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 2093

post #20921 of 22515
Quote:
Originally Posted by tolis626 View Post

Ok, first off, AIDA64's stress test is kind of garbage. My CPU can pass it at some ridiculous settings without crashing, but it will crash in most other things. So avoid it. Use x264 or x265 from the forums here, RealBench, OCCT (Although many suggest staying away from OCCT too, like Prime95) or similar. You want realistic loads as much as possible. Also, use manual voltage when testing, not adaptive.

Now, your settings seem fine. More than likely is that you need a bit more VCore. But since you're running DDR3 at 2400MHz (I assume at 1.65V), a bit more VCCSA may help, or it may not. Trying won't hurt and 0.92V isn't high to begin with, so you can go 0.95-1V and see what happens. You may also need a bit more VCCIN or way less, that's always worth trying. Your cache multi is a bit low, but leave it at that until you can stabilize everything else, then increase if you want.

My testing methodology goes like this. Dial in the settings then do a "quick" 10 loop test of x265 Stress Test. If it passes, drop the voltage a bit (I'm talking about VCore, but this applies to almost everything. I either decrease/increase clocks or decrease/increase voltages as I deem necessary) and repeat until I reach instability. Then go back up a step and try to run a 4 hour stress test in Realbench. If it passes, it's stable. If it doesn't, I go one more step up, but usually there's no need. My stability testing hasn't failed me since I started doing it like this, so give it a go. It does take while, though. smile.gif

Correction, adaptive is what you want to use. Its what keeps your voltage low at idle and where it needs to be at load. Its what ive been using for 4.5 and 4.8 on all cores at 1.2 and 1.3v respectively and it drops the voltages down at idle and to 1.2 at 4.5. It should be manual and adaptive. Theres also override which keeps the voltage at a set voltage and thats it. It also depends on the mobo as well.
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
post #20922 of 22515
Quote:
Originally Posted by bluej511 View Post

Correction, adaptive is what you want to use. Its what keeps your voltage low at idle and where it needs to be at load. Its what ive been using for 4.5 and 4.8 on all cores at 1.2 and 1.3v respectively and it drops the voltages down at idle and to 1.2 at 4.5. It should be manual and adaptive. Theres also override which keeps the voltage at a set voltage and thats it. It also depends on the mobo as well.

Hold on. I said use manual/override (same thing, really) when stress testing. I also revert to adaptive once I'm done testing. Although most mobos will drop VCore if C-States are enabled regardless of VID, but that's another story. smile.gif

Stress testing with adaptive voltage leads to the CPU sometimes using more voltage than needed, mostly during AVX loads. But regardless, when you want to test X frequency at Y voltage, you don't want either to fluctuate too much, so manual is the way to go here. For everyday use, I see no reason in using manual over adaptive, but some will disagree with me. To each their own I guess.
post #20923 of 22515
Quote:
Originally Posted by tolis626 View Post

Hold on. I said use manual/override (same thing, really) when stress testing. I also revert to adaptive once I'm done testing. Although most mobos will drop VCore if C-States are enabled regardless of VID, but that's another story. smile.gif

Stress testing with adaptive voltage leads to the CPU sometimes using more voltage than needed, mostly during AVX loads. But regardless, when you want to test X frequency at Y voltage, you don't want either to fluctuate too much, so manual is the way to go here. For everyday use, I see no reason in using manual over adaptive, but some will disagree with me. To each their own I guess.

I only stress test with intel tuning utility and its been the most realistic ever. When gaming my peak temps reach exactly what the stress test reaches. And since its a gaming pc can't get more accurate then that haha.

Ive checked with a DMM and my voltage stays a perfect 1.200 at 100% load when stress testing deviates to 1.198 when it drops to like 98% so ive never seen it overvolt.
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
post #20924 of 22515
Quote:
Originally Posted by bluej511 View Post

I only stress test with intel tuning utility and its been the most realistic ever. When gaming my peak temps reach exactly what the stress test reaches. And since its a gaming pc can't get more accurate then that haha.

