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

Haswell Overclocking Guide [With Statistics] - Page 1091

post #10901 of 19540
It's time to add my system to the chart (hopefully). But firstly I would like to thank Darkwizzie for starting this thread with its great guide. Big thanks to all others too for making this thread very informative vast and active.

This is my first overclock and also first Intel system. I had kinda tight budget so you wont be seeing any "gamers edition" motherboards or high clock XMP memkits. I actually didnt know what XMP was until I started to mess around with my bios. One thing that I did know though. Stock cooler wont do! So I went and bought the best air cooler that the local dealer could offer. Didnt pay that much of an attention to measurements because I have pretty roomy chassis. I still got little freaked out at the dealer when the clerk lifted the box on the table. I remember thinking: Okey, where's that suppose to fit? biggrin.gif Turned out, it was an easy install.
Okey, enough about the hardware.

I started testing with IBT very high 10 runs. I got as low as 1.229 VID with 45*core, uncore on auto (38), LLC etc. on auto (all maxed by BIOS) and with XMP. Then I started to bring up the uncore. first 45 with 1.229 V-uncore, froze up immediately. 44 uncore 1.2 V-uncore same deal. down to 43 uncore with 1.2 V, froze up after 3-4 runs. From 1.2 V-uncore to 1.224 V-uncore it was Antarctica all over. Then with 1.225 V-uncore it did 10 runs. Ok, lets try 20 runs. Out of nowhere BSOD makes it's return. VCCIN 1.85 -> 1.9, no effect. VCCIN 1.9 -> 1.85 and 1.235 VID -> BSOD. Kept increasing VID, until 1.245 stable. So at this point I decided to do the screenshot test with 45/43 1.250/1.225 1.85 VCCIN IBT very high 20 runs. Success!
An hour with Prime95 1344-1344 with 90 % available memory turned out to be a little tougher test. Ended up with 1.260 VID 1.230 V-uncore and 1.860 VCCIN. I increased the V-uncore to 1.230 just for the peace of mind and the VCCIN to 1.860 because I like it to be exact 0.6 higher than VID.
Overnight XTU was only for the sake of having a third different test.

Username: VeliManne
CPU Model: i5-4670k
Core Multiplier: 45
CPU VID: 1.260
Vcore: 1.264
Uncore Multiplier: 43
Uncore Voltage: 1.230
Cooling Solution: NH-D14
Stability Test: IBT-very high 20 runs, Prime95 27.9.1 1344-1344 ~1 hour. XTU-stress 8 hours
Batch Number: 3316b767
Ram Speed: 1600 MHz, 9 9 9 27 XMP
Ram Voltage: 1.65
Input Voltage: 1.860
LLC Setting: Max
Motherboard: Asus Z87-A

Then the pictures. Now, I know that it's not necessary to show them all to have the picture verification charted. But, since I ran them, I just would like to show them all. smile.gif






Edited by VeliManne - 3/9/14 at 11:39am
post #10902 of 19540
Quote:
Originally Posted by Darkwizzie View Post

If you want to throttle at 1.25v or under, be my guest. I failed Linpack at max with 1.25v due to temps on manual. On manual. Let's assume 0.1v hike, that means throttle at 1.15v. What kind of OC will you get with 1.15v? Not a whole lot. And some people in the past called my D14 "magical" for its low temps. 

Myself, I am not throttled at 1.25V. In your example if you are unstable at linpack at 1.25V then raise it until you are stable and you won't be at 1.15V. You will only hit those high volts when running something like linpack. But I said this wasn't for everybody. In my case the increase is not 0.1V, but .016-.02V. This boost in voltage obviously varies.
Quote:

We're back to the whole 'If you want to be FULLY stable you really ought to do this or that...' Yeah, well, that's your dogma. Stability is relative. How long do you run until it's "stable" for that test? That's right, somebody has to decide. The entire process is guess and check. I've gone through so many stupid arguments over this, I won't even go any further. This isn't rhetoric or said for dramatic effect, I'm being literal. It's like beating a dead horse and then ripping off the body parts and beating that up.

