Overclock.net › Forums › Cooling › Water Cooling › My Corsair H100 leaked on my new 680-GTX
New Posts  All Forums:Forum Nav:

My Corsair H100 leaked on my new 680-GTX - Page 17  

post #161 of 180
Quote:
Originally Posted by webdevii View Post

Also I am Over Clocked to 4.99 at 107' F running stable and smooth......

I'm a little concerned here -

you're saying 4.99ghz at 107 Fahrenheit ? That doesn't seem likely. What worries me is if it is 107 Celsius - that's a far more likely scenario, but I believe if your chip behaves anything like my 3820 (also a LGA2011 i7, but the quad core version), that is beyond the absolute maximum operating temp of the chip, and it's probably (a) throttling, and (b) working its way to complete failure if running at that temp long term.

My 3820 pushed to 4.87 on water full-on Prime95 stressed runs about 52 deg. C. What are you running for core volts on it?

That's an expensive chip you've got there, and especially after the chip seeing temps like that for any period of time, I would spend the $$ to buy the Intel Tuning Protection plan, where they will replace the chip once no questions asked - for reasons that would normally void the normal warranty - higher-than-spec volts, temp, etc.

Now normally I see the Tuning plan as a cash grab, but on a chip in that price range, it probably is well worth it. Plus it stays with the CPU, so it transfers if you ever sell the chip. I'm even going to buy it for my 3820 (it is a cheaper chip, but if I want to play with more volts, I want to have that card to play, as over-volting in some cases can leave some telltale physical evidence, contrary to what a lot of people say like "How the hell would intel know it was O/C vs just regular failure" -- now Intel has been pretty graceful according to some people, probably replacing chips that have the caps on the bottom of the package burnt open, overheated, etc -- but that may change and probably has as much to do with who is processing your RMA and what their day has been like.

But the bottom line is 107 degrees C is not a good place for the chip to be.
post #162 of 180
Thread Starter 
Quote:
Originally Posted by speck55 View Post

I'm a little concerned here -

you're saying 4.99ghz at 107 Fahrenheit ? That doesn't seem likely. What worries me is if it is 107 Celsius - that's a far more likely scenario, but I believe if your chip behaves anything like my 3820 (also a LGA2011 i7, but the quad core version), that is beyond the absolute maximum operating temp of the chip, and it's probably (a) throttling, and (b) working its way to complete failure if running at that temp long term.

My 3820 pushed to 4.87 on water full-on Prime95 stressed runs about 52 deg. C. What are you running for core volts on it?

That's an expensive chip you've got there, and especially after the chip seeing temps like that for any period of time, I would spend the $$ to buy the Intel Tuning Protection plan, where they will replace the chip once no questions asked - for reasons that would normally void the normal warranty - higher-than-spec volts, temp, etc.

Now normally I see the Tuning plan as a cash grab, but on a chip in that price range, it probably is well worth it. Plus it stays with the CPU, so it transfers if you ever sell the chip. I'm even going to buy it for my 3820 (it is a cheaper chip, but if I want to play with more volts, I want to have that card to play, as over-volting in some cases can leave some telltale physical evidence, contrary to what a lot of people say like "How the hell would intel know it was O/C vs just regular failure" -- now Intel has been pretty graceful according to some people, probably replacing chips that have the caps on the bottom of the package burnt open, overheated, etc -- but that may change and probably has as much to do with who is processing your RMA and what their day has been like.

But the bottom line is 107 degrees C is not a good place for the chip to be.

Ok, my brain computes in F, not C I set CPU-Z to F outputs, and I am running at 131'F. and that was after a while like 4 hours of runtime. Initially the start up temp was 107'F and slowly climbed to 131'F after about 3 hours and leveled out there the rest of the day.

For my Voltages I am running 1.512v
and the Multiplier is set at 101.000..

I can always back off the Multiplier and Voltage to 100.000 and 1.45v and drop the temp to around 100'F @ 4.5Ghz. Also on the Pump / Res I am using, are you supposed to fill them up to the top? correct. It is filled about 1/8" from the top of the Res It is the XSPC Res / Pump Combo on 1/2 x 3/4" Tubing Does that fill quantity sound about right? That was the only unsure part I had.

