F5F Stay Refreshed Power Users Overclocking Overclocking the 4790K

Overclocking the 4790K

Overclocking the 4790K

E
EatHam
Junior Member
5
04-22-2016, 02:05 PM
#1
I'm a new builder trying to overclock, so my knowledge is limited compared to others. My 4790K uses a Corsair H110i GT with some Noctua NF-A14 fans at 2000 rpm instead of the original ones. I managed to run Prime95 for 47 minutes at 4.7 GHz and 1.300 volts before the CPU overheated. Because the problem was temperature, I lowered the voltage to check how low it could go without crashing. I reached 1.235 and still ran Prime95 without issues. My temps started in the high 70s at first, but after just 16 minutes they jumped to the high 90s and stayed there. I cleaned and replaced the original thermal paste, and both fans and pump are running at full speed. Anyone have suggestions on what else I could try? Could it be that my cheap CPU cooler is failing? Any advice?
E
EatHam
04-22-2016, 02:05 PM #1

I'm a new builder trying to overclock, so my knowledge is limited compared to others. My 4790K uses a Corsair H110i GT with some Noctua NF-A14 fans at 2000 rpm instead of the original ones. I managed to run Prime95 for 47 minutes at 4.7 GHz and 1.300 volts before the CPU overheated. Because the problem was temperature, I lowered the voltage to check how low it could go without crashing. I reached 1.235 and still ran Prime95 without issues. My temps started in the high 70s at first, but after just 16 minutes they jumped to the high 90s and stayed there. I cleaned and replaced the original thermal paste, and both fans and pump are running at full speed. Anyone have suggestions on what else I could try? Could it be that my cheap CPU cooler is failing? Any advice?

F
fanfaro
Member
119
04-24-2016, 03:55 AM
#2
All programs listed are complimentary. Prime 26.6 causes issues with newer CPUs because of the avx code. 26.6 functions perfectly on Windows 10. Now you understand why your temperatures were unreasonable for anything older than version 26.6 – it’s extremely poor performance. Avoid it. For Aida, the experience is disappointing; it’s just not effective. Stay clear. I’ve run 3-hour passes with it followed by instant crashes during gaming or any stress test. The IBT (Intel burn test) will likely raise your CPU temperature more than any other mentioned method. It’s a simulation. I rely on this for extreme temperature scenarios and rigorous testing. Real-world benchmarks like X264 prove stability over time. Also, some users prefer Aida because it performs better under stress.
F
fanfaro
04-24-2016, 03:55 AM #2

All programs listed are complimentary. Prime 26.6 causes issues with newer CPUs because of the avx code. 26.6 functions perfectly on Windows 10. Now you understand why your temperatures were unreasonable for anything older than version 26.6 – it’s extremely poor performance. Avoid it. For Aida, the experience is disappointing; it’s just not effective. Stay clear. I’ve run 3-hour passes with it followed by instant crashes during gaming or any stress test. The IBT (Intel burn test) will likely raise your CPU temperature more than any other mentioned method. It’s a simulation. I rely on this for extreme temperature scenarios and rigorous testing. Real-world benchmarks like X264 prove stability over time. Also, some users prefer Aida because it performs better under stress.

Z
Zynxx
Member
61
04-24-2016, 05:29 AM
#3
I'm a beginner builder trying to overclock and have limited experience compared to others. My 4790K is cooled with a Corsair H110i GT and some Noctua NF-A14 (2000 rpm) instead of the original fans. I managed to run Prime95 for 47 minutes at 4.7 GHz and 1.300 volts before the CPU overheated. Initially, temperatures stayed in the high 70s during the test, but after just 16 minutes they jumped to the high 90s and didn’t drop. I cleaned and replaced the original thermal paste, and both fans and pump are now running at full speed. Anyone have suggestions on further improvements? Could it be that my CPU cooler is faulty? Any advice?