I am not suggesting everyone run off and make sure they are stable for everything. I am suggesting one way that they can do this that may or may not apply to them, so take it or leave it.
Quote:

About the 0.1v hike being necessary because the CPU demands it, that's simply assumption. You think Intel would do this and that. Many stress test provides such a high hike in Vcore under adaptive load. I've been Prime stable on 4.5. On override. And I would not be stable on adaptive due to thermal throttling. But wait a minute... Say that Prime 27.9 has AVX and the 0.1v hike is necessary for stability doesn't that mean starving the CPU of the 0.1v boost mean I'd see some level of instability? Some level being a gross understatement. But it goes further than that. Adaptive gives the CPU more voltage. Doesn't that by its very nature make the CPU more stable? Are we arguing that giving the CPU 0.1v+ decreases CPU stability? Because it's like saying a CPU is 1.35v stable, therefore it's 1.25v stable. Not necessarily true. But if it's 1.25v stable, it's very likely 1.35v stable. The first approach is what happens if you stress on adaptive, the latter if you did on manual. So as far as I can tell, it's harder to pass on manual than on adaptive.

You're just saying something is so without real evidence. But Intel made it this way so there must be a reason... that is akin to saying, delidding is pointless because the gap in the glue is on purpose. Intel is one of the most advanced technology companies in the world and they can't get a glue right? Obviously they can, so it must be a reason. You're saying that to truly know stability you must pass AVX on adaptive, which is also pure assumption. You have not and obviously cannot prove this, because I have not done so and I'm dead stable, synthetic Prime stable on manual, on this setting. My own case alone disproves your assertion. Sure, you might argue that it sure helps to stress on adaptive because it roots out instability if it exists but again 1) I've already stated, you're basically saying pumping extra 0.1v+ to the CPU is going to make it less stable 2) There isn't a widespread issue of OCs getting unstable once they start using adaptive, or simply because they did not pass AVX under adaptive (Probably because nobody does that, therefore meaning practically every OC so far refutes your point) 3) Even if you pass 1 and 2, the thermal hurdle alone is enough to turn this into a dead end, causing this method to be unfeasible.
Yes I am saying that and I did provide evidence - I can get a stable overclock against 1.184V fixed with non-AVX tests, but with p95 v28. it is not stable. However it is with AVX adjusted ot 1.2V non-avx load/1.216V AVX. This might not apply to everyone and I stated that in my earlier post.

I am sure there was a reason why Intel quit soldering their IHS and you can infer rightly why, can't you? And people did infer correctly and therefore did delid instead of saying oh,the temperature on this series of processors is too high, so we can't overclock any higher.
Quote:

This thread has closing up on 11,000 posts. Do you know how many times I've had this kind of conversation before? A person that is willing to seriously argue this point is already beyond debating. I'm not responding to you on this topic any further unless you provide something that totally blows my mind. Unlikely, because as I said, 11,000 posts. There is always the option to make your own guide if you're inclined. I'll leave this alone while I'm still civil. This is for your benefit and for my benefit.

Of course you can do what you want, I am not saying anything that will blow your mind.
post #10903 of 19540
Seems I might be on to something with my 4770k. I was having stability problems at 4.7ghz yesterday under stress tests. Played with voltages some more today and successfully ran Intel xtu benchmark and 10 passes of x264 with 1.25 vid and vcin at 1.915v. I'm shooting for 50 passes of x264 right now and so far so good.
post #10904 of 19540
Also found that uncore is where my true instability comes from when. So for now I'm just focusing on CPU speed and will play with uncore after all is said and done. Is there a correlation between high CPU multis and uncore? Meaning that the higher the multi the lower the uncore that is needed. If that makes any sense.
post #10905 of 19540
Quote:
Originally Posted by GeneO View Post

Yes I am saying that and I did provide evidence - I can get a stable overclock against 1.184V fixed with non-AVX tests, but with p95 v28. it is not stable. However it is with AVX adjusted ot 1.2V non-avx load/1.216V AVX. This might not apply to everyone and I stated that in my earlier post.