Ive checked with a DMM and my voltage stays a perfect 1.200 at 100% load when stress testing deviates to 1.198 when it drops to like 98% so ive never seen it overvolt.

Intel XTU is quite garbage for stress testing too, sorry. tongue.gif

The most realistic load, hands down, is x264, followed by Realbench (Ok, it's a bit of an extreme form of real world usage, but it's not synthetic). The reason your temps are lower when gaming or during general use compared to when stress testing is, among others, that the load is oscillating. For proper testing, you need 100% load 100% of the time. XTU, like AIDA64, is just not stressful enough and that's the reason it doesn't let temps increase. On my rig, Battlefield 4 reaches higher temps than XTU, so you get the idea. RealBench, for comparison, reaches 5-10C over BF4's temps.

Now, if XTU works for you, great, don't get me wrong. But objectively speaking, it's inadequate. It's on the other end of the spectrum compared to Prime95, which is over-taxing the CPU for no reason other than "hurr-durr it's stressful Arhh" (At least when not used properly). The other tools fall somewhere in the middle and they are the sweet spot. They make sure your system is stable without aggressively trying to fry your CPU in the process. smile.gif

PS : I've used both AIDA64 and XTU extensively in the past, so I'm talking from my own experience here, which just agrees with most of the people here. I'm not regurgitating what I've read around the forums. smile.gif

EDIT : To see the processor overvolt on its own with adaptive voltage, run either Prime95/IBT or OCCT's AVX enabled Linpack and watch that suckah burn. biggrin.gif
post #20925 of 22515
Quote:
Originally Posted by tolis626 View Post

Intel XTU is quite garbage for stress testing too, sorry. tongue.gif

The most realistic load, hands down, is x264, followed by Realbench (Ok, it's a bit of an extreme form of real world usage, but it's not synthetic). The reason your temps are lower when gaming or during general use compared to when stress testing is, among others, that the load is oscillating. For proper testing, you need 100% load 100% of the time. XTU, like AIDA64, is just not stressful enough and that's the reason it doesn't let temps increase. On my rig, Battlefield 4 reaches higher temps than XTU, so you get the idea. RealBench, for comparison, reaches 5-10C over BF4's temps.

Now, if XTU works for you, great, don't get me wrong. But objectively speaking, it's inadequate. It's on the other end of the spectrum compared to Prime95, which is over-taxing the CPU for no reason other than "hurr-durr it's stressful Arhh" (At least when not used properly). The other tools fall somewhere in the middle and they are the sweet spot. They make sure your system is stable without aggressively trying to fry your CPU in the process. smile.gif

PS : I've used both AIDA64 and XTU extensively in the past, so I'm talking from my own experience here, which just agrees with most of the people here. I'm not regurgitating what I've read around the forums. smile.gif

EDIT : To see the processor overvolt on its own with adaptive voltage, run either Prime95/IBT or OCCT's AVX enabled Linpack and watch that suckah burn. biggrin.gif

Well im also on water and bare die with CLU so theyll always be low haha.

And the games i play peg the cpu at 100% sometimes on all 4 cores, ie unity syndicate, rainbow six siege. Prime95 is totally outdated and dont know why people still use it.

And for me XTU pegs it at 100% all the time and thats confirmed with the DMM, task manager, and xtu.

