[SOLVED] 8700k issues

[tytanium]

was gaming one night and all was fine. was playing warzone and getting about 120fps. Next day go to play and im only getting 38 with drops down to like 7fps. all temps were fine.

trouble shot for a few days until i noticed my 4.8 overclock would drop down to 800mhz when i went into the game and then when leaving the game would only go back up to 4.3 and even a stresstest would not get it passed the 4.3. If i reboot it goes back up to the 4.7 until i try and game again.

Aida64 test had my temps at 66.

Someone please help

8700k
32gb 3600
evga 2070super
650w seasonic focus platinum
msi z370 gaming m5

 

[Darkbreeze]

Alright cpu's normally fluctuate depending on the load,

Now you're starting to get it.

 

[supra5mge]

I am going to go a direction no one is going here. Your power supply seems rather weak tbh. That may be causing some of your issue. In my experience, sometimes it can be the simplest of things. Truthfully, with that setup you have, I recommend 750 or 800 watts. If there is any bottleneck in your system, I think that is it.
You are at the minimum spec for that 2070 super. Plus throw in an overclock, she is working overtime I think.

 

[HaizRail007]

I have my associate degree in applied science for information technology and it was alot of sleepless night studying and making sure I knew the content. I will have to say there's been an explosion over the past several years in advancement which i love. I remember my first pc was a Pentium 1 with 150 mhz cpu, 4 Gig hhd and 512 mb's of ram, man though's were the days, nevertheless, I do know what I'm doing, but I'm only as good as the information I've been provided. and yes his problem did state that his 4.8 clock down clocked when under some sort of load, and all I was trying to get the guy to do is run a test with prime 95 while looking at his voltages,temps etc and watching his core clock to see if it dropped while doing the stress test, because yes while under load it's going to be boosted to it's max within prime 95 and hand brake as well. and if his cpu down clocked while running prime 95, then there would of been a problem. That's why it was vital for him to give us all the information so that we would of been aware of all the fact's. In addition, I think where we all went wrong was assuming information he hadn't provided. Nevertheless, No matter how much we know we always can learn more, that's the main reason why the comptia A+ Certification's are important. and keeping up with the latest tech.

 

[Darkbreeze]

I am going to go a direction no one is going here. Your power supply seems rather weak tbh. That may be causing some of your issue. In my experience, sometimes it can be the simplest of things. Truthfully, with that setup you have, I recommend 750 or 800 watts. If there is any bottleneck in your system, I think that is it.
You are at the minimum spec for that 2070 super. Plus throw in an overclock, she is working overtime I think.

His 2070 Super only begs 20w more than my 2060 Super XC ultra gaming, and his 8700k has a 4w higher TDP than my 6700k. Both my 6700k (4.6Ghz OC) and and 2060 Super are overclocked (Driving multiple 144hz 1440p displays), plus I have 7 140mm fans running, RGB controller and light strips, 5 drives, a Mellanox 10Gbps SFP+ network adapter card, two 16GB overclocked DIMMs and four external USB hubs, all running off a 650w Seasonic power supply that isn't all THAT much better than his Platinum Focus unit, and I can assure you that his 8700k plus 2070 Super is not using ANYWHERE near the capacity of that power supply. Unless the unit is faulty, then the PSU is 100% not the problem. It has plenty of capacity.

I would suggest that you download Prime95.

Open HWinfo and familiarize yourself with where the temperature sensor values for the core individual and package core temperatures are at as well as the VRM and PCH temperatures.

Run Prime95 and choose the "Small FFT" option. Not "smallest FFT", just "small FFT". Uncheck the box next to AVX2 and then the box next to AVX so that AVX instructions are disabled for the purpose of testing core temperature and VRMs. Keep an eye on core usage in HWinfo to make sure that no cores are dropping out and that they are all staying at 100% load. If any worker shows less than 100% load at any point during the 15 minutes of testing, then stop the test by going up to the file menu and choosing "Stop". Then choose "Exit". Do NOT simply click the X in the top right corner of the Prime95 application window as that will NOT stop the test. You must click STOP or EXIT from the drop down File menu to stop the test.

Assuming that doesn't happen, monitor the core and package temperatures, VRM temperatures and PCH temperatures and if at any point core temperatures exceeds 80°C, then stop the test. Otherwise, run it for 15 minutes. Once you've stopped the test (And also periodically during testing), check through the HWinfo sensor listings for the Core 0, 1, 2 and 3 "thermal throttling" listing to see if any thermal throttling was indicated.

 

[HaizRail007]

His 2070 Super only begs 20w more than my 2060 Super XC ultra gaming, and his 8700k has a 4w higher TDP than my 6700k. Both my 6700k (4.6Ghz OC) and and 2060 Super are overclocked (Driving multiple 144hz 1440p displays), plus I have 7 140mm fans running, RGB controller and light strips, 5 drives, a Mellanox 10Gbps SFP+ network adapter card, two 16GB overclocked DIMMs and four external USB hubs, all running off a 650w Seasonic power supply that isn't all THAT much better than his Platinum Focus unit, and I can assure you that his 8700k plus 2070 Super is not using ANYWHERE near the capacity of that power supply. Unless the unit is faulty, then the PSU is 100% not the problem. It has plenty of capacity.