I'm little lost so just trying to understand this (I bolded the key word in your post).

I'm reading the above as you're using Manual (or fixed whatever it's called on other boards) - not adaptive or offset. Right? Then you say that X programs run fine @ 1.184v (say x264 Cinebench, etc) but prime falls over. To get prime working you needed to increase your volts to 1.216v or so?

If all of the above is true then where's the argument? Prime just needed/wanted ~0.03v more than any other program. Overall, there is no crazy ~0.08-0.1v voltage jump in manual/fixed regardless what program you run.
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
post #10906 of 19540
Quote:
Myself, I am not throttled at 1.25V. In your example if you are unstable at linpack at 1.25V then raise it until you are stable and you won't be at 1.15V. You will only hit those high volts when running something like linpack. But I said this wasn't for everybody. In my case the increase is not 0.1V, but .016-.02V. This boost in voltage obviously varies.

If you can run linpack at 1.25v+ and get gflops way over 200, i'd like to see a screenshot - my high end air throttles too biggrin.gif

The ~0.02 increase is normal, the 0.1 is usually from adaptive voltage. AFAIK everybody gets such increase with adaptive
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 #10907 of 19540
Quote:
Originally Posted by Cyro999 View Post

If you can run linpack at 1.25v+ and get gflops way over 200, i'd like to see a screenshot - my high end air throttles too biggrin.gif

The ~0.02 increase is normal, the 0.1 is usually from adaptive voltage. AFAIK everybody gets such increase with adaptive

If you'd read my posts and look at the screenshots you would see that I do not get 0.1V I only get 0.016-0.032 with adaptive.

I am not talking about the normal rise ogf the core above VID, I am talking about the delta rise when using AVX. That is the point, not everyone gets such a rise, so don't take it for granted.

-
Edited by GeneO - 3/9/14 at 4:46pm
post #10908 of 19540
Quote:
Originally Posted by GeneO View Post

If you'd read my posts and look at the screenshots you would see that I do not get 0.1V I only get 0.016-0.032 with adaptive.

I am not talking about the normal rise ogf the core above VID, I am talking about the delta rise when using AVX. That is the point, not everyone gets such a rise, so don't take it for granted.

-

it scales with voltage. lower vcore lower adaptive voltage boost. On my board at 1.38volts it will run up to 1.49v I remember because it scared the crap outta me.
Cable box
(24 items)
 
Phat Bits
(16 items)
 
Lament Box
(16 items)
 
CPUMotherboardGraphicsRAM
i7-4790K 4.8ghz @ 1.245v Asus Maximus Hero VI 4GB Asus Strix GTX 980 Gskil Trident 2400 CL10 
Hard DriveHard DriveHard DriveOptical Drive
128GB sandisk extreme SSD 2 tb Western digital media intel 530 128 gb SSD External LG BD RW 
CoolingCoolingCoolingCooling
EK Coolstream 360mm PE Rad  Gentle typhoon AP-15 X 3 Fractal Design 140mm 1000RPM X3 Fractal Design Venturi High Flow Series 140mm F... 
CoolingCoolingCoolingCooling
Bitspower DDC Heatsink Black EK Coolstream XTC 140mm Rad EK Supremecy MX  EK FC980 Strix Water Block 
CoolingCoolingOSKeyboard
swiftech MCP-35X EK XRES 140 DDC Top WIN 7 ULTIMATE smk link 
PowerCaseMouseOther
Seasonic G-750  Fractal Design R5 G700s HDHOMERUN 3 TUNER CABLE CARD 
CPUMotherboardGraphicsRAM
4770k 4.5ghz @1.29v asus z87-A EVGA 4gb GTX 970 FTW 1510mhz 8000mem 1.23v Adata XPG 8GB @ 2400mhz 11-13-13-35-1 
Hard DriveHard DriveCoolingCooling
OCZ VERTEX 3 90GB 1tb 7200 Seagate hybrid  D5 vario Primo chill 120mm D5 Compression Res 
CoolingCoolingCoolingCooling
Swiftech MCR-320 XP 360mm rad Swiftech MCR-320 XP 120mm rad EK Supreme LTX Cpu block Biitspower ACX GTX 970 Water block 
OSKeyboardPowerCase
win7 64bit ultimate G13 Seasonic M211 620 watt Nzxt Guardian 
  hide details  
Reply
Cable box
(24 items)
 
