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

Haswell Overclocking Guide [With Statistics] - Page 1202

post #12011 of 19652
Quote:
Originally Posted by stolemyowncar View Post

^
Quick question folks... is this really the preferred method of stress testing these CPU's? I ran V1 for about 24 hours with no issues. Temps capping out at 71C for 4.4Ghz at 1.285Vcore (I could probably go lower on the vcore). Is it really a good stress tester? The temps are pretty low for a stresser... o.o

If you hit 71C at 1.285 on x264 you will probably get close to 90's on prime small fft's.
I'd say, if temps are not an issue, go for 3h+ of prime (preferably latest beta) small fft's. But at some point i ran prime small for ~2h 30min and similar on large fft's, manually aborting it thinking that it's enough.., only to bsod later that day after 2h 45min of video encoding (with megui which uses x264).
It is unlikely to bsod under a non synthetic load if you passed x264.

One question though, did you also multitask for any period of time during those 24h of test ?

Quote:
I'm running V2 right now. Is it going to be a better stresser?

marginally..
post #12012 of 19652
After 3 months of continuous tinkering and 2 mother boards later, I finally gave up on trying to push my i7-4770k and manage only a modest OC

2s19feb.jpg

Username: kangk81
CPU Model: 4770K
Core Multiplier: 44X
CPU VID: 1.35
Vcore: 1.36-1.49(adaptative)
Uncore Multiplier: Auto
Uncore Voltage: Auto
Input Voltage: 1.776 - 1.792 (adaptative)
Cooling Solution: custom loop running CPU(delided), PCH & VRM with 360x30mm rad
Stability Test: F@H for 19hrs (24 Apr 1600H, UTC +0 till 25 Apr 1100H, UTC +0)
Batch Number: Malay 323
Ram Speed: XMP 1600MHZ
Ram Voltage: 1.5V
LLC Setting: Auto
Motherboard: ASUS M6F - I only manage to get 44x on this. While I can get 45X stable on MSI GD65.
Shite got real
(13 items)
 
  
CPUMotherboardGraphicsGraphics
i7-4770k asus maximus Vi formula MSI GTX760 TF 4Gb MSI GTX760 TF 4Gb 
RAMHard DriveCoolingOS
Corsair Vengence Pro 2x8Gb Samsung EVO 500Gb SSD Corsair H80i Win 8.1 64bit 
MonitorKeyboardPowerCase
Philips 242G5 Gaming Monitor Razer Blackwidow Corsair AX1200i Corsair Carbide 500R 
Mouse
Razer Orochi 
  hide details  
Reply
Shite got real
(13 items)
 
  
CPUMotherboardGraphicsGraphics
i7-4770k asus maximus Vi formula MSI GTX760 TF 4Gb MSI GTX760 TF 4Gb 
RAMHard DriveCoolingOS
Corsair Vengence Pro 2x8Gb Samsung EVO 500Gb SSD Corsair H80i Win 8.1 64bit 
MonitorKeyboardPowerCase
Philips 242G5 Gaming Monitor Razer Blackwidow Corsair AX1200i Corsair Carbide 500R 
Mouse
Razer Orochi 
  hide details  
Reply
post #12013 of 19652
Quote:
Originally Posted by ConnorMcLeod View Post

From my little experience, i can tell you that you can save time starting tests with low voltages, because for example 5x1h tests with high voltages that pass tests take longer than 5x1h tests on low voltages that crashes in few minutes wink.gif

Well I knew that my OC was unstable at like 1.305 at 4.5Ghz, but just barely (would have been fine at 1.31 prolly). I'd been running fine for weeks but then had a crash on Civ 5 after running through turns like a mad person. Which might be a decent test, if someone could put together a benchmark that simulates Civ 5 turns lol.

So for this one I did a decrease of 25mV from where I knew my 4.5Ghz would be stable at which I already thought was a big enough decrease. I don't want to go over 1.3Vcore because I think it causes degradation. I know it's unstable at this voltage doing Prime 95 already. I tried it and got to like 9xC and then a prompt BSOD after about 30 secs. I don't think I could do Prime 95 ever. So considering I can't do Prime on it, I think it's safe to say that 1.285 is already on the lower end. I could try going lower and seeing if it'll at least pass x264, though. It seems to be much less demanding than Prime 95.
Quote:
Originally Posted by angelotti View Post

If you hit 71C at 1.285 on x264 you will probably get close to 90's on prime small fft's.
I'd say, if temps are not an issue, go for 3h+ of prime (preferably latest beta) small fft's. But at some point i ran prime small for ~2h 30min and similar on large fft's, manually aborting it thinking that it's enough.., only to bsod later that day after 2h 45min of video encoding (with megui which uses x264).
It is unlikely to bsod under a non synthetic load if you passed x264.

One question though, did you also multitask for any period of time during those 24h of test ?
marginally..

Yep, I'm doing all my forum posting on this computer while running the benchmarks. So Waterfox is always open. I'm doing gaming on my secondary rig while doing stress tests and web surfing on this. I actually did do a 13 hour stress test before the one I already did, but I noticed that at 8 threads the process for x264 wasn't taking up 100% of the CPU. Like 20% of it was taken up by Waterfox and some MS service. So I just redid it with 16 threads and well here I am. Seems to be fine.


