F5F Stay Refreshed Power Users Overclocking I tried overclocking my i5-4690K for the first time and believe I succeeded, but I would appreciate another perspective.

I tried overclocking my i5-4690K for the first time and believe I succeeded, but I would appreciate another perspective.

I tried overclocking my i5-4690K for the first time and believe I succeeded, but I would appreciate another perspective.

Pages (2): 1 2 Next
L
Left_RiteClick
Junior Member
41
04-11-2016, 07:59 AM
#1
So I chose to enhance my PC with a limited budget, diving into overclocking. It seems my CPU was unlocked, and the graphics card was quite standard, making it easy to look up information. Fifteen minutes ago my PC crashed with a blue screen—pretty concerning. I increased my CPU frequency from 3.5 GHz to 4.5 GHz, adjusted the voltage to 1.250, then raised it further to 1.265, hoping this would stabilize things. Temperatures stayed comfortably below 80 degrees, but the fan wasn’t doing much, possibly just maintaining a safe margin.

What do you think? I also boosted the GPU clock by around 165 cycles per core and 210 for memory, without changing the voltage since I don’t plan to sell it. I gathered this info from many videos and forums, so I’m not feeling shy about it. I’ve already seen huge improvements in Assassin's Creed Odyssey—about +10 frames per second in busy zones with lots of NPCs, and the stuttering is gone. Still, full performance was maintained in those areas, though that’s another topic.

Details: MSI GTX 970, CPU i5-4690K, MSI SLI KRAIT z79S, 16 GB RAM, 650W PSU, Cooler Master Hyper 212 Evo cooler.
L
Left_RiteClick
04-11-2016, 07:59 AM #1

So I chose to enhance my PC with a limited budget, diving into overclocking. It seems my CPU was unlocked, and the graphics card was quite standard, making it easy to look up information. Fifteen minutes ago my PC crashed with a blue screen—pretty concerning. I increased my CPU frequency from 3.5 GHz to 4.5 GHz, adjusted the voltage to 1.250, then raised it further to 1.265, hoping this would stabilize things. Temperatures stayed comfortably below 80 degrees, but the fan wasn’t doing much, possibly just maintaining a safe margin.

What do you think? I also boosted the GPU clock by around 165 cycles per core and 210 for memory, without changing the voltage since I don’t plan to sell it. I gathered this info from many videos and forums, so I’m not feeling shy about it. I’ve already seen huge improvements in Assassin's Creed Odyssey—about +10 frames per second in busy zones with lots of NPCs, and the stuttering is gone. Still, full performance was maintained in those areas, though that’s another topic.

Details: MSI GTX 970, CPU i5-4690K, MSI SLI KRAIT z79S, 16 GB RAM, 650W PSU, Cooler Master Hyper 212 Evo cooler.

A
AhkriinGamer
Junior Member
6
04-11-2016, 08:33 AM
#2
The weather is nice and your performance remains steady. You did well.
A
AhkriinGamer
04-11-2016, 08:33 AM #2

The weather is nice and your performance remains steady. You did well.

W
wilkes5
Member
68
04-28-2016, 04:10 PM
#3
I wouldn't suggest it until you've completed a proper stability test. Save and execute Realbench. Set the duration to 8 hours and opt for the 8GB configuration. Avoid using the machine during testing. Beforehand, download and run Prime95 version 26.6 with the small FFT option. Spend about 15 minutes if you stay under 80 degrees, you meet thermal requirements at least slightly. Preferably aim for the 70-75 range, though staying below 80 is safe—just expect a louder fan. More details are available here:
W
wilkes5
04-28-2016, 04:10 PM #3

I wouldn't suggest it until you've completed a proper stability test. Save and execute Realbench. Set the duration to 8 hours and opt for the 8GB configuration. Avoid using the machine during testing. Beforehand, download and run Prime95 version 26.6 with the small FFT option. Spend about 15 minutes if you stay under 80 degrees, you meet thermal requirements at least slightly. Preferably aim for the 70-75 range, though staying below 80 is safe—just expect a louder fan. More details are available here:

P
PerdyPotatoes
Member
145
04-28-2016, 10:17 PM
#4
Darkbreeze would not recommend it yet. Wait until you've completed a full stability check. Save and launch Realbench, set the duration to 8 hours and select the 8GB plan. Avoid using the machine during testing. Before that, install Prime95 version 26.6 and pick the small FFT option. Spend about 15 minutes if you stay under 80 degrees, and you'll be at least slightly compliant thermally. I prefer temperatures between 70-75, but anything below 80 is safe—just expect a noisy fan. More details are available here: It might be useful. Also, my temperature test with Prime95 without overclocking reached 90 degrees, while with overclock it hit 98. That’s concerning—it makes me worried, though it rarely exceeds that at other times. Furmark's CPU test shows readings between 85 and 88 degrees. Assassin's Creed Odyssey in busy zones: 75.
P
PerdyPotatoes
04-28-2016, 10:17 PM #4

Darkbreeze would not recommend it yet. Wait until you've completed a full stability check. Save and launch Realbench, set the duration to 8 hours and select the 8GB plan. Avoid using the machine during testing. Before that, install Prime95 version 26.6 and pick the small FFT option. Spend about 15 minutes if you stay under 80 degrees, and you'll be at least slightly compliant thermally. I prefer temperatures between 70-75, but anything below 80 is safe—just expect a noisy fan. More details are available here: It might be useful. Also, my temperature test with Prime95 without overclocking reached 90 degrees, while with overclock it hit 98. That’s concerning—it makes me worried, though it rarely exceeds that at other times. Furmark's CPU test shows readings between 85 and 88 degrees. Assassin's Creed Odyssey in busy zones: 75.

F
195
04-28-2016, 11:53 PM
#5
What Prime version were you using? Were you running a non-AVX release such as version 26.6, or a newer model that incorporates AVX instructions? Using an AVX version will lead to extremely inaccurate outcomes. If you were on version 26.6 and selected the small FFT option and achieved 90 degrees, then cooling issues may be present. It would be best if your cooler was properly installed. Refer to this guide I prepared: If you weren’t using a non-AVX version, please download the appropriate update and re-test. Games are not a valid method for thoroughly testing CPU thermal performance. Even Furmark cannot be considered sufficient here. Operating temperatures above 80°C can cause serious damage, so exceeding specifications is risky regardless of the software used.
F
Flashplayer551
04-28-2016, 11:53 PM #5

What Prime version were you using? Were you running a non-AVX release such as version 26.6, or a newer model that incorporates AVX instructions? Using an AVX version will lead to extremely inaccurate outcomes. If you were on version 26.6 and selected the small FFT option and achieved 90 degrees, then cooling issues may be present. It would be best if your cooler was properly installed. Refer to this guide I prepared: If you weren’t using a non-AVX version, please download the appropriate update and re-test. Games are not a valid method for thoroughly testing CPU thermal performance. Even Furmark cannot be considered sufficient here. Operating temperatures above 80°C can cause serious damage, so exceeding specifications is risky regardless of the software used.

M
MrSteely
Junior Member
7
04-29-2016, 12:12 AM
#6
Darkbreeze asked about the Prime version they were using. They mentioned running a non-AVX version like 26.6 or a newer one with AVX support, warning that AVX versions can lead to unrealistic results. They noted issues if they used version 26.6 with the small FFT option and achieved 90 degrees, suggesting a possible cooler installation problem. They referenced a guide and advised re-pasting and mounting the CPU cooler if needed. They emphasized that games aren't a proper thermal test for CPUs and Furmark is also unreliable. Running above 80°C risks serious damage, even with Furmark. The user later adjusted the paste pressure and improved performance to around 60 with OC, planning to reapply it at home.
M
MrSteely
04-29-2016, 12:12 AM #6

Darkbreeze asked about the Prime version they were using. They mentioned running a non-AVX version like 26.6 or a newer one with AVX support, warning that AVX versions can lead to unrealistic results. They noted issues if they used version 26.6 with the small FFT option and achieved 90 degrees, suggesting a possible cooler installation problem. They referenced a guide and advised re-pasting and mounting the CPU cooler if needed. They emphasized that games aren't a proper thermal test for CPUs and Furmark is also unreliable. Running above 80°C risks serious damage, even with Furmark. The user later adjusted the paste pressure and improved performance to around 60 with OC, planning to reapply it at home.