Prime95 doesn't provide a consistent workload. The stress varies as it processes your system. This spike in temperatures you're seeing is typical. Prime95 isn't ideal for assessing stability during overclocking today—it applies a much higher load than real-world applications. My experience shows that Prime95 reaches only about 70% of the thermal load generated, making it a decent test but not a perfect measure. I used a backup system and ran it normally; it required around three reboots with dynamic voltage adjustments, and stability lasted for eight months.

I currently run my 4790K at 4.5GHz and 1.17V under load, using a dynamic voltage of -0.115v. The Kraken x61 temperatures stayed between 50-60°C under full real-world loads, with fan speeds ranging from 30% to 50% across CPU cores. Video transcoding is one of the most demanding apps I run, but it hasn't matched the load of transcoding-intensive games. It's rare for a game to maintain 100% CPU usage across all eight threads for an hour or two.
Z
Zynxx
04-24-2016, 05:29 AM #3

I'm a beginner builder trying to overclock and have limited experience compared to others. My 4790K is cooled with a Corsair H110i GT and some Noctua NF-A14 (2000 rpm) instead of the original fans. I managed to run Prime95 for 47 minutes at 4.7 GHz and 1.300 volts before the CPU overheated. Initially, temperatures stayed in the high 70s during the test, but after just 16 minutes they jumped to the high 90s and didn’t drop. I cleaned and replaced the original thermal paste, and both fans and pump are now running at full speed. Anyone have suggestions on further improvements? Could it be that my CPU cooler is faulty? Any advice?

Prime95 doesn't provide a consistent workload. The stress varies as it processes your system. This spike in temperatures you're seeing is typical. Prime95 isn't ideal for assessing stability during overclocking today—it applies a much higher load than real-world applications. My experience shows that Prime95 reaches only about 70% of the thermal load generated, making it a decent test but not a perfect measure. I used a backup system and ran it normally; it required around three reboots with dynamic voltage adjustments, and stability lasted for eight months.

I currently run my 4790K at 4.5GHz and 1.17V under load, using a dynamic voltage of -0.115v. The Kraken x61 temperatures stayed between 50-60°C under full real-world loads, with fan speeds ranging from 30% to 50% across CPU cores. Video transcoding is one of the most demanding apps I run, but it hasn't matched the load of transcoding-intensive games. It's rare for a game to maintain 100% CPU usage across all eight threads for an hour or two.

S
Sentio_Cege
Member
246
04-26-2016, 01:06 PM
#4
Dizzario :
jpereira95 :
I'm a beginner in building and trying to overclock, so my knowledge is quite limited compared to others. My 4790K is cooled with a Corsair H110i GT and some Noctua NF-A14 (2000 rpm) instead of the original fans. I managed to run Prime95 for 47 minutes at 4.7 GHz and 1.300 volts before the CPU started getting hot. Since my problem was the temperature, I decided to lower the voltage to see how low I could go without crashing. I reached 1.235 without any issues and continued running Prime95. My temperatures were consistently in the high 70s at the start of the test, but after just 16 minutes they jumped to the high 90s and didn't drop down. I cleaned and replaced the original thermal paste, and both fans and the pump are now running at full speed. Anyone have any suggestions on what else I could try? Am I just using a very cheap build that overheats easily? Any advice?

Prime95 doesn't maintain a constant load. The workload changes as it processes on your system. This is the temperature spike you're observing. Prime95 isn't a reliable tool for assessing system stability during overclocking these days—it applies a much higher load than what you'd actually experience with real-world applications.

My build behaved similarly using Prime95 and no other real-world app I've ever run has reached more than 70% of the thermal load generated by it. While it's useful for testing system stability, it's better to use something more practical for checking your overclock results. I backed up my PC and just used it normally; it required about three reboots with freeze-ups and adjusted the dynamic voltage to a higher level. It's been stable for eight months now.