Thank You very much for your concerns on Chip Temps, I do appreciate it. As this is my first time doing the liquid cooled and OC'ing stuff.
Dev
Edited by webdevii - 5/13/13 at 2:24am
post #163 of 180
i would never go over 1.45v on sb-e under water. it doesnt tolerate voltage like sb does. 1.5+v is asking for quicker degradation.
w00t
(18 items)
 
   
CPUMotherboardGraphicsRAM
ph2 x20be asus m4a785-m sapphire vapor-x 4850  geil black dragon 1066 5-5-5-15-20 
Hard DriveOptical DriveCoolingOS
wd  asus dvd multi-write arctic cooling freezer 64 pro win 7 ultimate x64 
MonitorKeyboardPowerCase
samsung 42" lcd logitech wireless antec basic 500watt old dell crap 
MouseAudio
logitech wireless pioneer 7.1 
  hide details  
w00t
(18 items)
 
   
CPUMotherboardGraphicsRAM
ph2 x20be asus m4a785-m sapphire vapor-x 4850  geil black dragon 1066 5-5-5-15-20 
Hard DriveOptical DriveCoolingOS
wd  asus dvd multi-write arctic cooling freezer 64 pro win 7 ultimate x64 
MonitorKeyboardPowerCase
samsung 42" lcd logitech wireless antec basic 500watt old dell crap 
MouseAudio
logitech wireless pioneer 7.1 
  hide details  
post #164 of 180
Thread Starter 
Quote:
Originally Posted by pc-illiterate View Post

i would never go over 1.45v on sb-e under water. it doesnt tolerate voltage like sb does. 1.5+v is asking for quicker degradation.

Thank you,
I am going to drop the voltage back to 1.44
post #165 of 180
Quote:
Originally Posted by webdevii View Post

Ok, my brain computes in F, not C I set CPU-Z to F outputs, and I am running at 131'F. and that was after a while like 4 hours of runtime. Initially the start up temp was 107'F and slowly climbed to 131'F after about 3 hours and leveled out there the rest of the day.

For my Voltages I am running 1.512v
and the Multiplier is set at 101.000..

I can always back off the Multiplier and Voltage to 100.000 and 1.45v and drop the temp to around 100'F @ 4.5Ghz. Also on the Pump / Res I am using, are you supposed to fill them up to the top? correct. It is filled about 1/8" from the top of the Res It is the XSPC Res / Pump Combo on 1/2 x 3/4" Tubing Does that fill quantity sound about right? That was the only unsure part I had.

Thank You very much for your concerns on Chip Temps, I do appreciate it. As this is my first time doing the liquid cooled and OC'ing stuff.
Dev
OK let's look at some baselines - pretty much everyone looks at degrees celsius when we talk about CPU temps - I normally use F, for everyday things like weather, furnace settings, whatever. But I've learned Celcius as kind of a "separate" type of temperature scale if you follow - I couldn't tell you what X celcius was in F, but I can tell you some very few conversion points that give me reference. For example, I keep in mind that 100c = 212f. I can tell you water freezes at 0c and boils at 100c.

Now, let's take your temps: You don't say if you are running at stress or idle, or somewhere in between for CPU load.

You say you are getting 131F after it levels out at 4h of running. The "slow climb" from 107 to 131 tells me you are likely idling and/or doing lightweight tasks like web browsing where speedstep and other power saving features in the processor are basically almost turning the chip off. (Oversimplified, but suffice it to say if properly configured, your processor spends a lot of its life clocked down to 1600mhz or so for normal desktop browsing etc use).

So this would say you are giving us "idle temps".

To give you an idea of my temps (a 3820 O/C'd to 4.8ghz and change) at idle, the average core temp is somewhere around 28C-31C, or 82.4F- . That is on water, with the radiators in the basement below my machine, so ambient temps are much lower (50 degrees F, +/-, year-round), instead of room temperature.

My stress temps on heavy watercooling @ 1.386v, 4.8ghz and change, are around 52-54c, or 125.6F-129.2F. Pushing 50F water at a high flowrate through several radiators.

You see why I worry that you may be in F and getting C readings which would be more consistent with that extremely high voltage you are pushing through it.
Quote:
Originally Posted by pc-illiterate View Post

i would never go over 1.45v on sb-e under water. it doesnt tolerate voltage like sb does. 1.5+v is asking for quicker degradation.