C
cladff
Junior Member
24
04-30-2016, 04:54 AM
#7
Darkbreeze: Which Prime version were you using? Were you running a non-AVX release such as 26.6, or a newer model that supports AVX instructions? Using AVX versions can lead to unrealistic outcomes. If you used version 26.6 and selected the small FFT option, achieving 90 degrees could indicate cooling issues. I recommend checking if your cooler is installed properly. Refer to this guide I prepared: If you weren’t using a non-AVX version, try downloading the correct version and re-testing. Games aren’t a valid way to test CPU thermal performance under any circumstances. Neither is Furmark—you must perform a steady-state test with full core load and ensure you stay within the safe temperature range. Operating above 80°C can cause serious damage, so even Furmark’s specs don’t allow that. OMG dude I'm a MOD EDIT: Language I mistakenly thought I was using version 26.6, but I actually used it. I experienced high readings in the low 70s to low 80s during overclocking and the small test. Hope my CPU wasn’t harmed by reaching 100°C a few times. It doesn’t look like it, not in the long term. I’ve read that the maximum safe temperature for a CPU is around 85°C before problems arise. I saw a spike to 83, but during stress testing it was mostly 79. So I’d say it’s okay. Thanks for the tips. I’m unsure if I should do an 8-hour stress test—I might just proceed and monitor closely. Maybe run a stress test for a week to confirm everything is stable. Mod Note: Please keep your tone family-friendly, folks.
C
cladff
04-30-2016, 04:54 AM #7

Darkbreeze: Which Prime version were you using? Were you running a non-AVX release such as 26.6, or a newer model that supports AVX instructions? Using AVX versions can lead to unrealistic outcomes. If you used version 26.6 and selected the small FFT option, achieving 90 degrees could indicate cooling issues. I recommend checking if your cooler is installed properly. Refer to this guide I prepared: If you weren’t using a non-AVX version, try downloading the correct version and re-testing. Games aren’t a valid way to test CPU thermal performance under any circumstances. Neither is Furmark—you must perform a steady-state test with full core load and ensure you stay within the safe temperature range. Operating above 80°C can cause serious damage, so even Furmark’s specs don’t allow that. OMG dude I'm a MOD EDIT: Language I mistakenly thought I was using version 26.6, but I actually used it. I experienced high readings in the low 70s to low 80s during overclocking and the small test. Hope my CPU wasn’t harmed by reaching 100°C a few times. It doesn’t look like it, not in the long term. I’ve read that the maximum safe temperature for a CPU is around 85°C before problems arise. I saw a spike to 83, but during stress testing it was mostly 79. So I’d say it’s okay. Thanks for the tips. I’m unsure if I should do an 8-hour stress test—I might just proceed and monitor closely. Maybe run a stress test for a week to confirm everything is stable. Mod Note: Please keep your tone family-friendly, folks.

