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

Haswell Overclocking Guide [With Statistics] - Page 910

post #9091 of 19652
Thread Starter 
Quote:
Originally Posted by mandrix View Post


Not sure what you mean by chart, but some time back I did post my settings in this thread for various clock speeds.

x46 = 1.800 Vrin / 1.26 Vcore / 1.055 Vring / x35 uncore
x47 = 1.800 Vrin / 1.31 Vcore / 1.055 Vring / x35 uncore
x48 = 1.800 Vrin 1.900/ 1.33 Vcore / 1.055 Vring / x35 uncore
x49 = 1.900 Vrin / 1.40 Vcore / 1.055 Vring / x35 uncore

x49 is variable, depending on what stress tests I run.
As you can see on the first page, all Haswell OC settings are charted in the Google Doc by filling out this form. This is done for statistical purposes:
 
Username:
CPU Model:
Core Multiplier: [If you used Blck strap, put what Blck and mention your resulting frequency]
CPU VID: This is the CPU core voltage value you input into BIOS.
Vcore: This is the CPU Vcore reading from Hwinfo or HWMonitor under load. "Load" depends on what you're stressing.
Uncore Multiplier:
Uncore Voltage:
Cooling Solution:
Stability Test: [Any test is OK, synthetic or not. List how long test is run.]
Batch Number: [Not required. Malay or Costa Rica chip?]
Ram Speed: [Timings if you know them.]
Input Voltage: [aka VCCIN, Vrin, Eventual Input Voltage]
LLC Setting: [If you didn't change default, say AUTO]
Motherboard: [Optional. Not required to be charted, not required for picture verification.]

 

Quote:
Originally Posted by lombardsoup View Post

Giving up on 4.8ghz, not comfortable with 1.45 + vcore on air.

Here's my 4.5 Mr First Post. I'm too lazy to run prime and IBT again (stable with both but both take too long to run) so here's an x264 quickie.

Username: lombardsoup
CPU Model: 4670k
Core Multiplier: 45x
CPU VID: 1.25
Vcore: 1.28
Uncore Multiplier: 34x
Uncore Voltage: 1.024
Cooling Solution:Phanteks PH-TC14PE
Stability Test: x264
Batch Number: L315B373
Ram Speed: 1600 @ 9-9-9-24 1.5v
Input Voltage: 1.8
LLC Setting: AUTO
Motherboard: MSI Z87-G43 GAMING

You will be charted, thanks.

 

Quote:
Originally Posted by peakclimber View Post

I've back read a few 1000 posts and been following along on this thread. First thanks Darkwizzie for starting it especially the idea of keeping the data in one spreadsheet one can download.

One of the more interesting points I see emphasized here is the relation of Vccin with Vcore. While I read elsewhere the usual "keep Vccin +.4-.5 above Vcore", this thread suggests an even tighter sensitivity between the two. I took the data spreadsheet from the first post and decided to plot Vccin to CPU VID. While Vccin was only provided from some of the lines, I think there is enough to draw some conclusions.

From a linear relationship point of view, the data would suggest CPU VID + .6v is the sweet spot for Vccin.

Trying to apply a polynomial relationship, it might hint that even more is to be added above 1.325 Vcore, ramping even higher with larger Vcore. But I think the data above 1.4 Vcore is a bit questionable given the lack of data points.

Interesting, I'll take a closer look when I get back home.
Undelwalt (2017)
(29 items)
 
 
Celapaleis (2013)
(16 items)
 
CPUMotherboardGraphicsRAM
7600k @ 5.197/1.392 1.392v z170 Asus Hero EVGA 1080ti SC Black (2100/6250) x2 8gb GSkill Trident Z 3600 @ 3804 15-15-15-32-2T 
Hard DriveHard DriveHard DriveCooling
Samsung 950 Pro 512gb (Undelwalt) WD Red 2tb (Pack Yak II) Seagate Expansion Drive 5tb (Phoenix Down II) 3x560mm HardwareLabs Nemesis GTX 
CoolingCoolingCoolingCooling
XPSC Raystorm Pro Watercool Heatkiller 1080ti Full Cover Waterblock EK XTOP Revo Dual D5 (Serial, PWM, v4) EK ZMT (1/2 - 3/4 ID OD) 
CoolingCoolingCoolingCooling
14xEK ACF (Compression Fittings) EK x4 250 (v2) Distilled Water + PT Nuke (Copper Sulphate) 24x140mm Silent Wings 3 
OSMonitorKeyboardPower
Windows 10 Pro 64bit Catleap 27 Inch 2560x1440 IPS 60hz Display Coolermaster Storm Trigger (Brown Switches) EVGA P2 1000w 
CaseMouseMouse PadAudio
Corsair Air 540 Logitech G Pro Tek Syndicate "Raze the World" Desk Mat O2 + Odac by Mayflower Electronics (Rev A) 
AudioAudioAudioAudio
Sennheiser HD 800 2xKrk Rokit 6 G2 Blue Yeti (Grey) Rode PSA-1 Arm 
Other
Aquaero 6 LT + Aluminum Heatsink 
CPUMotherboardGraphicsRAM
i5 6600k Delid @ 4.848/4.848 Asus z170 Hero MSI GTX 980ti @ 1499/4082 1.25v/134% Power 2 x 8gb Gskills Ripjaws V 3131 16-16-16-32 
Hard DriveHard DriveHard DriveHard Drive
Samsung Pro Series 840 256gb (Celapaleis) 1 TB Samsung Drive (Pack Yak) Samsung Pro Series 850 256gb (Celapaleis Reprise) Phoenix Down (External Backup) 
CoolingCoolingOSMonitor
Noctua D14 Kraken x61 Push-Pull for GPU Windows 10 Pro Catleap 2560x1440 @ 60hz 
KeyboardPowerCaseMouse
CM Storm Trigger Brown Switches EVGA Supernova 1000w P2 Corsair 540 Air Razer Abyssus 
Mouse PadAudioAudioAudio
Steelseries QcK Mini Blue Yeti Microphone Objective Dac + Amp Sescom AB Switch 
AudioAudioAudioAudio
Krk Rokit 6 G2 x2 Radioshack Ground Loop Isolator (For Rokits) Wooaudio Aluminum Headphone Stand Sennheiser HD 800 
AudioOther
Logitech c920 Webcam NZXT Sentry Mix 2 Fan Controller 
  hide details  
Reply
Undelwalt (2017)
(29 items)
 
