• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

New to skylake (intel in general) overclocking. Need a bit of help.

Joined
Mar 30, 2013
Messages
7 (0.00/day)
Location
California, USA
System Name HIVEMIND
Processor R9 5950x
Motherboard Asus ROG Crosshair VIII Dark Hero
Cooling Arctic Liquid Freezer II 420
Memory Gskill 3600C18 128GB
Video Card(s) Asrock RX 6700 XT Challenger Pro 12GB OC
Storage 2TB Aorus 7000s + 1TB MX500 + 2x 4TB HDD + 240G SSD scratchdisk
Display(s) M28U
Case Phanteks P500A
Power Supply Seasonic Focus GM-750
Mouse G Pro X Superlight
Keyboard G Pro TKL
So, I finished my rig a few days ago and decided to OC a bit.

(temps are not too much of an issue, since I'm running a Corsair H100i v2 and have to replace the thermal compound in a while anyway)

Currently running my 6600k at 4.6ghz at 1.308-1.344 vcore (at least that's what the bios and other tools tell me it is).

I've done a bit of research, and in going for 4.5+ghz, normally forcing the vcore to 1.3ish just doesn't make it stable (and it is (at stable but a bit hotter at 4.6ghz when i manually force it to 1.38vcore in bios)

On the other hand, when i change the vcore via offset instead, I get rock solid 4.6 with 1.332vcore reported via cpu-z and hwmonitor and bios.

From what I noticed, the bios set vcore is a bit higher than what the bios and other tools report, so should I be worried?

I have not changed any other voltages/timings/settings besides enabling XMP, moving the multiplier to 46-47, changing the fclk multi, and locking ram voltage to 1.2.

Am I missing anything? (due to the lack of board settings, I cannot manually fix vdroop to a certain value, if that is an issue)

I don't intend to push beyond 1.4vcore (better if I can get 4.7-4.8, but I'll get there when I get there)
 
Joined
Jun 21, 2016
Messages
2,058 (0.72/day)
System Name AM4 / 775
Processor 2600x / C2D E7600
Motherboard B450 Aorus / ASUS P5G41C-M LX
Cooling TT Esports Duo / Chinesium cooler
Memory 16GB DDR4 3ghz / 4GB DDR2 800mhz
Video Card(s) 2060 Super / 5700-XT / GTX 650Ti
Storage 120GB + 1TB SSD / 160GB SSD
Display(s) Samsung CRG5 144hz QD
Case CiT shit chassis modded / Coolermaster Elite 430
Audio Device(s) Soundblaster FX / Audigy 2 ZX
Power Supply Superflower Leadex III GOLD / BeQuiet 450w bronze.
Mouse Razer Basilisk
Keyboard Read Dragon Kumara
Software Windows 10 Pro x64
Benchmark Scores 1 Billion
Put volts to 1.45 and put multi to x50 and enjoy 5ghz.

If you need to cool it just dunk it in the freezer for 30 mins prior to use.
 

cadaveca

My name is Dave
Joined
Apr 10, 2006
Messages
17,232 (2.62/day)
turning on XMP can also affect other voltages (VCCSA and I/O). Lowering those can affect temps and power consumption. But stability testing those is arduous.

Stay under 1.4V and you should be fine.

I'd also suggest not using offset, and try using adaptive.
 
Joined
Mar 30, 2013
Messages
7 (0.00/day)
Location
California, USA
System Name HIVEMIND
Processor R9 5950x
Motherboard Asus ROG Crosshair VIII Dark Hero
Cooling Arctic Liquid Freezer II 420
Memory Gskill 3600C18 128GB
Video Card(s) Asrock RX 6700 XT Challenger Pro 12GB OC
Storage 2TB Aorus 7000s + 1TB MX500 + 2x 4TB HDD + 240G SSD scratchdisk
Display(s) M28U
Case Phanteks P500A
Power Supply Seasonic Focus GM-750
Mouse G Pro X Superlight
Keyboard G Pro TKL
not sure if the board supports adaptive, but I do see a bit of fluctuation from the set vcore or offset.

I'm just wondering why the cpu is stable with little to no vcore fluctuation, and less volts shown/needed under offset mode, rather than the almost 1.4 vcore the thing eats when i manually input vcore.
 
Joined
Jun 21, 2016
Messages
2,058 (0.72/day)
System Name AM4 / 775
Processor 2600x / C2D E7600
Motherboard B450 Aorus / ASUS P5G41C-M LX
Cooling TT Esports Duo / Chinesium cooler
Memory 16GB DDR4 3ghz / 4GB DDR2 800mhz
Video Card(s) 2060 Super / 5700-XT / GTX 650Ti
Storage 120GB + 1TB SSD / 160GB SSD
Display(s) Samsung CRG5 144hz QD
Case CiT shit chassis modded / Coolermaster Elite 430
Audio Device(s) Soundblaster FX / Audigy 2 ZX
Power Supply Superflower Leadex III GOLD / BeQuiet 450w bronze.
Mouse Razer Basilisk
Keyboard Read Dragon Kumara
Software Windows 10 Pro x64
Benchmark Scores 1 Billion
If you value cooler temps at idle offset is a good way.
 

cadaveca

My name is Dave
Joined
Apr 10, 2006
Messages
17,232 (2.62/day)
If you value cooler temps at idle offset is a good way.
adaptive is just like running stock and leaves power-saving features enabled. The only difference is that it sets the voltage you choose at load. Offset is only needed if just BCLK OC'ing, and gives higher temps, since you raise voltage under all conditions, including idle.

not sure if the board supports adaptive, but I do see a bit of fluctuation from the set vcore or offset.

