Overclock.net › Forums › Intel › Intel CPUs › The Intel Devil's Canyon Owners Club
New Posts  All Forums:Forum Nav:

The Intel Devil's Canyon Owners Club - Page 2087

post #20861 of 22515
I dunno about you guys, but on my chip adding a bit of VRing helps stability overall, not only at high uncore frequencies. I think lowering it way down (along with the uncore clock, like 35x) to like 1.05V also helps, but lacks in performance (35x to 43x is quite jump, should yield similar gains to a 100MHz core overclock or thereabouts).

E.g. On my CPU, I can get 4.8GHz "stable" at 1.32V VID (1.344V Vcore according to HWiNFO) with my uncore at 43x and VRing at 1.2V (Uses about 1.21-1.23V under load according to HWiNFO). Go any lower on the VRing, and stability is much harder to attain, even at 1.335V core VID, requiring even more VCCIN than the 1.9V I have now, more VCCSA, etc. Adding a bit of VRing simplifies things on my chip and allows me to push it somewhat safely, as I start getting a bit scared above 1.3V for 24/7 use. I shouldn't, probably, as people are using way more 24/7 with no problems within the usable lifespan of the chip, but whatever. I expect it to last a few more years anyway. Even if I upgrade, it's going to go in a secondary rig I'll build around it, so I want it alive, preferably.

PS : I have brought this up before, but here it goes again. I dunno how accurate my HWiNFO readings are regarding voltages. Don't get me wrong, it's not HWiNFO's fault as all monitoring apps show more or less the same values. But take VCore for example. It only goes in steps. 1.296V to 1.312V to 1.328V to 1.344V, 1.360V, 1.376V and the list goes on. Now, setting a VID of, let's say, 1.305V will give me 1.328V Vcore on all cores consistently under load. Upping VID to 1.31V will give mostly 1.328V with cores 3 and 4 sometimes using 1.344V. Then upping the VID to 1.315V leads to all cores using mostly 1.344V while cores 1 and 2 will sometimes drop to 1.328V. Going to 1.32V VID will cause all cores to use 1.344V all the time. The same continues upwards. 1.325V VID leads to only cores 3 and 4 using 1.36V sometimes etc etc etc. You get the idea. My guess is it's a sensor reporting thing, but it could be right and that's how it is. "Get a multimeter" you'll say. Tbh, I can't be THAT bothered. I mean, I wanna know what's going on, but not to the point where I'll go full hardware on it.

If anyone has any insight, I'd aprreciate it. smile.gif
post #20862 of 22515
I have noticed similar on my 4790k with hwinfo. Seems to always jump by similar numbers. My motherboard boosts up quite a bit during load though I am set to 1.232v (4.8ghz) which is what the VID shows but under load it jumps to 1.272 vcore. If I raise the voltage though it just goes up additionally. I thought maybe I could find a median but it does it no matter what the LLC/voltage is set to. Not much control over this but I did notice the numbers it jumps to are always the same like you are saying. Curious how it works as well. I am guessing the regulator only jumps certain levels of volts and is probably motherboard dependent. Not sure though.
Edited by dmfree88 - 6/22/16 at 7:50pm
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
post #20863 of 22515
Quote:
Originally Posted by dmfree88 View Post

I have noticed similar on my 4790k with hwinfo. Seems to always jump by similar numbers. My motherboard boosts up quite a bit during load though I am set to 1.232v (4.8ghz) which is what the VID shows but under load it jumps to 1.272. If I raise the voltage though it just goes up additionally. I thought maybe I could find a median but it does it no matter what the LLC is set to. Not much control over this but I did notice the numbers it jumps to are always the same like you are saying. Curious how it works as well. I am guessing the regulator only jumps certain levels of volts and is probably motherboard dependent. Not sure though.

Yeah its mobo related.
It also depends on Just how heavy the load is. For exemple 100% load in prime Will give you a bigger offset of extra voltage bump than 100% load in Aida 64 probably. It shows 100% load for both When stresstest with them but eventual reached max vcore Will be different.

To me.. I think it can be really usfull knowing to around what actual vcore under a specific stresstest load of ur choice Will be reached. But using to many different kinds of programs to test might get confusing lol.
Edited by QuacK - 6/22/16 at 7:02pm
post #20864 of 22515
Quote:
Originally Posted by JackCY View Post

if the CPU is stable at this voltage with low ratio
Thank you. That's one part of the procedure I'd forgotten.
Quote:
Don't bother with RAM as testing RAM takes weeks, brings no benefit at all and it's a major PITA. Plus you need some magical RAM that is high clocks low latency but was sold with low clocks and high latency, good luck finding a company that bins the RAM that way, most of the time they just bin the RAM and sell you the high speed low latency RAM already OCed with proper XMP profiles.

