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

Haswell Overclocking Guide [With Statistics] - Page 85

post #841 of 19539
Quote:
Originally Posted by FtW 420 View Post

A screenshot taken while the cpu is under load should show any vdroop or vrise that occurs, along with a screen at the end.
There is always going to be a kind of honor system as far as the voltages go, software isn't completely accurate but it is all there is for most people, when haswell first came out no software could read the right voltage so a guy's word was all there was. Not posting any screens does seem weird though.
I was going to compare cpu-z with hwinfo & hwmonitor on the multimeter, I'll have to get that done yet. I've always used cpuz since it is basically the standard, hwinfo is handy to see some of the other voltages though. Not sure if I've used hwmon before...

I wasn't trying to implicate you or say that you did anything wrong, I was just using your screenshot as an example because it happened to be the most recent. I appreciate your (and other's) input and taking the time to post your results, cause I know how much time and work goes into it. I just figured that most everyone is posting the voltage reported by CPUz because it matches the manual input in BIOS, and some screenshots back this up. I just don't like people like Zvejnix trying to spin it like I am doctoring my results because that is not the case. My results are there for anyone to see and use if they like. We were told that manual shouldn't introduce any extra voltage, so I had no reason to believe any different. I too would appreciate any further insight into this and I am currently doing some inquiries of my own at Asus and Intel (but who knows if I will get a response).

In doing a little research I have found that HWMonitor is known for providing false readings on some boards, and they are still providing compatibility updates for some Sandy and Ivy chips. I am just hesitant to believe HWMonitor's Vcore offset voltage reading just yet because it behaves very strangely on my board. I will post a few screenshots below of the varying readings I am getting from different programs. Also the weird part is that the Vcore reading fluctuates (all the way down to 0.000V) even when in Manual mode, and the VID stays fixed where it should. Besides if we are all using Manual to stress then it is just easier and more relevant to report the voltage we set in BIOS and that appears in CPUz.

Also please excuse my outburst towards Zvejnix, it had been a long time coming. He seems to spend more time judging other's results than posting his own and I don't appreciate it. He was the source of my rant in the Haswell Owner's Thread last month about people with inadequacy issues telling others that their results are bogus or that they haven't stressed enough. This is not a contest, it is just posting results to help add to the data on Haswell OCs, that is all. And maybe we can all learn a little more about this generation in the process. I enjoy discussing the various nuances of overclocking Haswell, and I am here to learn and contribute to the discussion. I am not trying to stir the pot, I just don't like being called a liar or ignorant. If I were the former I wouldn't post my results and if were the ladder I would just blindly accept the different readings that all the 3rd party software reports. But anyways FTW 420 I wasn't trying to do the same thing to you and I just hope you didn't take it that way...

This photo shows that while in Manual mode the Vcore value still fluctuates while the VID (1.281V) remains static like it should. HWMonitor reports 1.136V and AI Suite is showing 0.000V.

While this picture shows that both AI Suite and HWinfo are showing different values for all 4 cores while in Adaptive. Also the voltage values seem a bit low, even for idle. I tend to believe CPUz over both of these values.

This photo just shows more variation while in Adaptive mode. The values nearly ever match up...

Edited by BangBangPlay - 8/18/13 at 4:57pm
Sager NP6350
(9 items)
 
   
CPUMotherboardGraphicsRAM
Intel i7-3630QM Clevo W350ETQ Nvidia GTX 660M 12 GB Samsung DDR3 1600 MHz 
Hard DriveHard DriveOptical DriveOS
Crucial M4 256 GB WD Black 750 GB Toshiba DVD-RW Drive Windows 7 (64 bit) 
Monitor
LG Phillips 
  hide details  
Reply
Sager NP6350
(9 items)
 
   
CPUMotherboardGraphicsRAM
Intel i7-3630QM Clevo W350ETQ Nvidia GTX 660M 12 GB Samsung DDR3 1600 MHz 
Hard DriveHard DriveOptical DriveOS
Crucial M4 256 GB WD Black 750 GB Toshiba DVD-RW Drive Windows 7 (64 bit) 
Monitor
LG Phillips 
  hide details  
Reply
post #842 of 19539
Thanks. biggrin.gif

From what I can tell temps are about par for the course. Highest temp I saw was when running the FPU only test in AIDA and that topped out at 92C.

I'll probably push for atleast 4.6 if not 4.7 depending on how much voltage I'll need and what my temps are like.

