The maximum multiplier when all 4 cores are active is probably 37. This is the default behavior for your CPU. With a K series CPU, most users go into the bios and adjust this so the CPU uses the same maximum multiplier whether 1, 2, 3 or 4 cores are active. That's a nice bonus overclocking feature but it's not how Intel ships these CPUs.

If you run a single threaded benchmark like the built in XS Bench, you should see a multiplier reported by RealTemp close to 39. You won't see the full 39 multiplier during this test because there are always going to be Windows background processes waking up additional cores which instantly lowers the maximum multiplier. If RealTemp does not report multipliers higher than 37.0 during the XS Bench then go into the bios and make sure that you have enabled the low power C3 and C6 C States.

RealTemp T|I Edition
http://www.overclock.net/t/1330144/realtemp-t-i-edition

Is your 4770K bad? It's probably not much different than all the other 4770K CPUs that run hot. Everyone knows the OEM heatsink and fan are inadequate so there is probably not a lot of data out there when running Prime95 to compare to.

 

[nickolb]

Upon visual inspection of the HSF/CPU interface everything appears to be normal and touching. However, I decided to see if something was amiss and discovered that when I physically press the HSF into the CPU the temperatures lower. Could it be that my HSF is not making proper contact with my CPU? How would I go about tightening the HSF to the CPU? I checked and tightened the screws to the HSF but they didn't really move because they are twisted all the way already. A friend suggested that I remove the HSF and apply a "Gob" of thermal paste. Would this work or complicate things?

 

[lp231]

Intel Stock cooler uses push pins. On each of the black push tabs there is a arrow. Turn so the arrow is pointing to the HSF itself, this is the unlock position. Do this to all of them. After that, pull the black tabs until they're release from the bottom of the board.
Next, gently in pull the HSF away from the CPU. Clean off the thermal paste and then apply a new thin layer to the CPU.
Turn the push tabs, where the arrow is facing away from the HSF, this is the lock position. Install the HSF to the CPU, push on the HSF itself until it's all the way down. Next push the black tabs until you hear a click, to lock it in place.
Do it in a X position, do not lock one side and then the other, this will cause 1 side of the HSF to loose contact with CPU.
Do not use a gob of paste, this will not aid in cooling of the CPU. What you want is to use a thin layer.
Need thermal paste, but don't know what to get? Check out this review
http://www.tomshardware.com/reviews/thermal-paste-heat-sink-heat-spreader,3600.html
http://www.tomshardware.com/reviews/thermal-paste-performance-benchmark,3616.html