I'm just wondering why the cpu is stable with little to no vcore fluctuation, and less volts shown/needed under offset mode, rather than the almost 1.4 vcore the thing eats when i manually input vcore.

Its possible that your board doesn't have adaptive because voltage regulation is now on the board, not the CPU as will Z97, but it should offer it...

You have to define "stable" and how it crashes before I can answer that question accurately.
 
Joined
Mar 30, 2013
Messages
7 (0.00/day)
Location
California, USA
System Name HIVEMIND
Processor R9 5950x
Motherboard Asus ROG Crosshair VIII Dark Hero
Cooling Arctic Liquid Freezer II 420
Memory Gskill 3600C18 128GB
Video Card(s) Asrock RX 6700 XT Challenger Pro 12GB OC
Storage 2TB Aorus 7000s + 1TB MX500 + 2x 4TB HDD + 240G SSD scratchdisk
Display(s) M28U
Case Phanteks P500A
Power Supply Seasonic Focus GM-750
Mouse G Pro X Superlight
Keyboard G Pro TKL
1 hour of intel burn test, some cinebench, P95 small FFTs, 1 hour OCCT, no crashes so far with the vcore manually set to 1.38 OR offset to 1.332

At least that's what preliminary testing tells me. When I get an OC that I think is as far as I can go, I'll move to 6/12 hour testing.

Could be wrong with the adaptive/offset/manual terminologies though, since the board just has auto, manual, and offset (+0.100v, etc)
 

cadaveca

My name is Dave
Joined
Apr 10, 2006
Messages
17,232 (2.62/day)
So when set to manual, when does it crash? Do you stability test the same way (test order etc) every time?

Its possible that offset has CPU current set higher or something, would explain the difference. How are temps compared between the two?

Oh, and FCLK changes? Any particular reason why? Have you actually seen any difference with it set higher than 800 MHz?
 
Joined
Mar 30, 2013
Messages
7 (0.00/day)
Location
California, USA
System Name HIVEMIND
Processor R9 5950x
Motherboard Asus ROG Crosshair VIII Dark Hero
Cooling Arctic Liquid Freezer II 420
Memory Gskill 3600C18 128GB
Video Card(s) Asrock RX 6700 XT Challenger Pro 12GB OC
Storage 2TB Aorus 7000s + 1TB MX500 + 2x 4TB HDD + 240G SSD scratchdisk
Display(s) M28U
Case Phanteks P500A
Power Supply Seasonic Focus GM-750
Mouse G Pro X Superlight
Keyboard G Pro TKL
Nope, but (again, I am new to skylake), a few guides said to push the fclk multi to 10 (or 100 bclk x 10 multi). It may or may not be a good idea, so I'm asking around.

When i set to manual at approx the same vcore (1.332vcore offset vs 1.332vcore manual), the manual crashes/fails the stresstests almost immediately.

For manual, i have to set a bit more juice.

I can try to set fclk to 800 later if it makes any difference
 

cadaveca

My name is Dave
Joined
Apr 10, 2006
Messages
17,232 (2.62/day)
Nope, but (again, I am new to skylake), a few guides said to push the fclk multi to 10 (or 100 bclk x 10 multi). It may or may not be a good idea, so I'm asking around.

When i set to manual at approx the same vcore (1.332vcore offset vs 1.332vcore manual), the manual crashes/fails the stresstests almost immediately.

For manual, i have to set a bit more juice.

I can try to set fclk to 800 later if it makes any difference
if you have most recent BIOS, FCLK should default to 1000 MHz anyway. Earlier BIOSes didn't work so well with 1000 MHz, but new ones should just fine. For most instances, the difference was like 1% for FCLK boost to 1000 MHz, although some weaker GPUs went a bit higher. 1% of 60 FPS is less than 1 FPS...

You definitely have a weird situation. I've tested quite a few Z170 boards and have never run into such stuff in OC, but I have seen quite a few chips that were basically broken in the box. :p

Makes me wonder if it isn't your board's doing though. Perhaps vdroop is higher on manual.
 
Joined
Mar 30, 2013
Messages
7 (0.00/day)
Location
California, USA
System Name HIVEMIND
Processor R9 5950x
Motherboard Asus ROG Crosshair VIII Dark Hero
Cooling Arctic Liquid Freezer II 420
Memory Gskill 3600C18 128GB
Video Card(s) Asrock RX 6700 XT Challenger Pro 12GB OC
Storage 2TB Aorus 7000s + 1TB MX500 + 2x 4TB HDD + 240G SSD scratchdisk
Display(s) M28U
Case Phanteks P500A
Power Supply Seasonic Focus GM-750
Mouse G Pro X Superlight
Keyboard G Pro TKL
And back.

After some tinkering with the mobo, I discovered a few things.

"High" vdroop crashes the entire pc upon starting P95 (at least when i pushed to 4.7ghz with 1.4ish volts). "standard" vdroop makes the stresstest fail, but no full pc crash.

I can stable 4.6 ghz with 1.332 vcore (will try to go lower, maybe) via offset. It went as low as 1.29ish stress test stable.

Still no luck with manual vcore for 4.7 ghz and above... with the exception of one peculiar case.

With around 1.40something vcore and 4.7ghz, i managed to run some stress tests... and for some reason, 1-2 cores were throttling in P95 (like wtf, they aren't even 80 yet) but at least the system wasn't crashing (and no, all powersaving features were turned off in the bios... i think). I can definitely lower temps tomorrow after removing the thermal thing that came with the H100i v2, and replace with AS5 (i should have probably done this in the first place).
 
Top