If you are bored or like lower temps at the end you can lower Vsa, Vioa, Viod and test again, lower one by one, test each time, tons of time to do proper.
Vccin to Vccin = Vmax (max of any voltage you used, core, ring, sa, ioa, iod) + 0.3V medium LLC. On DC Vccin works fine until Vmax + 0.2V but may depend on mobo or CPU. I don't think it runs that well on older Haswell.
RAM I have already put the time in to OC. Got 1600 9-9-9 Hynix to 2800 11-14-14 and tight tertiaries. Super proud of that, honestly. At this stage I'm more searching for a magic VCCIN + LLC combo that the core likes.
post #20865 of 22515
whats interesting is these same numbers are sticking in my mind from my AMD cpu as well. Not sure if it is a similar regulation method but I seem to remember the same numbers popping up. I generally use those numbers as well when I set voltage. 1.232 is what the regulator jumps up to during load if I am set around 1.195-1.2 (give or take) in bios so I ended up at 1.232 for stability when I crashed at 1.221 (which jumped up to 1.272 and bounced between that and 1.242 IIRC, Now is stable 1.272 during load)

During gaming today (played crysis 3 and project cars) it peaked at 1.264v so as you mentioned quack it really seems to be dependent on the amount of load. All my stabilization testing is just done with intel burn test (which is what I was using when I said "during load" above). I run normal while I am playing around and until I am happy with stability then do one run on high after I have settled in (which is 10x runs technically) then I start gaming and if it works it works. Don't really want to spend too much time stress testing if I can avoid it. I do consider any appearance of the loading symbol or even minimal freezing of the flame to be failure.
Edited by dmfree88 - 6/22/16 at 8:13pm
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
post #20866 of 22515
It might be superstition, but I believe things to be more stable when the unloaded VIDs are as close to one another as possible. I'll load up command center and move the vcore in .001 increments to see what single core voltage setting gets them most closely clumped. It's not always a clean .01v step.
post #20867 of 22515
Quote:
Originally Posted by dmfree88 View Post

whats interesting is these same numbers are sticking in my mind from my AMD cpu as well. Not sure if it is a similar regulation method but I seem to remember the same numbers popping up. I generally use those numbers as well when I set voltage. 1.232 is what the regulator jumps up to during load if I am set around 1.195-1.2 (give or take) in bios so I ended up at 1.232 for stability when I crashed at 1.221 (which jumped up to 1.272 and bounced between that and 1.242 IIRC, Now is stable 1.272 during load)

During gaming today (played crysis 3 and project cars) it peaked at 1.264v so as you mentioned quack it really seems to be dependent on the amount of load. All my stabilization testing is just done with intel burn test (which is what I was using when I said "during load" above). I run normal while I am playing around and until I am happy with stability then do one run on high after I have settled in (which is 10x runs technically) then I start gaming and if it works it works. Don't really want to spend too much time stress testing if I can avoid it. I do consider any appearance of the loading symbol or even minimal freezing of the flame to be failure.

If you are testing now running normal, i would just stick to normal and just Let raise the amount of runs to pass.


High Will use more ram and put more load on ur system, so where your settings are stable at "normal" it might crash When switching to high.
post #20868 of 22515
I usually load up custom with nearly all of ram used and run for 5 minutes after all is said and done if I really want to check stability (which I did do for a moment the other day). If it is not stable on high and normal it is likely not actually stable at all. Temps are definitely higher and the runs themselves take much longer. On normal it takes about 7 seconds per run which barely has time for the CPU to warm up which is why I switch to high afterwards (normal for testing ramping up and down ability, high for temperature stressing). I do extreme/custom as this helps verify your ram is stable, the run takes way too long though I don't wait for it to finish I just load it up (which it takes a few moments to load up the ram) and wait a couple minutes to verify it works. It is much rougher on the system but it still should be stable on any setting I would think. If yours is only stable on normal but not on high you might have a ram problem or are just on the fine line of stability which doesn't mean it will crash during regular use and may be stable enough but isn't necessarily 100% stable.
Edited by dmfree88 - 6/22/16 at 9:04pm
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
post #20869 of 22515
Quote:
Originally Posted by dmfree88 View Post