Phat Bits
(16 items)
 
Lament Box
(16 items)
 
CPUMotherboardGraphicsRAM
i7-4790K 4.8ghz @ 1.245v Asus Maximus Hero VI 4GB Asus Strix GTX 980 Gskil Trident 2400 CL10 
Hard DriveHard DriveHard DriveOptical Drive
128GB sandisk extreme SSD 2 tb Western digital media intel 530 128 gb SSD External LG BD RW 
CoolingCoolingCoolingCooling
EK Coolstream 360mm PE Rad  Gentle typhoon AP-15 X 3 Fractal Design 140mm 1000RPM X3 Fractal Design Venturi High Flow Series 140mm F... 
CoolingCoolingCoolingCooling
Bitspower DDC Heatsink Black EK Coolstream XTC 140mm Rad EK Supremecy MX  EK FC980 Strix Water Block 
CoolingCoolingOSKeyboard
swiftech MCP-35X EK XRES 140 DDC Top WIN 7 ULTIMATE smk link 
PowerCaseMouseOther
Seasonic G-750  Fractal Design R5 G700s HDHOMERUN 3 TUNER CABLE CARD 
CPUMotherboardGraphicsRAM
4770k 4.5ghz @1.29v asus z87-A EVGA 4gb GTX 970 FTW 1510mhz 8000mem 1.23v Adata XPG 8GB @ 2400mhz 11-13-13-35-1 
Hard DriveHard DriveCoolingCooling
OCZ VERTEX 3 90GB 1tb 7200 Seagate hybrid  D5 vario Primo chill 120mm D5 Compression Res 
CoolingCoolingCoolingCooling
Swiftech MCR-320 XP 360mm rad Swiftech MCR-320 XP 120mm rad EK Supreme LTX Cpu block Biitspower ACX GTX 970 Water block 
OSKeyboardPowerCase
win7 64bit ultimate G13 Seasonic M211 620 watt Nzxt Guardian 
  hide details  
Reply
post #10909 of 19540
Quote:
Originally Posted by GeneO View Post

If you'd read my posts and look at the screenshots you would see that I do not get 0.1V I only get 0.016-0.032 with adaptive.

BIOS set to x39 @ 1.2v and Adaptive.

First picture is using Cinebench, it's asking for 1.184v which is little weird seeing as that's less than what I've put in BIOS.

Cinebench 15:


Second picture is using Prime 28.5, it's asking for 1.264v which is exactly that expected ~0.08v (1.184v + 0.08v) increase when using Adaptive.

Prime 28.5:


Nothing new to see here.
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
post #10910 of 19540
Quote:
Originally Posted by Wirerat View Post

it scales with voltage. lower vcore lower adaptive voltage boost. On my board at 1.38volts it will run up to 1.49v I remember because it scared the crap outta me.

Thanks for that. Now that makes a little sense. I could not test mine at higher voltages because I lost the lottery and have air cooling. Only thing I could try momentarily was to raise the power to 130W by running P95 v28 small fft, and it didn't scale with power, and 4.2 vs 4.3 GHz, which I didn't see a difference.

It doesn't scale linearly though, in that case.
-
Edited by GeneO - 3/9/14 at 7:08pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Haswell Overclocking Guide [With Statistics]