I run my 4790K at 4.5GHz, 1.17V under load with a dynamic voltage of -0.115V (negative). The Kraken x61 temperatures stayed between 50-60°C under full real-world loads and a fan speed of 30-50% across the CPU cores. Video transcoding is one of the most demanding apps I run, but games haven't matched its load. It's rare for a game to run at 100% CPU usage across all 8 threads for an hour or two.

So, I should just stick to normal operation and see what happens, since my temperatures are usually in the 70s when running Prime95. Sorry for the long explanation, but what's the difference between the regular voltage (1.17V) and the dynamic voltage (-0.115V)? Is my 1.235 enough to push temperatures into the 70s compared to your 50-60°C at 1.17V? Also, keep in mind that your Noctua fans run at full speed in the 70s instead of 50%, so I should be seeing lower temps—why don't they?
S
Sentio_Cege
04-26-2016, 01:06 PM #4

Dizzario :
jpereira95 :
I'm a beginner in building and trying to overclock, so my knowledge is quite limited compared to others. My 4790K is cooled with a Corsair H110i GT and some Noctua NF-A14 (2000 rpm) instead of the original fans. I managed to run Prime95 for 47 minutes at 4.7 GHz and 1.300 volts before the CPU started getting hot. Since my problem was the temperature, I decided to lower the voltage to see how low I could go without crashing. I reached 1.235 without any issues and continued running Prime95. My temperatures were consistently in the high 70s at the start of the test, but after just 16 minutes they jumped to the high 90s and didn't drop down. I cleaned and replaced the original thermal paste, and both fans and the pump are now running at full speed. Anyone have any suggestions on what else I could try? Am I just using a very cheap build that overheats easily? Any advice?

Prime95 doesn't maintain a constant load. The workload changes as it processes on your system. This is the temperature spike you're observing. Prime95 isn't a reliable tool for assessing system stability during overclocking these days—it applies a much higher load than what you'd actually experience with real-world applications.

My build behaved similarly using Prime95 and no other real-world app I've ever run has reached more than 70% of the thermal load generated by it. While it's useful for testing system stability, it's better to use something more practical for checking your overclock results. I backed up my PC and just used it normally; it required about three reboots with freeze-ups and adjusted the dynamic voltage to a higher level. It's been stable for eight months now.

I run my 4790K at 4.5GHz, 1.17V under load with a dynamic voltage of -0.115V (negative). The Kraken x61 temperatures stayed between 50-60°C under full real-world loads and a fan speed of 30-50% across the CPU cores. Video transcoding is one of the most demanding apps I run, but games haven't matched its load. It's rare for a game to run at 100% CPU usage across all 8 threads for an hour or two.

So, I should just stick to normal operation and see what happens, since my temperatures are usually in the 70s when running Prime95. Sorry for the long explanation, but what's the difference between the regular voltage (1.17V) and the dynamic voltage (-0.115V)? Is my 1.235 enough to push temperatures into the 70s compared to your 50-60°C at 1.17V? Also, keep in mind that your Noctua fans run at full speed in the 70s instead of 50%, so I should be seeing lower temps—why don't they?

D
dazzlinglogan
Member
67
04-27-2016, 12:59 AM
#5
Setting your voltage to fixed mode is recommended ideally. Based on your Prime95 temperatures, it seems they aren't accurate. Could you confirm which version you're running? Anything newer than v26.6 would be problematic for Haswell and newer chips. For example, my 4790k handled 4.8v at 1.26v without issues under Prime95 with a Corsair H105 for hours, and it stayed under 82°C. My 5930k at 4.3ghz with 1.3v never exceeded 85°C during stress tests. I use IBT at max for 15 passes, 4 hours on Prime95, 2 hours on Asus real bench, and 10 hours of x264 stress testing. Stress tests like IBT, real bench, Prime95 26.6, and x264 are essential. If you pass all of them, you're fully stable. I personally struggle with any instability. Long gaming sessions and multi-hour renders make a BSD in the middle highly desirable.
D
dazzlinglogan
04-27-2016, 12:59 AM #5