+1 on this - And the closer you get to that 1.45V, the the LGA2011 chips get VERY hot VERY QUICKLY on a steep curve.

Generally, you want your voltage to be AS LOW as the clocks you are running will be stable at.
post #166 of 180
Thread Starter 
Quote:
Originally Posted by speck55 View Post

OK let's look at some baselines - pretty much everyone looks at degrees celsius when we talk about CPU temps - I normally use F, for everyday things like weather, furnace settings, whatever. But I've learned Celcius as kind of a "separate" type of temperature scale if you follow - I couldn't tell you what X celcius was in F, but I can tell you some very few conversion points that give me reference. For example, I keep in mind that 100c = 212f. I can tell you water freezes at 0c and boils at 100c.

Now, let's take your temps: You don't say if you are running at stress or idle, or somewhere in between for CPU load.

You say you are getting 131F after it levels out at 4h of running. The "slow climb" from 107 to 131 tells me you are likely idling and/or doing lightweight tasks like web browsing where speedstep and other power saving features in the processor are basically almost turning the chip off. (Oversimplified, but suffice it to say if properly configured, your processor spends a lot of its life clocked down to 1600mhz or so for normal desktop browsing etc use).

So this would say you are giving us "idle temps".

To give you an idea of my temps (a 3820 O/C'd to 4.8ghz and change) at idle, the average core temp is somewhere around 28C-31C, or 82.4F- . That is on water, with the radiators in the basement below my machine, so ambient temps are much lower (50 degrees F, +/-, year-round), instead of room temperature.

My stress temps on heavy watercooling @ 1.386v, 4.8ghz and change, are around 52-54c, or 125.6F-129.2F. Pushing 50F water at a high flowrate through several radiators.

You see why I worry that you may be in F and getting C readings which would be more consistent with that extremely high voltage you are pushing through it.
+1 on this - And the closer you get to that 1.45V, the the LGA2011 chips get VERY hot VERY QUICKLY on a steep curve.

Generally, you want your voltage to be AS LOW as the clocks you are running will be stable at.

Ok, there were a few things you just mentioned that I found very very sensible and correct
To SPEK55
1. Using C instead of F
2. How the Speed Stepping throttles down the processor during light loads (That was confusing me because at a full Idle I was running at 1.32v @1200 Mgz after the machine was booted up for a few minutes. Right after boot up it did show 4.1 Ghz then throttled down and I thought that was instability due to too low of voltage therefore I raised it to 1.51. Bad and based on an uninformed decision by me). as I was watching it on CPU-Z.

Here is the link to my CPU-Z specs
http://valid.canardpc.com/show_oc.php?id=2781787

Current Temps


3. The temps I was mentioning were during a Backup from a NAS to a Raid Array on the Overclocked PC.

I do believe in the computer business I see now why they would rather use C instead of F, Accuracy, as the conversion factor is all over the place depending on what temp range you are dealing with
ie: 0C = 32F which is a factor of x3 whereas at 100C = 212F is a factor of x2. This tells me that C is much more accurate and stable than F readings are.

The Dev learned and respected this info greatly and I thank you for that.

I am not going to fire up the machine rite now because it is almost 90' F outside rite now and just too damn hot. but when I do the first thing I am going to do is drop the voltage back to around 1.40 and see what happens and maybe bump up the multiplier to 120.000 instead of 100.000 as that is what things kinda defaulted to on the original overclock that is displayed in my CPU-z specs....

If there are any other settings and specs you can share I would greatly appreciate it You seem to know your stuff when it comes to this. Most everyone here says that each scenario is different therefore do not want to advise, and I respect that. (I promise I wont hold you to anything if something goes sideways hehe haah)..

Regards
Dev
Edited by webdevii - 5/13/13 at 12:50pm
post #167 of 180
Quote:
Originally Posted by webdevii View Post

Ok, there were a few things you just mentioned that I found very very sensible and correct
To SPEK55
1. Using C instead of F
2. How the Speed Stepping throttles down the processor during light loads (That was confusing me because at a full Idle I was running at 1.32v @1200 Mgz after the machine was booted up for a few minutes. Right after boot up it did show 4.1 Ghz then throttled down and I thought that was instability due to too low of voltage therefore I raised it to 1.51. Bad and based on an uninformed decision by me). as I was watching it on CPU-Z.