Honestly I might need to redo my thermal paste application at some point. I think the Kraken X40 can do better than this.
post #12014 of 19652
With my i7 4770k my temps are usually from 35-42 on idle with max temp of 82 degrees , ambient temp anywhere from 21-26 depending on the air con in the room, its in a fractal design xl r2 case so it doesnt have the best of air flow due to the design of case. Are these temps normal?

I had a really strange shut down before when i was playing the elder scrolls online , the computer just shut itself down and then powered back up, not sure if it was a power surge or what, no error log in event viewer and no error in bsod viewer either. Its also on a asrock z87 oc formula mainboard.

Volts are in bios,
12v - 12.091
3v - 3.36
5v - 5.112

Its a brand new corsair cs750m

I also notice when i game my volts go from the standard 12 down to 11.985, is this normal?

Many thanks
post #12015 of 19652
Quote:
Originally Posted by darkelixa View Post

With my i7 4770k my temps are usually from 35-42 on idle with max temp of 82 degrees , ambient temp anywhere from 21-26 depending on the air con in the room, its in a fractal design xl r2 case so it doesnt have the best of air flow due to the design of case. Are these temps normal?

I had a really strange shut down before when i was playing the elder scrolls online , the computer just shut itself down and then powered back up, not sure if it was a power surge or what, no error log in event viewer and no error in bsod viewer either. Its also on a asrock z87 oc formula mainboard.

Volts are in bios,
12v - 12.091
3v - 3.36
5v - 5.112

Its a brand new corsair cs750m

I also notice when i game my volts go from the standard 12 down to 11.985, is this normal?

Many thanks

I guess the 12V going down a little should be normal. I'm using AX1200i and I run FAH 24/7. my 12V rail is reading 11.904V from HWMonitor and 5V is at 4.960
Shite got real
(13 items)
 
  
CPUMotherboardGraphicsGraphics
i7-4770k asus maximus Vi formula MSI GTX760 TF 4Gb MSI GTX760 TF 4Gb 
RAMHard DriveCoolingOS
Corsair Vengence Pro 2x8Gb Samsung EVO 500Gb SSD Corsair H80i Win 8.1 64bit 
MonitorKeyboardPowerCase
Philips 242G5 Gaming Monitor Razer Blackwidow Corsair AX1200i Corsair Carbide 500R 
Mouse
Razer Orochi 
  hide details  
Reply
Shite got real
(13 items)
 
  
CPUMotherboardGraphicsGraphics
i7-4770k asus maximus Vi formula MSI GTX760 TF 4Gb MSI GTX760 TF 4Gb 
RAMHard DriveCoolingOS
Corsair Vengence Pro 2x8Gb Samsung EVO 500Gb SSD Corsair H80i Win 8.1 64bit 
MonitorKeyboardPowerCase
Philips 242G5 Gaming Monitor Razer Blackwidow Corsair AX1200i Corsair Carbide 500R 
Mouse
Razer Orochi 
  hide details  
Reply
post #12016 of 19652
Ah yes im pretty sure 1% is quite normal drop for a psu, what are the temps that you get for your cpu?
post #12017 of 19652
Is hyper-pi a good stress test for RAM ?
Is it a good test to determine if it is better to have higher frequencies+higher timings or lower frequencies+lower timings ?

From tests i made i had those results (average 32M/normal time) :

RAM Kingston 12800 CL9 2x4Go http://www.kingston.com/datasheets/KHX1600C9D3K2_8GX.pdf :

1600MHz :
9-9-9-27 CR2 : 8min50 (XMP)
9-9-9-24 CR1 : 8min39

1866MHz :
9-10-9-24 CR1 : 8min16

2133MHz
10-11-10-28 CR1 : 7min58

So i should stay with the 2133 setting ?


Also, the tutorial i followed said that if hyper-pi test is passed, setting is ok, but it doesn't seem to be able to pass same p95 tests (1344) with 1866 setting than with 1600 (not xmp)
My actual settings are core ratio 45, vid 1.30, cache ratio 42, ring voltage 1.23, vccin 2.1 (will lower than one later), dram 1.65

So should i try to play with another setting ? Raise dram ? stay with xmp profile ?

What about XTU memory test and AIDA64 mem only test ?
Edited by ConnorMcLeod - 4/25/14 at 6:36am
 
Sia's rig
(15 items)
 
LGA775
(15 items)
 
CPUMotherboardGraphicsRAM
i5-4670k msi z87-g45 gaming Gygabyte gtx 760 Kingston Genesis 1600 2x4Go CL9 
Hard DriveHard DriveCoolingOS
SSD Samsung PRO 128Go western digital caviar green 500Go cooler master hyper 212 evo Windows 7 x64 
MonitorKeyboardPowerCase
Samsung SyncMaster 2233 Logitech Ultra Flat Corsair CX600 V2 Fractal Design Core 3300 
MouseMouse PadOther
Logitech g500 SteelSeries 4hd SteelSeries Siberia v2 
CPUMotherboardGraphicsRAM
E8400 Asus P5Q-E msi gtx 560 se cdiscount 2x2Go 
RAMHard DriveCoolingOS
kingston 1x2Go Seagate 500Go Cooler Master Hyper 212 EVO Windows 7 64 premium sp1 
MonitorKeyboardPowerCase
Samsung syncmaster 2233z logitech k310 corsair cx500 Antec NSK4480 
MouseMouse PadAudio
logitech g5 steelseries S&S Steelseries siberia v1 
  hide details  
Reply
 