Setting your voltage to fixed mode is recommended ideally. Based on your Prime95 temperatures, it seems they aren't accurate. Could you confirm which version you're running? Anything newer than v26.6 would be problematic for Haswell and newer chips. For example, my 4790k handled 4.8v at 1.26v without issues under Prime95 with a Corsair H105 for hours, and it stayed under 82°C. My 5930k at 4.3ghz with 1.3v never exceeded 85°C during stress tests. I use IBT at max for 15 passes, 4 hours on Prime95, 2 hours on Asus real bench, and 10 hours of x264 stress testing. Stress tests like IBT, real bench, Prime95 26.6, and x264 are essential. If you pass all of them, you're fully stable. I personally struggle with any instability. Long gaming sessions and multi-hour renders make a BSD in the middle highly desirable.

D
DockLate
Junior Member
48
04-27-2016, 03:13 AM
#6
You should ideally set your voltage to fixed mode. Based on your Prime95 temperatures, they don’t seem accurate. Could you confirm which version you’re running? Anything newer than v26.6 is problematic for Haswell and newer chips. For example, my 4790k ran at 4.8v with 1.26v and never exceeded 82°C under Prime for hours using a Corsair H105. My 5930k at 4.3GHz with 1.3V never hit 85°C either. This was tested with stress tests like IBT, real bench, Prime95 26.6, and x264. If you pass all of these, you’re fully stable. I personally can’t tolerate any instability. I often play long gaming sessions and do 10+ hour renders, so having a BSD in the middle is not feasible.

The voltage is currently set to manual mode at 1.235 and overclocked to 4.7GHz, but these numbers may change as I adjust them for optimal performance. Both cooling units are similar, so that’s unlikely the issue. Thank you for sharing the program names—I’ll look into further testing. I checked my Prime95 version (28.7), which might explain the problem. Also, consider that Windows 10 is installed, and I’m unsure if it conflicts with older versions like 26.6. You should also know whether there’s a free version of this software and if IBT, real bench, and x264 are free alternatives.

Appreciate the assistance!
D
DockLate
04-27-2016, 03:13 AM #6

You should ideally set your voltage to fixed mode. Based on your Prime95 temperatures, they don’t seem accurate. Could you confirm which version you’re running? Anything newer than v26.6 is problematic for Haswell and newer chips. For example, my 4790k ran at 4.8v with 1.26v and never exceeded 82°C under Prime for hours using a Corsair H105. My 5930k at 4.3GHz with 1.3V never hit 85°C either. This was tested with stress tests like IBT, real bench, Prime95 26.6, and x264. If you pass all of these, you’re fully stable. I personally can’t tolerate any instability. I often play long gaming sessions and do 10+ hour renders, so having a BSD in the middle is not feasible.

The voltage is currently set to manual mode at 1.235 and overclocked to 4.7GHz, but these numbers may change as I adjust them for optimal performance. Both cooling units are similar, so that’s unlikely the issue. Thank you for sharing the program names—I’ll look into further testing. I checked my Prime95 version (28.7), which might explain the problem. Also, consider that Windows 10 is installed, and I’m unsure if it conflicts with older versions like 26.6. You should also know whether there’s a free version of this software and if IBT, real bench, and x264 are free alternatives.

Appreciate the assistance!

