F5F Stay Refreshed Power Users Overclocking i9700k - 5ghz tuning assistance needed

i9700k - 5ghz tuning assistance needed

i9700k - 5ghz tuning assistance needed

D
Darknaki33
Junior Member
12
02-24-2018, 07:13 PM
#1
Hello everyone,

I’m reaching out for your help with fine-tuning my overclock configuration. I’m using this PC solely for gaming and streaming, which currently uses about 10% more CPU than normal. I’ve followed the YouTube tutorial and adjusted the settings accordingly.

Details:
- Processor: i7 9700K
- Motherboard: Asus TUF Z390M-Pro Gaming (Wi-Fi)
- RAM: 16GB at 3200MHz
- Cooler: Noctua dh-15, single fan only
- Case: Zalman I3 black
- BIOS settings (based on the guide):
- Ai Overclock Tuner: XMP
- Avx instruction core negative offset: 0
- Asus multi-core enhancement: Auto (BIOS optimizes)
- BCLK aware voltage disabled
- SVID behavior: Typical
- Core ratio limit: 48
- DRAM frequency: DDR4-3200MHz
- Power-saving & performance mode: Auto
- Loadline calculation: 5
- CPU current capacity: auto
- CPU power management: as per video, except IA loadlines are 0.01
- Uncore ratio: auto (43x)

I tested with Apex Legends and streaming at the same time and achieved a stable performance.

Results:
- At 4.8GHz, adaptive voltage target is set to 1.30V
- Idle temperature/voltage: 31°C, 0.625V
- Load temperature/voltage: 70–75°C, peaks at 1.295V and 1.33V
- At 5GHz, adaptive target is 1.35V
- Manual setting at 5GHz gives 1.35V but caused instability with BSODs
- Load temp/voltage: 70–80°C, peaks at 1.335V and 1.34V

I’m puzzled about the discrepancy between the target voltage (1.30V) and the actual values I see in BIOS. My BIOS doesn’t have an option to enable/disable SVID support, even though it’s mentioned on the video. Could this be the issue?

Could you suggest adjustments to help reach 5GHz without overloading the cooler too much? Also, what might be missing from my configuration that’s affecting the voltage stability?

Thanks in advance for your advice.
Best,
Bjzor
D
Darknaki33
02-24-2018, 07:13 PM #1

Hello everyone,

I’m reaching out for your help with fine-tuning my overclock configuration. I’m using this PC solely for gaming and streaming, which currently uses about 10% more CPU than normal. I’ve followed the YouTube tutorial and adjusted the settings accordingly.

Details:
- Processor: i7 9700K
- Motherboard: Asus TUF Z390M-Pro Gaming (Wi-Fi)
- RAM: 16GB at 3200MHz
- Cooler: Noctua dh-15, single fan only
- Case: Zalman I3 black
- BIOS settings (based on the guide):
- Ai Overclock Tuner: XMP
- Avx instruction core negative offset: 0
- Asus multi-core enhancement: Auto (BIOS optimizes)
- BCLK aware voltage disabled
- SVID behavior: Typical
- Core ratio limit: 48
- DRAM frequency: DDR4-3200MHz
- Power-saving & performance mode: Auto
- Loadline calculation: 5
- CPU current capacity: auto
- CPU power management: as per video, except IA loadlines are 0.01
- Uncore ratio: auto (43x)

I tested with Apex Legends and streaming at the same time and achieved a stable performance.

Results:
- At 4.8GHz, adaptive voltage target is set to 1.30V
- Idle temperature/voltage: 31°C, 0.625V
- Load temperature/voltage: 70–75°C, peaks at 1.295V and 1.33V
- At 5GHz, adaptive target is 1.35V
- Manual setting at 5GHz gives 1.35V but caused instability with BSODs
- Load temp/voltage: 70–80°C, peaks at 1.335V and 1.34V

I’m puzzled about the discrepancy between the target voltage (1.30V) and the actual values I see in BIOS. My BIOS doesn’t have an option to enable/disable SVID support, even though it’s mentioned on the video. Could this be the issue?

Could you suggest adjustments to help reach 5GHz without overloading the cooler too much? Also, what might be missing from my configuration that’s affecting the voltage stability?