Here is the link to my CPU-Z specs
http://valid.canardpc.com/show_oc.php?id=2781787

Current Temps


3. The temps I was mentioning were during a Backup from a NAS to a Raid Array on the Overclocked PC.

I do believe in the computer business I see now why they would rather use C instead of F, Accuracy, as the conversion factor is all over the place depending on what temp range you are dealing with
ie: 0C = 32F which is a factor of x3 whereas at 100C = 212F is a factor of x2. This tells me that C is much more accurate and stable than F readings are.

The Dev learned and respected this info greatly and I thank you for that.

I am not going to fire up the machine rite now because it is almost 90' F outside rite now and just too damn hot. but when I do the first thing I am going to do is drop the voltage back to around 1.40 and see what happens and maybe bump up the multiplier to 120.000 instead of 100.000 as that is what things kinda defaulted to on the original overclock that is displayed in my CPU-z specs....

If there are any other settings and specs you can share I would greatly appreciate it You seem to know your stuff when it comes to this. Most everyone here says that each scenario is different therefore do not want to advise, and I respect that. (I promise I wont hold you to anything if something goes sideways hehe haah)..

Regards
Dev

OK so those are idle temps in Celsius. Those are pretty high as far as idle temps go - your chip doing no work is running about the same temperature as mine working 100% to the max. With (4) GTX 680's in the same loop.

Speedstep basically is a power saving measure, where the system will throttle the CPU multiplier down when there is no need/demand for processor power by a program or game. There is no sense in having your processor running full bore at 4+ Ghz to run a web browser or a screen saver, or typing a letter in Microsoft Word, right? So the processor waits until it get some load at the lowest step - once it becomes busy at the lowest step, it will bump up to the next notch. Once it gets busy at that notch, it will bump up again to the next notch up. If it is still busy, it will go up. And so on, until you hit your full clock rate (the one you expect to see). Power management can step through all of these notches and back quicker than you can blink.

Now you mention throttling due to temperature being too high - this may well be happening - we will find out when I have you run Prime95, but with idle temperatures like you have, I'm not going to have you do that just yet. I'm certain that you at this point will exceed the absolute limits of the chip and possibly kill it. Excess volts + excess temp + certain BIOS settings can be lethal to the CPU, and sometimes the motherboard - particularly the power delivery portion (PWM/MOSFETs).

So now we know your idle temps from that RealTemp screen shot.

Need to know what you have in your BIOS for settings for BCLK, Gear Ratio, Multiplier, Turbo Boost Ratios, CPU Core Voltage, Memory Voltage, any other voltages you have changed as well.

While you are in there, you may as well drop the BCLK back to 100.0 mhz, and the gear ratio back to 1.0 x - there is little reason for you to be using the base clock to over clock, since you have the "K" CPU, you can do most if not all of your over clock by just changing the multiplier. My "partially locked" 3820 I got 4.8 ghz out of without taking the base clock above 100.00. When you change BCLK, you are not only overclocking the CPU itself, you are overclocking a lot of other things - memory controllers, PCI express interfaces, memory -- many many things are based on this 100.00 base clock value.

Also what waterblock do you have for your CPU, what brand thermal paste did you use, and how much of it?


Honestly a 6C/12T $600 chip isn't a very good candidate for your first over clock. It's a potentially very expensive learning experience, especially at the voltages you've had it at. I mean, feel free to practice on my 3820, I'll just hang on to your 3930k until you get the swing of things. thumb.gif Seriously though, the more cores you have in a processor, the less you are going to get an overclock overall, because as you have 2 more cores and 4 more threads generating heat in the same space. What will you be using this machine for when we get you stable?
post #168 of 180
Ok, now I'm worried.. Which I rarely, and I mean very rarely do get worried.. My H100 in my HAF X(below the top MFlows) is obviously exactly above my GTX 690..