Sia's rig
(15 items)
 
LGA775
(15 items)
 
CPUMotherboardGraphicsRAM
i5-4670k msi z87-g45 gaming Gygabyte gtx 760 Kingston Genesis 1600 2x4Go CL9 
Hard DriveHard DriveCoolingOS
SSD Samsung PRO 128Go western digital caviar green 500Go cooler master hyper 212 evo Windows 7 x64 
MonitorKeyboardPowerCase
Samsung SyncMaster 2233 Logitech Ultra Flat Corsair CX600 V2 Fractal Design Core 3300 
MouseMouse PadOther
Logitech g500 SteelSeries 4hd SteelSeries Siberia v2 
CPUMotherboardGraphicsRAM
E8400 Asus P5Q-E msi gtx 560 se cdiscount 2x2Go 
RAMHard DriveCoolingOS
kingston 1x2Go Seagate 500Go Cooler Master Hyper 212 EVO Windows 7 64 premium sp1 
MonitorKeyboardPowerCase
Samsung syncmaster 2233z logitech k310 corsair cx500 Antec NSK4480 
MouseMouse PadAudio
logitech g5 steelseries S&S Steelseries siberia v1 
  hide details  
Reply
post #12018 of 19652
Quote:
Originally Posted by RedKnight7 View Post

Thanks again Cyro, let me play with that then. I will be working with my 4.4 Ghz situation ...

So I should set Cache and RAM to Auto (low) settings to make them behave, then see if I can lower VID while I ... what exactly? Raise Vrin from +0.5 to +0.6 over VID, and/or raise LLC ... and all the while I'll see if it lets me reduce VID (and still be stable), right?

I am asking real specifically 'cuz maybe my problem is, a basic flaw in my approach.

I live in a vast OC desert! arrowheadsmiley.png Nobody local OCs.

Yea pretty much

1.38 to 1.32 is a massive reduction, you just need to play around with vcore and core multiplier while other stuff is at stock, and occasionally double-check that your VRIN isn't causing problems. Past about 1.2 to 1.25vcore, it can help to work up in 100mhz steps, noting each one, which at least gives you a trend line

If you're using prime, it's a lot harder on CPU than x264 so it will kill you for most smallish mistakes, i only use it to double-check final vcore's when i can run stuff like x264, with stuff like fft1344-1344 only

Quote:
Originally Posted by stolemyowncar View Post

^
Quick question folks... is this really the preferred method of stress testing these CPU's? I ran V1 for about 24 hours with no issues. Temps capping out at 71C for 4.4Ghz at 1.285Vcore (I could probably go lower on the vcore). Is it really a good stress tester? The temps are pretty low for a stresser... o.o

I'm running V2 right now. Is it going to be a better stresser?

If you're willing to bump voltages a little afterwards or adjust to a few crashes, it's hard to beat - linpack can be at 100c while x264 is still in the 50's thumb.gif
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
post #12019 of 19652
Quote:
Originally Posted by ConnorMcLeod View Post

Is hyper-pi a good stress test for RAM ?
Is it a good test to determine if it is better to have higher frequencies+higher timings or lower frequencies+lower timings ?

From tests i made i had those results (average 32M/normal time) :

RAM Kingston 12800 CL9 2x4Go http://www.kingston.com/datasheets/KHX1600C9D3K2_8GX.pdf :

1600MHz :
9-9-9-27 CR2 : 8min50 (XMP)
9-9-9-24 CR1 : 8min39

1866MHz :
9-10-9-24 CR1 : 8min16

2133MHz
10-11-10-28 CR1 : 7min58

So i should stay with the 2133 setting ?


Also, the tutorial i followed said that if hyper-pi test is passed, setting is ok, but it doesn't seem to be able to pass same p95 tests (1344) with 1866 setting than with 1600 (not xmp)
My actual settings are core ratio 45, vid 1.30, cache ratio 42, ring voltage 1.23, vccin 2.1 (will lower than one later), dram 1.65

So should i try to play with another setting ? Raise dram ? stay with xmp profile ?

What about XTU memory test and AIDA64 mem only test ?
Quote:
Is hyper-pi a good stress test for RAM ?
No.

If it can't pass Prime, it's unstable. Unstable RAM is not a very good thing.

Use tests like Maxxmem to benchmark performance as you go, i like to test the RAM module capability with CPU @4ghz (~1.07vcore) because if it works there it'll work at your CPU OC, unless your IMC does not allow you to be stable which is a different issue and not related to etc memory voltage
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
post #12020 of 19652
i clocked down my ram and tightened timings.
still at the (high) stock voltage, but better than expected:

1866 @ 8-8-9-24 1.65V
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]