Thanks in advance for your advice.
Best,
Bjzor

R
ratelslang3
Member
167
03-08-2018, 09:00 PM
#2
Test using XMP enabled. It acts as an overclock, which is where I usually begin. This will put extra stress on the CPU (memory controller) and memory. Run a 30-minute RealBench session while gaming or streaming to assess stability.
Overclocking properly requires time, testing, and patience. The main reason for frustration during overclocking is impatience. You appear ready to follow the right steps and conduct thorough tests. After establishing a stable baseline, you can gradually increase your CPU frequency and test again to find a balanced setting.
R
ratelslang3
03-08-2018, 09:00 PM #2

Test using XMP enabled. It acts as an overclock, which is where I usually begin. This will put extra stress on the CPU (memory controller) and memory. Run a 30-minute RealBench session while gaming or streaming to assess stability.
Overclocking properly requires time, testing, and patience. The main reason for frustration during overclocking is impatience. You appear ready to follow the right steps and conduct thorough tests. After establishing a stable baseline, you can gradually increase your CPU frequency and test again to find a balanced setting.

T
TotO_Gaming_
Member
59
03-09-2018, 03:17 AM
#3
You're checking the relevant metric for voltages, noting the distinction between VID and Vcore. Also, if the LLC 5 setting indicates a significant voltage increase under load, consider reducing it to a minimal rise or just enough to prevent further drop.
T
TotO_Gaming_
03-09-2018, 03:17 AM #3

You're checking the relevant metric for voltages, noting the distinction between VID and Vcore. Also, if the LLC 5 setting indicates a significant voltage increase under load, consider reducing it to a minimal rise or just enough to prevent further drop.

K
56
03-09-2018, 08:49 AM
#4
Thank you for your prompt response.
The voltages come from HWinfo via the CPU's VID; the Mobo voltage is based on the Vcore from the MOBO part. For the LLC, there are 7 levels where a higher number indicates a smaller vdrop, so I plan to reduce it gradually.
K
KinoahKrazyK31
03-09-2018, 08:49 AM #4

Thank you for your prompt response.
The voltages come from HWinfo via the CPU's VID; the Mobo voltage is based on the Vcore from the MOBO part. For the LLC, there are 7 levels where a higher number indicates a smaller vdrop, so I plan to reduce it gradually.

S
star11931
Junior Member
12
03-09-2018, 04:52 PM
#5
The primary use of the LLC configuration is typically limited to LN2 overclocking.
S
star11931
03-09-2018, 04:52 PM #5

The primary use of the LLC configuration is typically limited to LN2 overclocking.

B
blueunicorn05
Junior Member
30
03-12-2018, 04:22 PM
#6
The 1.35v setting might be slightly excessive for your existing cooling setup. What stability level are you aiming for? Are you interested in Prime95 with small FFT and AVX stable? Avoid using games as stability checks since they add complexity. There are many factors to consider. A reliable stress test tool is RealBench. Execute the RealBench stress test at normal speeds and settings, using half your RAM capacity for 30 minutes, and check if it succeeds. Also monitor the temperatures reaching their limits. You might also run HWiNFO64 in the background (just sensors) to verify if any components are flagged under 'performance limit reasons.'
B
blueunicorn05
03-12-2018, 04:22 PM #6

The 1.35v setting might be slightly excessive for your existing cooling setup. What stability level are you aiming for? Are you interested in Prime95 with small FFT and AVX stable? Avoid using games as stability checks since they add complexity. There are many factors to consider. A reliable stress test tool is RealBench. Execute the RealBench stress test at normal speeds and settings, using half your RAM capacity for 30 minutes, and check if it succeeds. Also monitor the temperatures reaching their limits. You might also run HWiNFO64 in the background (just sensors) to verify if any components are flagged under 'performance limit reasons.'

G
GameBoosh
Senior Member
470
03-13-2018, 04:06 AM
#7
I'm not very confident about the stability, but the maximum load I'd apply is running with 13% CPU usage while playing Apex Legends, which uses AVX. Other games I play don't consume much CPU, so I'm okay even if it doesn't pass the benchmarks. Regardless of what I do, Prime95 has consistently failed on the small version.