S
szymongraczpl
Junior Member
9
04-30-2016, 05:54 AM
#8
andreas.scott :
I dont know if I'm gonna do the 8 hour stress test. I might just run with it, and adjust if I get a blue screen.
THAT, is a very bad idea. I'll tell you why. Don't cut corners. DO the 8 hour test. You can run it overnight so it doesn't affect you, or some time when you'll be gone for 8 hours. Since it's at least marginally thermally compliant, there is no reason you have to be present while the test runs, you just need to see if it matches all hashes and passes or if it does not. Here's why it's important.
andreas.scott :
IF I get a blue screen
Sounds good, doesn't work that way though. It might seem ok until three months later when your system starts spewing errors or misbehaving and you can't figure out what is causing it. You'll end up doing this every three to six months and blame it on Microsoft sending out bad updates. Then at some point you'll realize most of your personal files are corrupt too after a six months to a year of this.
NOT WORTH THE RISK. Test correctly or don't overclock the cpu or use memory settings different than the default or XMP values. Even for XMP settings, testing should still be done. In most cases, the XMP values are both an overclock and a tightening of timings and need to be tested for stability.
An hour isn't enough to do anything besides give you a false sense of security no matter what you are using and most utilities do not actually DO what they are intended to do, which is to ensure stability running processes or calculations known to commonly create errors or corruption on unstable configurations.
Instability does not have to be to the degree where you see blue screens or freezing. It can be at the micro-error level where nothing fails to continue running but small errors like a zero here where a one should be or a couple of ones where zeros should be, and before long you end up with a completely fubared OS or personal files.
S
szymongraczpl
04-30-2016, 05:54 AM #8

andreas.scott :
I dont know if I'm gonna do the 8 hour stress test. I might just run with it, and adjust if I get a blue screen.
THAT, is a very bad idea. I'll tell you why. Don't cut corners. DO the 8 hour test. You can run it overnight so it doesn't affect you, or some time when you'll be gone for 8 hours. Since it's at least marginally thermally compliant, there is no reason you have to be present while the test runs, you just need to see if it matches all hashes and passes or if it does not. Here's why it's important.
andreas.scott :
IF I get a blue screen
Sounds good, doesn't work that way though. It might seem ok until three months later when your system starts spewing errors or misbehaving and you can't figure out what is causing it. You'll end up doing this every three to six months and blame it on Microsoft sending out bad updates. Then at some point you'll realize most of your personal files are corrupt too after a six months to a year of this.
NOT WORTH THE RISK. Test correctly or don't overclock the cpu or use memory settings different than the default or XMP values. Even for XMP settings, testing should still be done. In most cases, the XMP values are both an overclock and a tightening of timings and need to be tested for stability.
An hour isn't enough to do anything besides give you a false sense of security no matter what you are using and most utilities do not actually DO what they are intended to do, which is to ensure stability running processes or calculations known to commonly create errors or corruption on unstable configurations.
Instability does not have to be to the degree where you see blue screens or freezing. It can be at the micro-error level where nothing fails to continue running but small errors like a zero here where a one should be or a couple of ones where zeros should be, and before long you end up with a completely fubared OS or personal files.

A
aalto01
Junior Member
6
05-15-2016, 01:04 AM
#9
Please keep your language appropriate for a family-friendly site.
A
aalto01
05-15-2016, 01:04 AM #9

Please keep your language appropriate for a family-friendly site.

S
Silberflug
Member
203
05-22-2016, 02:57 PM
#10
Another unrelated issue popped up. I was stuck in a boot loop. I took out the CMOS battery for five minutes, kept the power button pressed during that time, and then put it back in. Now I’m seeing a post screen! F1 to start the setup, f2 to keep going, alt+F1 to run the OC profile. It’s detecting all my devices.
Did this result from my overclock? Can I safely restore the OC settings once it passed the stress test at +0.9 GHz?
Edit: selected f2 and it started looping again after I powered off the PC...
Back to POST. No changes made.
S
Silberflug
05-22-2016, 02:57 PM #10

Another unrelated issue popped up. I was stuck in a boot loop. I took out the CMOS battery for five minutes, kept the power button pressed during that time, and then put it back in. Now I’m seeing a post screen! F1 to start the setup, f2 to keep going, alt+F1 to run the OC profile. It’s detecting all my devices.
Did this result from my overclock? Can I safely restore the OC settings once it passed the stress test at +0.9 GHz?
Edit: selected f2 and it started looping again after I powered off the PC...
Back to POST. No changes made.

Pages (2): 1 2 Next