Overclock.net › Forums › Intel › Intel CPUs › Kaby Lake Overclocking Guide [With Statistics]
New Posts  All Forums:Forum Nav:

Kaby Lake Overclocking Guide [With Statistics] - Page 174

post #1731 of 2900
Quote:
Originally Posted by stryfetew View Post

I did turn the XMP off and it did restart again.

Turning XMP off after you enabled it will leave the settings like they were set by XMP. Load the BIOS default settings first then start playing with the multiplier / Vcore.
Quote:
Originally Posted by stryfetew View Post

I think I see part of the problem now. The actual VID of the chip itself never went above 1.230 so I don't believe the offset was working it was 100% set to auto. Per the guide I set it to 48 with 1.350 on adaptive voltage with XMP turned off. Now HWMonitor shows 1.255v at the VID with the OC at 48. I'm running a stress test now in realbench to see if it's stable, IF it is stable with the overclock I'll give it 24-48 hours to verify it dont restart itself again.

As you said, it's only a guide. You can't copy every setting and expect it to work. Every CPU is different. Some need less then 1.3 V for stable 5 GHz, others need 1.45 V or can't even reach 5 GHz. You have to find it out the hard and long way. The guide give just some starting safe values. Forget the VID, it's not an important metric, it's just what the CPU is requesting. Vcore is the one that you're setting with adaptive voltage and that is needed to be monitored to not be too much off from your setting in BIOS. I don't think 1.35 V is necessary for 48. Maybe on potato chips but I can get away with 1.25 V.


@SweWiking

Memory timings are frequency and voltage dependent. Too have tigher timings without changing the frequency you need to increase the voltage. If you don't or can't increase the voltage, you have to lower the frequency. Here's some graphs and tables to give you a rough idea: http://hw-db.com/memory/2768/g-skill-f4-3200c16d-8gvk-review/2. Benchmark your memory with HyperPi 8M at least (fast), preferably 32M (slow). If you're really into memory OC, read this: http://hwbot.org/newsflash/3058_advanced_skylake_overclocking_tune_ddr4_memory_rtlio_on_maximus_viii_with_alexaros_guide
Edited by dirtyred - 4/8/17 at 9:25am
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
post #1732 of 2900
Disabling XMP and loading automatic puts the memory back at the stock timings and voltages. As for the VCORE nothing seems to be consistent. Like for instance right now VID is 1.255v and VCORE is .994v on one 1.208v on another and CPU VCORE is .552v. Which none of these match anything close to what I set the bios to or what the vid is requesting.. Based on that statement none of these values are right.

post #1733 of 2900
Quote:
Originally Posted by SweWiking View Post

I tried some memory overclock today and yesterday. They are 3200mhz corsair memory. I turned on XMP and tried them at 3200mhz (original) 3333mhz, 34xxmhz, 3600mhz, 3733mhz.

But no matter what speed i try, there is no difference at all in gaming preformance. I tried 3dmark, timespy, valley bench, gta5 benchmark and i get pretty much exactly the same score no matter what the ddr4 is running at. Is there something els i need to change butr the mhz of the memory to get some preformance gain from it ?

well
practically no

like dude suggested you could try to tighten timings further, but that works against the higher speed (and might not show any difference either, except if they're set unreasonably high)

it's hard to find anything in the real world showing a difference beyond 3000

CPU-Z has a score, that should be higher with higher speed

real world usage, mmm
off the top of my head I only know Fallout 4 to be more sensitive to higher memeroy speeds


what you could do is make sure that command rate is set to 1 and not 2
that could increase performace a tad if it isn't set at 1 already


http://www.overclock.net/t/1586767/digital-foundry-memory-overclocking-and-how-it-affects-fps-in-8-different-games
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
post #1734 of 2900
VID has pretty much nothing to do with the Vcore you set in BIOS. Use HFiNFO not HWMonitor as those values seem to be inaccurate (I'm sure there's an explanation for that, maybe your MB is not supported). Vcore and CPU Vcore should be the same value. Also it's important to measure Vcore under full CPU load. When idle the Vcore should drop to around 700-800 mV (0.7-0.8 V).

@peter2k you're right, in games memory speed is not that relevant. But try some memory intensive apps like 3d rendering. You'll get measurable real life gains as few seconds every minute or few minutes every hour between every full memory ratio. Between 2133 and 3200 MHz there could be 3-6 seconds every minute, which is huge (5-10%). Translate that to an 8 hour 3d rendering and it becomes really important.
Edited by dirtyred - 4/8/17 at 10:06am
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 7700 K @ 4.7 GHz ASUS Prime Z270-K Gigabyte GeForce GTX 1060 WINDFORCE OC 6G  G.Skill Ripjaws V 16 GB 
Hard DriveCoolingOSMonitor
Samsung 940 EVO 256 GB Phanteks PH-TC14PE Windows 10 Pro Dell U2312HM 
PowerCaseMouse
Seasonic M12II-620 EVO Fractal Design Define C Razer Taipan 
  hide details  
Reply
post #1735 of 2900
Quote:
Originally Posted by stryfetew View Post