Ok I'm a bit worried now..
Koniakki
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 6700K@4.7Ghz ASUS ROG MAXIMUS VIII Z170 HERO Zotac Geforce GTX 1080Ti FE 32GB G.Skill TridentZ 3733@CL15 
Hard DriveHard DriveHard DriveHard Drive
Samsung 950 PRO 256GB M.2 SSD Seagate Archive v2 8TB Seagate ST4000DM000 4TB Seagate Archive v2 8TB 
CoolingOSMonitorPower
Alphacool NexXxoS Cool Answer 480 D5/UT60 W10 PRO x64 OEM DIGITAL Samsung PS-59D550 Corsair RMx 750W GOLD 
CaseMouseAudio
Thermatake Core P5 Logitech MK700/MK710 LG BH9540TW 9.1 Home Theater 
  hide details  
Koniakki
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 6700K@4.7Ghz ASUS ROG MAXIMUS VIII Z170 HERO Zotac Geforce GTX 1080Ti FE 32GB G.Skill TridentZ 3733@CL15 
Hard DriveHard DriveHard DriveHard Drive
Samsung 950 PRO 256GB M.2 SSD Seagate Archive v2 8TB Seagate ST4000DM000 4TB Seagate Archive v2 8TB 
CoolingOSMonitorPower
Alphacool NexXxoS Cool Answer 480 D5/UT60 W10 PRO x64 OEM DIGITAL Samsung PS-59D550 Corsair RMx 750W GOLD 
CaseMouseAudio
Thermatake Core P5 Logitech MK700/MK710 LG BH9540TW 9.1 Home Theater 
  hide details  
post #169 of 180
Thread Starter 
Quote:
Originally Posted by Koniakki View Post

Ok, now I'm worried.. Which I rarely, and I mean very rarely do get worried.. My H100 in my HAF X(below the top MFlows) is obviously exactly above my GTX 690..

Ok I'm a bit worried now..

I know the H100 does sound worrysome however from what I have been told there is only a .5% failure rate on these. And Corsair does honor there warranty....

Although I would recommend investing in and building your own custom loop. I had Joe at Frozen CPU assist myself in doing it and it is working awesome however the cost was about 80% more than the H100 was roughly $600.00.

Good Luck and if you need any basic advise feel free to ask I will be glad to share what I do know as well as finding out if I do not.

Dev
post #170 of 180
Quote:
Originally Posted by webdevii View Post

I know the H100 does sound worrysome however from what I have been told there is only a .5% failure rate on these. And Corsair does honor there warranty....

Although I would recommend investing in and building your own custom loop. I had Joe at Frozen CPU assist myself in doing it and it is working awesome however the cost was about 80% more than the H100 was roughly

Good Luck and if you need any basic advise feel free to ask I will be glad to share what I do know as well as finding out if I do not.

Dev

Just my luck. I just bought a brand new H100 a week ago and I got an offer from a friend for a 1month old H100i for the same price. F**k! tongue.gif
Koniakki
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 6700K@4.7Ghz ASUS ROG MAXIMUS VIII Z170 HERO Zotac Geforce GTX 1080Ti FE 32GB G.Skill TridentZ 3733@CL15 
Hard DriveHard DriveHard DriveHard Drive
Samsung 950 PRO 256GB M.2 SSD Seagate Archive v2 8TB Seagate ST4000DM000 4TB Seagate Archive v2 8TB 
CoolingOSMonitorPower
Alphacool NexXxoS Cool Answer 480 D5/UT60 W10 PRO x64 OEM DIGITAL Samsung PS-59D550 Corsair RMx 750W GOLD 
CaseMouseAudio
Thermatake Core P5 Logitech MK700/MK710 LG BH9540TW 9.1 Home Theater 
  hide details  
Koniakki
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 6700K@4.7Ghz ASUS ROG MAXIMUS VIII Z170 HERO Zotac Geforce GTX 1080Ti FE 32GB G.Skill TridentZ 3733@CL15 
Hard DriveHard DriveHard DriveHard Drive
Samsung 950 PRO 256GB M.2 SSD Seagate Archive v2 8TB Seagate ST4000DM000 4TB Seagate Archive v2 8TB 
CoolingOSMonitorPower
Alphacool NexXxoS Cool Answer 480 D5/UT60 W10 PRO x64 OEM DIGITAL Samsung PS-59D550 Corsair RMx 750W GOLD 
CaseMouseAudio
Thermatake Core P5 Logitech MK700/MK710 LG BH9540TW 9.1 Home Theater 
  hide details  
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Water Cooling
This thread is locked  
Overclock.net › Forums › Cooling › Water Cooling › My Corsair H100 leaked on my new 680-GTX