I don't intend on delidding atleast for the time being. My wife would kill me if I stuffed up and had to fork out for another CPU. lol
post #843 of 19539
Quote:
Originally Posted by Forceman View Post

There really isn't much point (from a performance perspective) in pushing the cache speed - if you were stable with it at 35 then I would either leave it there or slowly move it up until you run into problems. Jumping straight to 42 probably isn't the best course of action. But, in addition to Vring, you can also try bumping up VCCSA and VCCIOD and see if those help with stability. You may also need more Vcore to run at the higher cache speed.

Thx. I rolled back vring to 35 and as a test lower my memory oc back. A rep for the help. At this point I still consider that I am in testing phrase. Haswell does have more setting to play with. I also feel that some of my monitoring software might not be accurate since the chip is so new. For example in core temp, I only get temp, frequency, VID and CPUID but all the other info are blank. However, I would still like some proof that people are getting the result they are saying, if not anything, it gives me something to compare my chip to and to get a general idea where mine is performance wise.
ThermalTron
(16 items)
 
  
CPUMotherboardGraphicsRAM
4790k MSI Z87 Mpower MSI 7870 Team Xtreem Dark 
Hard DriveHard DriveCoolingOS
Western digital Samsung 840 pro H100 Windows 8 64bit enterprise 
MonitorMonitorMonitorKeyboard
Apple HCD Dell LG Corsair K90 
PowerCaseMouseMouse Pad
Kingwin 850 Platinum Thermaltake Chaser MK1 TT Theron Roccat Sense Glacier Blue 
  hide details  
Reply
ThermalTron
(16 items)
 
  
CPUMotherboardGraphicsRAM
4790k MSI Z87 Mpower MSI 7870 Team Xtreem Dark 
Hard DriveHard DriveCoolingOS
Western digital Samsung 840 pro H100 Windows 8 64bit enterprise 
MonitorMonitorMonitorKeyboard
Apple HCD Dell LG Corsair K90 
PowerCaseMouseMouse Pad
Kingwin 850 Platinum Thermaltake Chaser MK1 TT Theron Roccat Sense Glacier Blue 
  hide details  
Reply
post #844 of 19539
Quote:
Originally Posted by kinzx View Post

Thx. I rolled back vring to 35 and as a test lower my memory oc back. A rep for the help. At this point I still consider that I am in testing phrase. Haswell does have more setting to play with. I also feel that some of my monitoring software might not be accurate since the chip is so new. For example in core temp, I only get temp, frequency, VID and CPUID but all the other info are blank. However, I would still like some proof that people are getting the result they are saying, if not anything, it gives me something to compare my chip to and to get a general idea where mine is performance wise.

Yeah well despite posting plenty of documentation there will always be someone who wants to discredit you around here, especially if your voltage happens to be a little lower then theirs. I focus on the value entered in BIOS because that is what others can immediately relate to, whether there is an offset that occurs in the integrated voltage regulator inside the die.

Anyways in my experience raising the vring (cache ratio on Asus boards) will require a small bump in core voltage to stabilize. Raising my cache from 38 to 42 on 4.6 GHz required at bump from 1.218V to to 1.223V to stabilize. I also raised the cache voltage too. It is advised that you find your core voltage first without overclocking the cache or memory and then dial those in afterwards. That way you can focus on each part of your OC one at a time and if you run into problems you know which values to change more or less. Raising the cache will effect stress testing because it bottleneck less, but it won't effect real world usage much. So if you are looking for less overall voltage and better thermals then leave the cache at or below your turbo (38 or 39). I like to try to keep mine within 300-400 GHz of my OC if I can. Good luck!
Sager NP6350
(9 items)
 
   
CPUMotherboardGraphicsRAM
Intel i7-3630QM Clevo W350ETQ Nvidia GTX 660M 12 GB Samsung DDR3 1600 MHz 
Hard DriveHard DriveOptical DriveOS
Crucial M4 256 GB WD Black 750 GB Toshiba DVD-RW Drive Windows 7 (64 bit) 
Monitor
LG Phillips 
  hide details  
Reply
Sager NP6350
(9 items)
 
   
CPUMotherboardGraphicsRAM
Intel i7-3630QM Clevo W350ETQ Nvidia GTX 660M 12 GB Samsung DDR3 1600 MHz 
Hard DriveHard DriveOptical DriveOS
Crucial M4 256 GB WD Black 750 GB Toshiba DVD-RW Drive Windows 7 (64 bit) 
Monitor
LG Phillips 
  hide details  
Reply
post #845 of 19539
Quote:
Originally Posted by combatant3219 View Post

Ok so at 1.22 vcore, 45x multi I've passed IBT, Aida64 (CPU/FPU/Cache/Memory) for 12 hrs, Aida64 (FPU Only) for 12 hours, could encode in handbrake no problems and prime 95 ran for several hours. This included browsing etc while running these.