 
Celapaleis (2013)
(16 items)
 
CPUMotherboardGraphicsRAM
7600k @ 5.197/1.392 1.392v z170 Asus Hero EVGA 1080ti SC Black (2100/6250) x2 8gb GSkill Trident Z 3600 @ 3804 15-15-15-32-2T 
Hard DriveHard DriveHard DriveCooling
Samsung 950 Pro 512gb (Undelwalt) WD Red 2tb (Pack Yak II) Seagate Expansion Drive 5tb (Phoenix Down II) 3x560mm HardwareLabs Nemesis GTX 
CoolingCoolingCoolingCooling
XPSC Raystorm Pro Watercool Heatkiller 1080ti Full Cover Waterblock EK XTOP Revo Dual D5 (Serial, PWM, v4) EK ZMT (1/2 - 3/4 ID OD) 
CoolingCoolingCoolingCooling
14xEK ACF (Compression Fittings) EK x4 250 (v2) Distilled Water + PT Nuke (Copper Sulphate) 24x140mm Silent Wings 3 
OSMonitorKeyboardPower
Windows 10 Pro 64bit Catleap 27 Inch 2560x1440 IPS 60hz Display Coolermaster Storm Trigger (Brown Switches) EVGA P2 1000w 
CaseMouseMouse PadAudio
Corsair Air 540 Logitech G Pro Tek Syndicate "Raze the World" Desk Mat O2 + Odac by Mayflower Electronics (Rev A) 
AudioAudioAudioAudio
Sennheiser HD 800 2xKrk Rokit 6 G2 Blue Yeti (Grey) Rode PSA-1 Arm 
Other
Aquaero 6 LT + Aluminum Heatsink 
CPUMotherboardGraphicsRAM
i5 6600k Delid @ 4.848/4.848 Asus z170 Hero MSI GTX 980ti @ 1499/4082 1.25v/134% Power 2 x 8gb Gskills Ripjaws V 3131 16-16-16-32 
Hard DriveHard DriveHard DriveHard Drive
Samsung Pro Series 840 256gb (Celapaleis) 1 TB Samsung Drive (Pack Yak) Samsung Pro Series 850 256gb (Celapaleis Reprise) Phoenix Down (External Backup) 
CoolingCoolingOSMonitor
Noctua D14 Kraken x61 Push-Pull for GPU Windows 10 Pro Catleap 2560x1440 @ 60hz 
KeyboardPowerCaseMouse
CM Storm Trigger Brown Switches EVGA Supernova 1000w P2 Corsair 540 Air Razer Abyssus 
Mouse PadAudioAudioAudio
Steelseries QcK Mini Blue Yeti Microphone Objective Dac + Amp Sescom AB Switch 
AudioAudioAudioAudio
Krk Rokit 6 G2 x2 Radioshack Ground Loop Isolator (For Rokits) Wooaudio Aluminum Headphone Stand Sennheiser HD 800 
AudioOther
Logitech c920 Webcam NZXT Sentry Mix 2 Fan Controller 
  hide details  
Reply
post #9092 of 19652
Quote:
Originally Posted by Aedaric View Post

Mandrix, what voltage do you have your LLC set to and which sensor is it monitored by in HWiNFO if you don't mind?
Are you asking me if I know what the hell I'm doing? smile.gif Again, I'm sure of the voltages and also sometimes monitor with my DMM.....the voltages displayed by HWINFO are very close to what the DMM shows.
LLC is not a voltage, it's load line calibration that (for Haswell) affects vrin/vccin, whichever you want to call it. I have this set to Extreme.