L
Lips
Senior Member
624
05-01-2016, 10:29 PM
#7
All programs listed are complimentary. Prime 26.6 is essential because the avx code damages new CPUs. 26.6 functions perfectly on Windows 10. You now understand why your temperatures were unreasonable for anything newer than version 26.6—it's extremely problematic. Avoid it. For Aida, the experience is poor; it's disappointing. Stay clear. I've run 3-hour sessions with it followed by instant crashes during gaming or any stress test. The IBT (Intel burn test) will significantly raise your CPU temperature compared to other tests. It's a simulation. I rely on this for extreme heat scenarios and rigorous testing. Real benchmarks and X264 provide genuine insights for long-term stability. Also, some prefer Aida because it's simple to pass, though in my view it doesn't push the CPU sufficiently and isn't a true real-world test. Those results are common: my highest readings stay below 85°C, 5 hours of IBT pushed me to 76°C. After 2 hours on real bench, I reached around 72°C, and X264 after 10 hours hit 74°C—this was during a hot summer in a 24°C room.
L
Lips
05-01-2016, 10:29 PM #7

All programs listed are complimentary. Prime 26.6 is essential because the avx code damages new CPUs. 26.6 functions perfectly on Windows 10. You now understand why your temperatures were unreasonable for anything newer than version 26.6—it's extremely problematic. Avoid it. For Aida, the experience is poor; it's disappointing. Stay clear. I've run 3-hour sessions with it followed by instant crashes during gaming or any stress test. The IBT (Intel burn test) will significantly raise your CPU temperature compared to other tests. It's a simulation. I rely on this for extreme heat scenarios and rigorous testing. Real benchmarks and X264 provide genuine insights for long-term stability. Also, some prefer Aida because it's simple to pass, though in my view it doesn't push the CPU sufficiently and isn't a true real-world test. Those results are common: my highest readings stay below 85°C, 5 hours of IBT pushed me to 76°C. After 2 hours on real bench, I reached around 72°C, and X264 after 10 hours hit 74°C—this was during a hot summer in a 24°C room.

M
MONSTERmoose91
Senior Member
526
05-03-2016, 02:29 PM
#8
All programs listed are free. Use prime 26.6. The updated version causes issues with newer CPUs because of the avx code. Version 26.6 should work perfectly on Windows 10. Now you understand why your temperatures were unreasonable—anything newer than version 26.6 is problematic. Avoid it. For Aida, things are worse. It really doesn’t perform well under stress. I’ve run 3-hour passes with it, then experienced instant crashes during gaming or other tests. The IBT (Intel burn test) will likely heat your CPU more than any other test mentioned. It’s a synthetic benchmark. I rely on this for extreme temperature readings and tough testing. X264 and real bench tests are solid for long-term stability checks. Also, some people prefer Aida because it’s easier to pass, even though it’s not a true-world test. In my view, it’s soft, doesn’t push the CPU hard enough, and isn’t reliable for real conditions. Those who like it do so because it’s simple to pass. For more details, my highest temperatures stay below 85°C with IBT, and after 5 hours at 26.6 I reached around 76°C. Real bench tests over 2 hours kept me near 72°C, and X264 after 10 hours hit 74°C—this was during a summer session in a 24°C room.
M
MONSTERmoose91
05-03-2016, 02:29 PM #8

All programs listed are free. Use prime 26.6. The updated version causes issues with newer CPUs because of the avx code. Version 26.6 should work perfectly on Windows 10. Now you understand why your temperatures were unreasonable—anything newer than version 26.6 is problematic. Avoid it. For Aida, things are worse. It really doesn’t perform well under stress. I’ve run 3-hour passes with it, then experienced instant crashes during gaming or other tests. The IBT (Intel burn test) will likely heat your CPU more than any other test mentioned. It’s a synthetic benchmark. I rely on this for extreme temperature readings and tough testing. X264 and real bench tests are solid for long-term stability checks. Also, some people prefer Aida because it’s easier to pass, even though it’s not a true-world test. In my view, it’s soft, doesn’t push the CPU hard enough, and isn’t reliable for real conditions. Those who like it do so because it’s simple to pass. For more details, my highest temperatures stay below 85°C with IBT, and after 5 hours at 26.6 I reached around 76°C. Real bench tests over 2 hours kept me near 72°C, and X264 after 10 hours hit 74°C—this was during a summer session in a 24°C room.