I may try dropping vcore a little to somewhere between 1.21-1.22 but otherwise I think I'm gonna call that stable and move on to uncore.

What do you think?

Yea try lowering the VCore, you may actually have a decent chip.
Quote:
Originally Posted by Bartouille View Post

2400MHz memory at 1.6v is that good?? First experience overclocking memory redface.gif

It's good, what clocks did you set?
post #846 of 19539
I went back to 2133 because for some reason 24 multiplier increased my timings and the memory-copy was actually slower!!
junkman
(17 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-4770K @ 4600MHz ASUS Z97-A/USB 3.1 HIS R9 280 IceQ X² OC @ 1200/1500MHz HIS R9 280 IceQ X² OC @ 1200/1500MHz 
RAMHard DriveHard DriveHard Drive
G.SKILL F3-2133C9D-16GXH @ 2400MHz 10-12-12-31-1T Crucial MX300 275GB SAMSUNG Spinpoint F3 1TB SAMSUNG Spinpoint F3 1TB 
CoolingCoolingCoolingOS
[CPU] Noctua NH-D14 (NF-P14/NF-P12) [GPU1] Asetek 570LC (Gentle Typhoon AP-30) [GPU2] Asetek 570LC (Gentle Typhoon AP-30) Windows 7 Professional 64-bit 
MonitorMonitorPowerCase
ViewSonic XG2401 Dell UltraSharp U2713H EVGA SuperNOVA 1300 G2 CORSAIR 300R 
Other
Scythe Kaze Q-12 Fan Controller 
  hide details  
Reply
junkman
(17 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-4770K @ 4600MHz ASUS Z97-A/USB 3.1 HIS R9 280 IceQ X² OC @ 1200/1500MHz HIS R9 280 IceQ X² OC @ 1200/1500MHz 
RAMHard DriveHard DriveHard Drive
G.SKILL F3-2133C9D-16GXH @ 2400MHz 10-12-12-31-1T Crucial MX300 275GB SAMSUNG Spinpoint F3 1TB SAMSUNG Spinpoint F3 1TB 
CoolingCoolingCoolingOS
[CPU] Noctua NH-D14 (NF-P14/NF-P12) [GPU1] Asetek 570LC (Gentle Typhoon AP-30) [GPU2] Asetek 570LC (Gentle Typhoon AP-30) Windows 7 Professional 64-bit 
MonitorMonitorPowerCase
ViewSonic XG2401 Dell UltraSharp U2713H EVGA SuperNOVA 1300 G2 CORSAIR 300R 
Other
Scythe Kaze Q-12 Fan Controller 
  hide details  
Reply
post #847 of 19539
Yep, if you can't tighten the timings, then stock is always better.
post #848 of 19539
Quote:
Originally Posted by Bartouille View Post

I went back to 2133 because for some reason 24 multiplier increased my timings and the memory-copy was actually slower!!

If you load the XMP profile for 2133 on your memory, then don't touch anything and override the speed to let it run at 2400, the board will probably translate all your XMP profile timings to that new speed. If some timing is 28 at 2133, that would end up 28/2133*2400 = 31.5... rounded to 31 or 32. Check if it does something like that.

After you put things to run at 2400 and tested that your memory will actually run fine at that speed, perhaps after you increased voltage some, you have to tighten all your timings manually. If timings are set manually to a value, they won't change if you change speed.
Edited by deepor - 8/18/13 at 7:53pm
post #849 of 19539
So do I totally discount Prime95 results do you think? Just ran blend again and it crapped out after 3-4 hours. Not sure how much emphasis to put on Prime.

All other tests seem to indicate stability, as per my previous posts.

This is with everything set to Auto except for vcore at 1.22 for 4.5Ghz. My motherboard automatically sets uncore to 39x.
Edited by combatant3219 - 8/18/13 at 7:29pm
post #850 of 19539
Quote:
Originally Posted by combatant3219 View Post

So do I totally discount Prime95 results do you think? Just ran blend again and it crapped out after 3-4 hours. Not sure how much emphasis to put on Prime.

All other tests seem to indicate stability, as per my previous posts.

This is with everything set to Auto except for vcore at 1.22 for 4.5Ghz.

LLC at auto too? What kinda temps are you getting?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]