Since this is a gaming PC i pretty much just stress test while gaming anyways, im not mining or computing with it so any other test is pointless. I don't even stress test more then an hr. Its a total waste and destroys CPUs.
Edited by bluej511 - 7/5/16 at 9:40am
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
R7 1700X asus crosshair 6 Sapphire R9 390 Nitro CMK16GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Kingston HyperX 3K Seagate ST2000DM006 EKWB Supremacy Evo EK-XRES 100 Revo D5 - Acetal 
CoolingCoolingCoolingCooling
Alphacool GPX r9 390 m01 alphacool xt45 240 Alphacool xt45 360 Alphacool D5 PWM 
OSMonitorKeyboardPower
Windows 10 LG 29UM68 Corsair Strafe Corsair RM1000 
CaseMouseMouse Pad
Thermaltake X5 Corsair M65 Pro RGB Razer Goliathus Speed 
  hide details  
Reply
post #20926 of 22515
Quote:
Originally Posted by tolis626 View Post

Ok, first off, AIDA64's stress test is kind of garbage. My CPU can pass it at some ridiculous settings without crashing, but it will crash in most other things. So avoid it. Use x264 or x265 from the forums here, RealBench, OCCT (Although many suggest staying away from OCCT too, like Prime95) or similar. You want realistic loads as much as possible. Also, use manual voltage when testing, not adaptive.

Now, your settings seem fine. More than likely is that you need a bit more VCore. But since you're running DDR3 at 2400MHz (I assume at 1.65V), a bit more VCCSA may help, or it may not. Trying won't hurt and 0.92V isn't high to begin with, so you can go 0.95-1V and see what happens. You may also need a bit more VCCIN or way less, that's always worth trying. Your cache multi is a bit low, but leave it at that until you can stabilize everything else, then increase if you want.

My testing methodology goes like this. Dial in the settings then do a "quick" 10 loop test of x265 Stress Test. If it passes, drop the voltage a bit (I'm talking about VCore, but this applies to almost everything. I either decrease/increase clocks or decrease/increase voltages as I deem necessary) and repeat until I reach instability. Then go back up a step and try to run a 4 hour stress test in Realbench. If it passes, it's stable. If it doesn't, I go one more step up, but usually there's no need. My stability testing hasn't failed me since I started doing it like this, so give it a go. It does take while, though. smile.gif

So I tried the following today. Setting my bios to default then edit this in the bios

Vcore: Manual 1.25V
Core: 45x
Memory: XMP
chache ratio: 40x
Cache voltage 1.15V
rest of the settings on auto

It tested fine in asus realbench stress test but then I decided to try 46x and my pc didn't even boot up. I had iot freeze on me when trying to boot and also tried resetting my bios by presseing the clear cmos button on the board with no luck. Keep giving me the boot device LED. I tried and tried and tried to reset the cmos but nothing. I also plugged out pc pressing the clr cmos button and waiting untill all lights are gone and still no boot. I then unplugged all of my HDD's and SSD's still same issue, boot device LED RED. I fiddled with random stuff and couldn't get it to boot and magically it booted.

I had the same issue before when I was overclocking my system and for some reason just before windows loads I think where it loads the storage devices it takes a while for it to go past that. Now I'm on default so I don't know if I have a faulty board or maybe even a faulty CPU or whatever. I checked my Storage devices for errors and nothing. Samsung magician also shows nothing wrong so I don't think it's my Storage devices failing. I would really like to overclock my system cause when I had it overclocked last time I could see a speed boost in GTA V going from 4.4GHz to 4.6GHz
Main Gaming RIG
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790k ASUS VII Ranger Galax hof gtx 980 ti 16GB 1600mhz adata ram 
Hard DriveHard DriveHard DriveCooling
128GB SSD Samsung 840 evo 1TB WD blue Liteon 256GB SSD Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 ASUS PG279Q Cooler master Quickfire XT Corsair AX850w 
CaseMouseMouse PadAudio
Corsair Air carbide 540 Steelseries Ikari Laser Steelsereies QCK+ sound Blaster Z 
Audio
Sennheiser HD 598 
  hide details  
Reply
Main Gaming RIG
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790k ASUS VII Ranger Galax hof gtx 980 ti 16GB 1600mhz adata ram 
Hard DriveHard DriveHard DriveCooling
128GB SSD Samsung 840 evo 1TB WD blue Liteon 256GB SSD Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 ASUS PG279Q Cooler master Quickfire XT Corsair AX850w 
CaseMouseMouse PadAudio
Corsair Air carbide 540 Steelseries Ikari Laser Steelsereies QCK+ sound Blaster Z 
Audio
Sennheiser HD 598 
  hide details  
Reply
post #20927 of 22515
Quote:
Originally Posted by DiceAir View Post