I usually load up custom with nearly all of ram used and run for 5 minutes after all is said and done if I really want to check stability (which I did do for a moment the other day). If it is not stable on high and normal it is likely not actually stable at all. Temps are definitely higher and the runs themselves take much longer. On normal it takes about 7 seconds per run which barely has time for the CPU to warm up which is why I switch to high afterwards (normal for testing ramping up and down ability, high for temperature stressing). I do extreme/custom as this helps verify your ram is stable, the run takes way too long though I don't wait for it to finish I just load it up and wait a couple minutes to verify it works. It is much rougher on the system but it still should be stable on any setting I would think. If yours is only stable on normal but not on high you might have a ram problem or are just on the fine line of stability which doesn't mean it will crash during regular use and may be stable enough but isn't necessarily 100% stable.
Quote:
Originally Posted by dmfree88 View Post

whats interesting is these same numbers are sticking in my mind from my AMD cpu as well. Not sure if it is a similar regulation method but I seem to remember the same numbers popping up. I generally use those numbers as well when I set voltage. 1.232 is what the regulator jumps up to during load if I am set around 1.195-1.2 (give or take) in bios so I ended up at 1.232 for stability when I crashed at 1.221 (which jumped up to 1.272 and bounced between that and 1.242 IIRC, Now is stable 1.272 during load)

During gaming today (played crysis 3 and project cars) it peaked at 1.264v so as you mentioned quack it really seems to be dependent on the amount of load. All my stabilization testing is just done with intel burn test (which is what I was using when I said "during load" above). I run normal while I am playing around and until I am happy with stability then do one run on high after I have settled in (which is 10x runs technically) then I start gaming and if it works it works. Don't really want to spend too much time stress testing if I can avoid it. I do consider any appearance of the loading symbol or even minimal freezing of the flame to be failure.

If you are testing now running normal, i would just stick to normal and just Let raise the amount of runs to pass.


High Will use more ram and put more load on ur system, so where your settings are stable at "normal" it might crash When switching to high.
Quote:
Originally Posted by dmfree88 View Post

I usually load up custom with nearly all of ram used and run for 5 minutes after all is said and done if I really want to check stability (which I did do for a moment the other day). If it is not stable on high and normal it is likely not actually stable at all. Temps are definitely higher and the runs themselves take much longer. On normal it takes about 7 seconds per run which barely has time for the CPU to warm up which is why I switch to high afterwards (normal for testing ramping up and down ability, high for temperature stressing). I do extreme/custom as this helps verify your ram is stable, the run takes way too long though I don't wait for it to finish I just load it up and wait a couple minutes to verify it works. It is much rougher on the system but it still should be stable on any setting I would think. If yours is only stable on normal but not on high you might have a ram problem or are just on the fine line of stability which doesn't mean it will crash during regular use and may be stable enough but isn't necessarily 100% stable.

I dont use ibt at all. Ive used it in the past though.

My point was.. when switching to a higher load, you might crash while on normal everything is fine.

Since you mentioned you dont want to spend all that much time stresstesting.

Just loading it up for a minute and closing it before a run is passed isnt really effective I think, but whatever works for you I say go for it wink.gif
post #20870 of 22515
It would probably take 10+ minutes to do a 14+gb custom single run. Not sure but the purpose is not to test if it passes the test as the test has already been passed 20+ times from running normal then high 10 or more times each. The purpose is to assure it doesn't BSOD or freeze up when the ram is heavily loaded and the test is running.

My point is if you crash on high but not normal you are not actually stable. Loading up more ram should not make you unstable. If it does you are probably not actually fully stable and it most likely has to do with your ram or the ram controller because that is where the additional load is mostly added when increasing difficulty of IBT. Although the Vcore may still need more volts as well but I see no reason you shouldn't pass on any setting if it is really stable.
Edited by dmfree88 - 6/22/16 at 9:33pm
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
Intel Overclocker
(12 items)
 
Media PC
(8 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 4790k Delidded + Lapped ASRock Z97 Extreme4 MSI Twin Frozr GTX 660  Corsair Vengeance Pro 2400mhz DDR3 
Hard DriveHard DriveOptical DriveCooling
OCZ Deneva 2 R series 400GB SSD WD Green 4TB HDD Asus DRW-24B1ST Phanteks PH-TC14PE_RD 
OSPowerCaseMouse
Windows 10 Seasonic 650w M12II Phanteks Enthoo Pro M with Acrylic side Cobra e-blue 
CPUMotherboardRAMHard Drive
intel core 2 quad q8200 Dell studio stock Dell studio Stock Corsair Force 3 SSD 
Hard DriveOptical DriveCoolingOS
External powered Seagate expansion desk Dell studio stock Stock Windows 10 
CPUMotherboardGraphicsRAM
FX-8350 GA-990FXA-UD5 MSI 7870 Hawk Edition Gskill 1600 2x4 GB 
Hard DriveOptical DriveCoolingOS
Corsair Force Series 3 22x Samsung w/ lightscribe dvd/cd burner Noctua NH-D14 Win 7 pro x64 
MonitorPowerCase
Acer x173w Seasonic M12II 650watt Semi-Modular PSU Apevia x-dreamer4 (with side window and red) 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › The Intel Devil's Canyon Owners Club