I would suggest that you download Prime95.

Open HWinfo and familiarize yourself with where the temperature sensor values for the core individual and package core temperatures are at as well as the VRM and PCH temperatures.

Run Prime95 and choose the "Small FFT" option. Not "smallest FFT", just "small FFT". Uncheck the box next to AVX2 and then the box next to AVX so that AVX instructions are disabled for the purpose of testing core temperature and VRMs. Keep an eye on core usage in HWinfo to make sure that no cores are dropping out and that they are all staying at 100% load. If any worker shows less than 100% load at any point during the 15 minutes of testing, then stop the test by going up to the file menu and choosing "Stop". Then choose "Exit". Do NOT simply click the X in the top right corner of the Prime95 application window as that will NOT stop the test. You must click STOP or EXIT from the drop down File menu to stop the test.

Assuming that doesn't happen, monitor the core and package temperatures, VRM temperatures and PCH temperatures and if at any point core temperatures exceeds 80°C, then stop the test. Otherwise, run it for 15 minutes. Once you've stopped the test (And also periodically during testing), check through the HWinfo sensor listings for the Core 0, 1, 2 and 3 "thermal throttling" listing to see if any thermal throttling was indicated.

I couldn't agree more, besides a 650 watt seasonic is a very good power supply and at the platinum rating it's going to be extremely power efficient, now if it were a bronze's I'd be worried. Nevertheless I'd also download the overclocking checking tool, Even though it's for overclocking it allows you to monitor voltages,temps, power usage and etc. and allow you to narrow down if you do have a problem. However, considering you did state that the turbo features etc was enabled, I'm thinking your cpu is running normally. and that you actually don't have a issue. Nevertheless, it wouldn't hurt to run the tests and report back.

 

[Darkbreeze]

Actually, this is one of those areas where IT people tend to sometimes have a lack of understanding of enthusiast hardware. No offense.

In reality, whether the unit is bronze or Titanium really doesn't make the least bit of difference when it comes to anything other than a little extra heat (Because it's not AS efficient) and a little extra power consumption (Because it's not AS efficient). It wouldn't worry me at all to run his system on a bronze 650w unit IF it was a bronze 650w unit (An thus, platform) that we knew (By virtue of a positive professional review) had good performance, able to supply it rated capacity at 40°C, low ripple and used a decent internal capacitor selection. A 650w bronze unit SHOULD be just as capable of delivering it's power reliably as a Gold, Platinum or Titanium unit.

It just may be a bit louder because it is creating more heat and it's cooling will need to be a bit more aggressive than if it was a higher efficiency model, but that shouldn't affect performance, IF the performance is good to begin with.

If his cores are being abnormally throttled, meaning it's dropping down in frequency when it should be maintaining full frequency 100% core usage behavior, then it will show up in HWinfo during the Prime95 test. Either some cores will not show at 100% or it will specifically SAY there is "thermal throttling" happening in there among the other sensor listings.

 

[HaizRail007]

Actually, this is one of those areas where IT people tend to sometimes have a lack of understanding of enthusiast hardware. No offense.

In reality, whether the unit is bronze or Titanium really doesn't make the least bit of difference when it comes to anything other than a little extra heat (Because it's not AS efficient) and a little extra power consumption (Because it's not AS efficient). It wouldn't worry me at all to run his system on a bronze 650w unit IF it was a bronze 650w unit (An thus, platform) that we knew (By virtue of a positive professional review) had good performance, able to supply it rated capacity at 40°C, low ripple and used a decent internal capacitor selection. A 650w bronze unit SHOULD be just as capable of delivering it's power reliably as a Gold, Platinum or Titanium unit.

It just may be a bit louder because it is creating more heat and it's cooling will need to be a bit more aggressive than if it was a higher efficiency model, but that shouldn't affect performance, IF the performance is good to begin with.

If his cores are being abnormally throttled, meaning it's dropping down in frequency when it should be maintaining full frequency 100% core usage behavior, then it will show up in HWinfo during the Prime95 test. Either some cores will not show at 100% or it will specifically SAY there is "thermal throttling" happening in there among the other sensor listings.

yes I agree with what your saying, but a lower rated power supply will create more dirty power and will have more of a arch which can result in unstable spikes as the psu ages or the user put's more load on it. Nevertheless, this can be said of a Titanium rated psu as well, but would be less noticeable. The only real difference's I've seen between the ratings besides price, was the capacitors and etc that was used. The higher rated psu's use better components at least in theory and should last long under the same loads. However, this isn't always the case either.