So I tried the following today. Setting my bios to default then edit this in the bios

Vcore: Manual 1.25V
Core: 45x
Memory: XMP
chache ratio: 40x
Cache voltage 1.15V
rest of the settings on auto

It tested fine in asus realbench stress test but then I decided to try 46x and my pc didn't even boot up. I had iot freeze on me when trying to boot and also tried resetting my bios by presseing the clear cmos button on the board with no luck. Keep giving me the boot device LED. I tried and tried and tried to reset the cmos but nothing. I also plugged out pc pressing the clr cmos button and waiting untill all lights are gone and still no boot. I then unplugged all of my HDD's and SSD's still same issue, boot device LED RED. I fiddled with random stuff and couldn't get it to boot and magically it booted.

I had the same issue before when I was overclocking my system and for some reason just before windows loads I think where it loads the storage devices it takes a while for it to go past that. Now I'm on default so I don't know if I have a faulty board or maybe even a faulty CPU or whatever. I checked my Storage devices for errors and nothing. Samsung magician also shows nothing wrong so I don't think it's my Storage devices failing. I would really like to overclock my system cause when I had it overclocked last time I could see a speed boost in GTA V going from 4.4GHz to 4.6GHz

Strange that you mention it, because I've been having similar issues with my Maximus VII Formula. Like one in every few boot-ups will result in a delayed boot-up time. It usually either does what you described with the red boot LED, or, more frequently, it will go past the POST screen and get stuck where it should start loading Windows and show q-core b1 on the mobo screen. It's a strange thing and it's at least frustrating when it happens. Thing is, if I just leave it do its thing, it will boot up fine after a while. At least most of the time.

For the longest time I thought this had something to do with memory retraining, but I highly doubt it at this point. I'm at a loss. But once it boots up, it's fine, so no worries I guess.

Anyway, that your PC doesn't even boot up at 4.6GHz at 1.25V is strange. But I'd suggest trying to run your memory at 1600MHz 9-9-9-27 and not XMP until you figure it out. Or even 1333MHz. It won't be permanent, it will just make your life easier. At 2400MHz memory and 4.6GHz core it may be your IMC that craps out and not the core itself. If it works with the memory downclocked, it probably just needs a bump in VCCSA or I/O voltages. Probably VCCSA. So try one thing at a time and you'll figure it out. Anything else you need, ask away. smile.gif
post #20928 of 22515
Quote:
Originally Posted by deegzor View Post

Welcome and nice job! I have 4770k with the same mob. Just a friendly notice keep an eye for your voltages for a day or so with some monitoring program (i recommend hwinfo). Atleast for me the voltages would randomly spike over my set limit during gaming or encoding. They wen't from 1.35v to 1.48v and that's waaaay too much even with my delidded water setup.

Cheers and welcome thumb.gif

I actually did (thanks for the heads up) I found out I was voltage dropping on heavy loads to 1.185v which was unstable doing x264 bench, so I upped the voltage 0.020v and now it drops to 1.201 and when it volts high it does 1.220v which is rock solid stable. Temps don't go above 72ºC with air cooling... so a win win!
post #20929 of 22515
vSA and IO voltages have a huge impact on my startup speed. To the point where I can recognize unstable settings by how it boots.

.01/.01/.02 SA/IOA/IOD has the windows loading loading wheel spin about a half turn and instantly to desktop. Unstable though.

.09/.01/.09 adds a minor delay between my secondary monitor coming up and has about a 2 revolution spin before loading but is x264 stable. IOA anywhere above +.01v requires more vCore and seems generally destabilizing.