I just finished testing out a bunch of x254 loops @ 4.9 & tweaking, and my final "stable" settings were:
vccin/vrin 2.100
vcore 1.395 (set) 1.404 (load)
vring 1.15 (set) 1.177 (load)
uncore x39





BTW, I did notice that while in the BIOS, the vring reads 2.100v. In Windows of course it's what I stated above.
Edited by mandrix - 1/26/14 at 4:24pm
post #9093 of 19652
Old question asked many times in different generations of i7 biggrin.gif

Has anyone with a non-delidded haswell tried to use CLU/CLP between the IHS and waterblock and compared it to regular TIM between IHS and waterblock?

Any answer from before-after delidding would be appreciated. Wanted to see if it would be more than a 1-2C difference without delidding.

Going to get the loop up and running tonight so wanted some thoughts before I fiddle with the CLU I have left biggrin.gif (no i will not whack this cpu with a hammer just yet!)
  
Reply
  
Reply
post #9094 of 19652
Quote:
Originally Posted by INCREDIBLEHULK View Post

Old question asked many times in different generations of i7 biggrin.gif

Has anyone with a non-delidded haswell tried to use CLU/CLP between the IHS and waterblock and compared it to regular TIM between IHS and waterblock?

Any answer from before-after delidding would be appreciated. Wanted to see if it would be more than a 1-2C difference without delidding.

Going to get the loop up and running tonight so wanted some thoughts before I fiddle with the CLU I have left biggrin.gif (no i will not whack this cpu with a hammer just yet!)

I don't have experience, but I'm going to go ahead and tell you that it will still be a huge difference. The issues isn't the TIM, the issue is the gap between the IHS and the die. No matter what TIM you use on the IHS, if there is still the gap between the die and IHS, it won't make a difference. You probably won't see a major difference between CLU and whatever TIM you were using before on the IHS, so don't bother testing or wasting your CLU. There is a good reason people mainly use CLU only on the dies of chips.
post #9095 of 19652
Quote:
Originally Posted by INCREDIBLEHULK View Post

Old question asked many times in different generations of i7 biggrin.gif

Has anyone with a non-delidded haswell tried to use CLU/CLP between the IHS and waterblock and compared it to regular TIM between IHS and waterblock?

Any answer from before-after delidding would be appreciated. Wanted to see if it would be more than a 1-2C difference without delidding.

Going to get the loop up and running tonight so wanted some thoughts before I fiddle with the CLU I have left biggrin.gif (no i will not whack this cpu with a hammer just yet!)

It makes a pretty small difference, and generally speaking, isn't worth the hassle. Use CLU under the IHS, but don't bother with it on top.
Quote:
Originally Posted by jameyscott View Post

I don't have experience, but I'm going to go ahead and tell you that it will still be a huge difference. The issues isn't the TIM, the issue is the gap between the IHS and the die. No matter what TIM you use on the IHS, if there is still the gap between the die and IHS, it won't make a difference. You probably won't see a major difference between CLU and whatever TIM you were using before on the IHS, so don't bother testing or wasting your CLU. There is a good reason people mainly use CLU only on the dies of chips.

Your answer is correct, but your first sentence does really relate to your answer (it seems to be answering the question of whether it is worth it to delid). There is a marginal difference between CLU and non-CLU on top of the IHS. There is a huge difference between delidding and not-delidding, but the TIM you use on top of the IHS makes little difference.
Edited by Forceman - 1/26/14 at 5:24pm
post #9096 of 19652
Thanks guys, I figured it was same as the other chips. That good ole INTEL TIM and GLUE holding the IHS on top of the chip.

I would normally whack the chip and delidd for the temperatures but I do want to give it a run with my loop and see how it handles temps not-delidded.
(I also have it up for sale so I don't want to risk delidding just yet biggrin.gif )
  
Reply
  
Reply
post #9097 of 19652
Does anyone have any thoughts about whether the x264 Benchmark still makes a good stability test if you lower its priority from high to normal so the system is still usable? I wouldn't do it as a final test of stability, but I hope to use it as a rough guide as I work my way up.
post #9098 of 19652
Quote:
Originally Posted by fritzdis View Post

Does anyone have any thoughts about whether the x264 Benchmark still makes a good stability test if you lower its priority from high to normal so the system is still usable? I wouldn't do it as a final test of stability, but I hope to use it as a rough guide as I work my way up.

It's best to leave it how it is and test for stability overnight. That way it doesn't interfere with your daily tasks and you can fully test for stability.
post #9099 of 19652
IT'S ALIVE!!! IT'S ALIIIIIIIIIVE! It still works! Here's a couple pics. Pretty happy with the results. I'm using NT-H1 on the die (for now, CLU on the way) and IHS.




Corsair Air 540
(13 items)
 
  
Reply
Corsair Air 540
(13 items)
 
  
Reply
post #9100 of 19652
just a quick question does c7 state really require the psu's that support it? cuz mine doesnt support it
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]