I've managed to bring LLC down to 2 with the 4.8ghz adaptive setting and it runs smoothly, though it causes a BSOD during the 8GB RAM benchmark after just a minute. With the 5ghz adaptive setting, it crashes at level 4 when opening Firefox after booting.

Update: LLC 2 with 4.8ghz adaptive has BSODs.
Update: LLC 3 with 4.8ghz works well, I adjusted SVID to best-case, disabled multicore enhancement, set target voltage to 1.28; Vcore is around 1.27 and spikes up to 1,288/1,296 under load as the maximum.
G
GameBoosh
03-13-2018, 04:06 AM #7

I'm not very confident about the stability, but the maximum load I'd apply is running with 13% CPU usage while playing Apex Legends, which uses AVX. Other games I play don't consume much CPU, so I'm okay even if it doesn't pass the benchmarks. Regardless of what I do, Prime95 has consistently failed on the small version.

I've managed to bring LLC down to 2 with the 4.8ghz adaptive setting and it runs smoothly, though it causes a BSOD during the 8GB RAM benchmark after just a minute. With the 5ghz adaptive setting, it crashes at level 4 when opening Firefox after booting.

Update: LLC 2 with 4.8ghz adaptive has BSODs.
Update: LLC 3 with 4.8ghz works well, I adjusted SVID to best-case, disabled multicore enhancement, set target voltage to 1.28; Vcore is around 1.27 and spikes up to 1,288/1,296 under load as the maximum.

M
MichaelFW
Member
171
03-13-2018, 07:35 AM
#8
Used version HWBOT_RealBench_v.243; all settings reset to defaults via F5 in BIOS, left XMP turned off. Stress test succeeded with 8GB RAM. Temperatures ranged from 37 to 82, average 76°C. Within performance limits, only utilization met "yes"; all other conditions failed at the end. Full log with HWinfo attached: https://easyupload.io/qbhkkc
In the spreadsheet, additional performance limit entries also showed "yes".
M
MichaelFW
03-13-2018, 07:35 AM #8

Used version HWBOT_RealBench_v.243; all settings reset to defaults via F5 in BIOS, left XMP turned off. Stress test succeeded with 8GB RAM. Temperatures ranged from 37 to 82, average 76°C. Within performance limits, only utilization met "yes"; all other conditions failed at the end. Full log with HWinfo attached: https://easyupload.io/qbhkkc
In the spreadsheet, additional performance limit entries also showed "yes".

J
JEMY
Junior Member
30
03-29-2018, 04:31 PM
#9
Test using XMP enabled. This is essentially an overclock attempt, which I usually begin with. It puts extra strain on the CPU (memory controller) and memory. Run a 30-minute RealBench session while gaming or streaming to assess stability. True overclocking requires time, testing, and patience. The main reason for frustration during overclocking is rushing through it. You appear ready to follow the right steps and conduct thorough tests. After establishing a stable baseline, you can gradually raise your CPU frequency and see where it balances best. A helpful resource is available here: https://forums.
J
JEMY
03-29-2018, 04:31 PM #9

Test using XMP enabled. This is essentially an overclock attempt, which I usually begin with. It puts extra strain on the CPU (memory controller) and memory. Run a 30-minute RealBench session while gaming or streaming to assess stability. True overclocking requires time, testing, and patience. The main reason for frustration during overclocking is rushing through it. You appear ready to follow the right steps and conduct thorough tests. After establishing a stable baseline, you can gradually raise your CPU frequency and see where it balances best. A helpful resource is available here: https://forums.

T
TehFruitNinja
Junior Member
15
03-29-2018, 05:44 PM
#10
Thank you for the feedback
@alceryes
I managed to locate a 4.8 all-core setup with up to 1.296v usage and a max temperature of 70°C while streaming and gaming. Although it fails under benchmark stress tests, the reduced temperature and quieter performance meet my needs perfectly for my use case.
T
TehFruitNinja
03-29-2018, 05:44 PM #10

Thank you for the feedback
@alceryes
I managed to locate a 4.8 all-core setup with up to 1.296v usage and a max temperature of 70°C while streaming and gaming. Although it fails under benchmark stress tests, the reduced temperature and quieter performance meet my needs perfectly for my use case.