I would say play with your IO volts a little and see if it doesn't give you more consistent startups. Your setup might be like mine where the fastest settings to desktop aren't necessarily the most stable.
post #20930 of 22515
Quote:
Originally Posted by tolis626 View Post

Strange that you mention it, because I've been having similar issues with my Maximus VII Formula. Like one in every few boot-ups will result in a delayed boot-up time. It usually either does what you described with the red boot LED, or, more frequently, it will go past the POST screen and get stuck where it should start loading Windows and show q-core b1 on the mobo screen. It's a strange thing and it's at least frustrating when it happens. Thing is, if I just leave it do its thing, it will boot up fine after a while. At least most of the time.

For the longest time I thought this had something to do with memory retraining, but I highly doubt it at this point. I'm at a loss. But once it boots up, it's fine, so no worries I guess.

Anyway, that your PC doesn't even boot up at 4.6GHz at 1.25V is strange. But I'd suggest trying to run your memory at 1600MHz 9-9-9-27 and not XMP until you figure it out. Or even 1333MHz. It won't be permanent, it will just make your life easier. At 2400MHz memory and 4.6GHz core it may be your IMC that craps out and not the core itself. If it works with the memory downclocked, it probably just needs a bump in VCCSA or I/O voltages. Probably VCCSA. So try one thing at a time and you'll figure it out. Anything else you need, ask away. smile.gif

Dude you a legend. Running my memory at stock actually made my cpu able to run Asus realbench for 10min at 4.6ghz 1.25V. I will try to go lower later or maybe even higher cpu core but aiming for 4.6GHz for now.

Actually to be honest I forget what happened to me last year. I had a MSI z97 gaming 5 but my memory wasn't stable at 2400mhz so had to run something like 1600mhz - 1866mhz to get it stable so actually this board is helping me a little bit with the memory stability but anyway this made me think. ok memroy stock (what I should do from the start when overclocking anyway stupid me...lol) So now I did at least 10min of asus realbench before I decided to stop.

Even my bootup is so much faster now. The little dots doesn't go around for more than 1 turn and I'm in windows. I wish I bought 1866mhz instead of this 2400mhz. Nothing to do with the memory but think my IMC might be a little bit more stable at 1600mhz-1866mhz. Anyway I can always manually clock it to 1866mhz and test but for now I first want to work at getting my core almost fully stable.

So what do you suggest for voltage? Should I run my voltage on either manual + c-states giving me lower voltage on idle as measured by hwinfo under vcore not vid or should I try for adaptive rather? I kinda leaning towards manual because I like it that my pc won't exceed the maximum voltage as set in bios even when using avx instructions and I think it might give me more stability but I don't really know. Do you also think 4.4GHz 2400mhz ram will give me slower performance than 4.6GHz 1600mhz?
Main Gaming RIG
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790k ASUS VII Ranger Galax hof gtx 980 ti 16GB 1600mhz adata ram 
Hard DriveHard DriveHard DriveCooling
128GB SSD Samsung 840 evo 1TB WD blue Liteon 256GB SSD Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 ASUS PG279Q Cooler master Quickfire XT Corsair AX850w 
CaseMouseMouse PadAudio
Corsair Air carbide 540 Steelseries Ikari Laser Steelsereies QCK+ sound Blaster Z 
Audio
Sennheiser HD 598 
  hide details  
Reply
Main Gaming RIG
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-4790k ASUS VII Ranger Galax hof gtx 980 ti 16GB 1600mhz adata ram 
Hard DriveHard DriveHard DriveCooling
128GB SSD Samsung 840 evo 1TB WD blue Liteon 256GB SSD Phanteks PH-TC14PE 
OSMonitorKeyboardPower
Windows 8.1 ASUS PG279Q Cooler master Quickfire XT Corsair AX850w 
CaseMouseMouse PadAudio
Corsair Air carbide 540 Steelseries Ikari Laser Steelsereies QCK+ sound Blaster Z 
Audio
Sennheiser HD 598 
  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