Disabling XMP and loading automatic puts the memory back at the stock timings and voltages. As for the VCORE nothing seems to be consistent. Like for instance right now VID is 1.255v and VCORE is .994v on one 1.208v on another and CPU VCORE is .552v. Which none of these match anything close to what I set the bios to or what the vid is requesting.. Based on that statement none of these values are right.


sigh

err
as I'm on the go at the moment

you wouldn't have set you're voltage to something not manual (means set to adaptive or offset) and you're power plan is set to balanced or energy saver? (I think that low voltage is for when the CPU is at 800Mhz, as it's idle)

to stop confusion
maybe set it to manual for a desired voltage
set power plan to max performace

set you're memory to default
OC you're CPU as desired
and when you know which voltage is needed you could switch to offset or adaptive and try reaching that voltage

then move onto memeroy OC

you can leave the memory at default
the timings and other settings for my memeroy were noted on the packaging it came in

you could set voltage, speed and timings(the main ones) manually

and start working from there (or leave it if it's fine as is)
Edited by peter2k - 4/8/17 at 10:11am
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
post #1736 of 2900
I set the voltage to adaptive and instead of allowing it to be auto I set it to 1.350 with no offset. HWiNFO also shows the lower voltage as well. It's worth noting I haven't failed any stress test, this whole problem occurs when little to no load is on the CPU. Power plan was on balanced jsut changed it to High Performance.

EDIT: Since I did make those voltages changes Real Bench has been running for 98m with no issue.
post #1737 of 2900
Quote:
Originally Posted by dirtyred View Post

VID has pretty much nothing to do with the Vcore you set in BIOS. Use HFiNFO not HWMonitor as those values seem to be inaccurate (I'm sure there's an explanation for that, maybe your MB is not supported). Vcore and CPU Vcore should be the same value. Also it's important to measure Vcore under full CPU load. When idle the Vcore should drop to around 700-800 mV (0.7-0.8 V).

@peter2k you're right, in games memory speed is not that relevant. But try some memory intensive apps like 3d rendering. You'll get measurable real life gains as few seconds every minute or few minutes every hour between every full memory ratio. Between 2133 and 3200 MHz there could be 3-6 seconds every minute, which is huge (5-10%). Translate that to an 8 hour 3d rendering and it becomes really important.

I'm not sure if I edited it in
but I mentioned beyond 3000

also note he didn't mention any software

only 3dmark and games

I mean WinRar would show a difference too

I'm sure professional software would show something

but then you should be using a 6800k or better or a Ryzen if saving time equals more money (as in productivity)
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
post #1738 of 2900
Quote:
Originally Posted by stryfetew View Post

I set the voltage to adaptive and instead of allowing it to be auto I set it to 1.350 with no offset. HWiNFO also shows the lower voltage as well. It's worth noting I haven't failed any stress test, this whole problem occurs when little to no load is on the CPU. Power plan was on balanced jsut changed it to High Performance.

EDIT: Since I did make those voltages changes Real Bench has been running for 98m with no issue.

there could be issues with certain voltage settings and switching between low and high power states
maybe

I'm just thinking loudly here


but maybe when the system is idle it has a low setting in voltage, which is normal
but a background task could make the CPU spike to a higher freqency (think browser, antivirus, explorer, sensor software anything really)

and at those spikes there's not enough voltage making the system unstable

but under high load (stresstest) it's max voltage is applied evenly, making the system stable as long as the voltage doesn't change to lower states


well either moving to manual or getting better control over how and how much is applied would yield better results

means leaving less things at auto


I'm kinda old school OC

I start with manual and see where things go that way
and when I have time try to dabble with the other settings

shrugs
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
post #1739 of 2900
Temperature aside, what is the max Vcore for not shorten the CPU life using it 12h / day for 7 days / week? The canonic 1.4?
post #1740 of 2900
Quote:
Originally Posted by fevion View Post

Temperature aside, what is the max Vcore for not shorten the CPU life using it 12h / day for 7 days / week? The canonic 1.4?

if we just knew that thumb.gif

gonna say yes on 1.4 (I think that's also what der8auer recommends)
though if you love hammering you're CPU with prime even less

but frankly if you're in it for fun and just try to find the highest clock you'll notice the sweet spot

I need to bump voltage from 1.375 to 1.45 to get from 5.2 to 5.3
definitely beyond the sweet spot biggrin.gif

but if you're "only" in it for the magical 5Ghz then you might have luck with not even getting near 1.4
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
Black n RGB
(17 items)
 
  
CPUMotherboardGraphicsRAM
i5 7600K 5.2 Ghz delid at 1.376v MAXIMUS VIII FORMULA NVIDIA GeForce GTX 1060 3GB Kingston HyperX OC to 3200 CL17 
RAMHard DriveHard DriveCooling
Kingston HyperX OC to 3200 CL17 samsung 840 evo Samsung 830 pro NZXT Kraken X62 
OSMonitorKeyboardPower
Windows 10 LG 29EA93-P Cougar Vantar Gaming Sharkoon WPM600 
CaseMouseMouse PadOther
Thermaltake Core P5 Logitech Performance MX Asus ROG Sheath MX Desk Mount 
Other
NZXT RGB HUE+ 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Kaby Lake Overclocking Guide [With Statistics]