i5-4690k core speed won't go higher than 3.5Ghz under full load?

skyflake

Honorable
Apr 13, 2016
5
0
10,510
Hi!

For some reason my i5-4690k core speed won't go higher than 3.5Ghz when stressed with Prime95 (turbo boost is enabled). I think the core speed should go to at least 3.7Ghz under full load. Core voltage under full load is 1.010v, which is quite low :??:
BIOS settings are all in "auto". BIOS is up-to-date.

I'm using CPUID HWMonitor and CPU-Z for monitoring.

Any ideas what might be causing this?

PC specs:
MB: Asus Z97-A USB-3.1
CPU: i5-4690k with Noctua NH-U12S
RAM: Kingston 4GB HyperX Fury, DDR3
HDD: Seagate Barracuda, 1T
GPU: Asus Strix GTX 970
PSU: Seasonic S12G-650
OS: Win7 Professional
 
Solution
If you really want to see boost in action use a light to moderate load, or single core load as suggested above. Full load you will (eventually) see default clock speed. You can try CoreTemp (standalone install, don't use the installer it tries to install chrome or something :/) it allows a taskbar clockspeed indicator for every core.
then it is working exactly as designed. Try loading a single instance of Prime95, so only one core will be stressed. THen the single core will turbo-boost to a higher frequency. Until TDP limitations throttle it, so your cooling should be up to the task.
 
If you really want to see boost in action use a light to moderate load, or single core load as suggested above. Full load you will (eventually) see default clock speed. You can try CoreTemp (standalone install, don't use the installer it tries to install chrome or something :/) it allows a taskbar clockspeed indicator for every core.
 
Solution

skyflake

Honorable
Apr 13, 2016
5
0
10,510
Prime95 with only one core stressed, the core speed stays rock solid at 3.5Ghz.

Then I tried SuperPI and the core speed went up to 3.9Ghz, so it's working as designed I think.

Prime95 won't trigger the turbo boost for some reason.


Thank you for your replies.
 
which version of Prime95 did you run? After a specific version of Prime95 (details here: http://www.tomshardware.com/forum/id-1800828/intel-temperature-guide.html) the instruction set cause the processor to heat up in a specific insttruction set location on the chip. So you run into issues with the TDP envelope of the processor and throttling to keep the processor within that.
 

king3pj

Distinguished


This depends on your motherboard. What you are describing is the way it is supposed to work but my Gigabyte GA-Z97X-UD3H-BK automatically applies the full 3.9GHz turbo boost to all 4 cores on my i5-4690k. I have tested this under gaming load and stress testing load.

When I bought this setup I was planning to overclock so I also got a Hyper 212 EVO cooler. This is enough to keep my turbo boost active when under full load 100% of the time. The best stable overclock I have been able to achieve at reasonable voltages is 4.4GHz. I haven't found a game where 3.9GHz isn't enough yet so I'm not running my overclock until I do.
 


I'm well aware that some motherboards push the Turbo speed. Not a great fan of those motherboards.
Even so, how are your observations relevant to someone who posted that they cannot achieve the turbo clock speeds?
 

king3pj

Distinguished


It's not. His question was already marked as solved when I posted that. I was simply adding to the discussion about the way turbo boost works for other people who may read this thread. I was not questioning the answers that you or anyone else gave the OP.
 

king3pj

Distinguished
To the OP, I know this isn't exactly what you asked about but if you are open to suggestions you could try a modest overclock since you have the cooler and motherboard to support it.

You could most likely change the multiplier to 39 without even touching the voltage. This would give you the full 3.9GHz turbo boost speed on all 4 cores without too much impact on temperature.

I think I was able to get mine to 4.0GHz stable before I had to start messing with voltages. Just start with 3.9GHz and then stress test to make sure you don't get any crashes. If this is stable you can either leave it alone or try increasing the multiplier by 1 until it is no longer stable. When you hit that point back off 1 and you are good to go.
 

skyflake

Honorable
Apr 13, 2016
5
0
10,510


I'm using ver 28.7.
 


Right, as per the link I provided you, V 26.6 might exhibit expected behavior. http://windows-downloads-center.blogspot.com/2011/04/prime95-266.html

The instruction set used in later Prime95 versions might cause the processor to run too hot and thermally throttle the processor.

Unless I overlooked this, do you have a non-stock aka aftermarket cooler? That may also factor in to your findings.

 

skyflake

Honorable
Apr 13, 2016
5
0
10,510


I just ran v26.6 of Prime95. Core speed went up to 3.7Ghz under full load (4 active cores) and it runs ~10C cooler.
So v28.7 produces much higher temps.

I have Noctua NH-U12S, so thermal throttling shouldn't be a problem.

Thank you for the provided info
 


Now temperatures are in play? :)

What were the original temperatures when you couldn't get past 3.5 GHz? if they are now 10C cooler, then thermal throttling could have been a factor.

 

skyflake

Honorable
Apr 13, 2016
5
0
10,510


The original temperatures were 56-60C. So temperatures are ok :)