Overclock.net › Forums › Intel › Intel Motherboards › EVGA 750i FTW Overclock Thread
New Posts  All Forums:Forum Nav:

EVGA 750i FTW Overclock Thread - Page 241

post #2401 of 2885
Again I get schooled on temps. Well maybe ;-). So it seems at some conferance last year Intel released the TJmax specs which yes indeed do say 90 for the G0.
The link you sent though I would have to disagree with (nothing to do with you of course, more on the writer). Here's why.
Page 15 states the G0 TJmax is 90C. This is also what they say it si for the B3. In the bottom of the page notes they have a blurp that the G0 TJmax was upped to allow for quieter cooling blah blah. So if it was upped from the B3 (which is stated at 90) and it shows the same..... I guess they missed the 80C TJMax for the B3 somewhere.
Looking at page 11 with the Intel recommended temp tools, they have sceen shots of Everest, core temp and real temp. The first 2 are reading the 9650 at 105C TJMax, and Real temp shows 95C which is what the doc say's a little farther down. Maybe this is part of the reason so many people get confused on this crap.
So from what I have read TJMax is the point where themal throttling begins and actually counts down. I'm going somewhere with all this.
Real Temp which shows core temps + distance to TJmax is a good thing to demonstarte this. Using TJmax as 100, temp is 30, distance to TJmax 61
Changre the TJMax to 90 and temp is 29. Distance to TJMax is still 61. I never realized it all before but core temp is not a real temp at all. It is reading how close you are to TJmax and based on what it beleves the TJMax is giving you a temperature. Makes sense (a little more to me now).
So now here is my next thing. Currrent temp in my room is 26.8C, setting to a TJmax in Real temp to 90 my cores are 28C? Yeah ok I am on water but that means the CPU is 2.2C warmer than outside air? (pausing for few minutes, just threw a digital thermometer inside case to see what ambient is in there as well as current room temp).

So now that that is done, Although this would seem to be good news it doesn't really affect a thing. Using a TJmax of 90 shows my core temps at 28C, who cares what it says the temp is, it doesn't matter, why do they give this number on software I do not know. The real number that cannot be mucked up is distance to TJmax, software is taking this number and subtracting it from what it believes the TJmax is for the CPU and saying this is the temp. Across a couple different softwares core temp varies 10C, but distance to TJmax is constant (makes more sense now). No software can accurately give a corresponding temperature from distance to TJMax since it is calibrated on a per processor basis at the factory. This could explain why my processor is saying it is the same temp as the air in case and water temp in tower. Which is 2C less than ambient air. That would be the perfect cooling system (although only good for idle). I figured it would have to be TJMax at 100 since the cores would have to be warmer than the water, if only bya little bit. The core temp is math done against number given out by Intel. But they cannot account for the calibration on how that changes things throughout the temp range. The only reliable number really is distance to TJmax which when it hits zero it starts to take steps to cool itself.
And of course now I find this link
http://forums.guru3d.com/showthread.php?t=290704
thermal test of Q6600 with pictures.

Hope I didn't confuse anyone even more by this post. I will read some more about it later, but prolly won't bore ya all with it.
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
post #2402 of 2885
Quote:
Originally Posted by blkrnbw View Post
Again I get schooled on temps. Well maybe ;-). So it seems at some conferance last year Intel released the TJmax specs which yes indeed do say 90 for the G0.
The link you sent though I would have to disagree with (nothing to do with you of course, more on the writer). Here's why.
Page 15 states the G0 TJmax is 90C. This is also what they say it si for the B3. In the bottom of the page notes they have a blurp that the G0 TJmax was upped to allow for quieter cooling blah blah. So if it was upped from the B3 (which is stated at 90) and it shows the same..... I guess they missed the 80C TJMax for the B3 somewhere.
Looking at page 11 with the Intel recommended temp tools, they have sceen shots of Everest, core temp and real temp. The first 2 are reading the 9650 at 105C TJMax, and Real temp shows 95C which is what the doc say's a little farther down. Maybe this is part of the reason so many people get confused on this crap.
So from what I have read TJMax is the point where themal throttling begins and actually counts down. I'm going somewhere with all this.
Real Temp which shows core temps + distance to TJmax is a good thing to demonstarte this. Using TJmax as 100, temp is 30, distance to TJmax 61
Changre the TJMax to 90 and temp is 29. Distance to TJMax is still 61. I never realized it all before but core temp is not a real temp at all. It is reading how close you are to TJmax and based on what it beleves the TJMax is giving you a temperature. Makes sense (a little more to me now).
So now here is my next thing. Currrent temp in my room is 26.8C, setting to a TJmax in Real temp to 90 my cores are 28C? Yeah ok I am on water but that means the CPU is 2.2C warmer than outside air? (pausing for few minutes, just threw a digital thermometer inside case to see what ambient is in there as well as current room temp).

So now that that is done, Although this would seem to be good news it doesn't really affect a thing. Using a TJmax of 90 shows my core temps at 28C, who cares what it says the temp is, it doesn't matter, why do they give this number on software I do not know. The real number that cannot be mucked up is distance to TJmax, software is taking this number and subtracting it from what it believes the TJmax is for the CPU and saying this is the temp. Across a couple different softwares core temp varies 10C, but distance to TJmax is constant (makes more sense now). No software can accurately give a corresponding temperature from distance to TJMax since it is calibrated on a per processor basis at the factory. This could explain why my processor is saying it is the same temp as the air in case and water temp in tower. Which is 2C less than ambient air. That would be the perfect cooling system (although only good for idle). I figured it would have to be TJMax at 100 since the cores would have to be warmer than the water, if only bya little bit. The core temp is math done against number given out by Intel. But they cannot account for the calibration on how that changes things throughout the temp range. The only reliable number really is distance to TJmax which when it hits zero it starts to take steps to cool itself.
And of course now I find this link
http://forums.guru3d.com/showthread.php?t=290704
thermal test of Q6600 with pictures.

Hope I didn't confuse anyone even more by this post. I will read some more about it later, but prolly won't bore ya all with it.
You got it!!!! The only true way to monitor temps is distance from TJmax. It doesn't matter what you have TJmax calibrated to in any software program, the distance from TJmax is still going to be the same. Try using RealTemp which will show you the distance.
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
post #2403 of 2885
Quick OC question!

How safe is it to run 1.35 on the FSB on this board?
I ask because i thought I had hit a wall with my quad and I was running 1.4v on the vcore and 1.3 on the FSB. I just decided to try bumping up the FSB to 1.35 (which is red) and now I've been able to lower my voltages to 1.325 on the vcore and still have it stable in IBT! So how safe is 1.35 on FSB?
NotSoSLi
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Asus Maximus Hero VII EVGA GTX 980ti G.Skill 16GB DDR3 1600 
Hard DriveOSMonitorKeyboard
Samsung 840 EVO Windows 8.1 x64 ACER 1440p 144hz G15 
PowerCaseMouse
Corsair 850tx Stormtrooper MX518 
  hide details  
Reply
NotSoSLi
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Asus Maximus Hero VII EVGA GTX 980ti G.Skill 16GB DDR3 1600 
Hard DriveOSMonitorKeyboard
Samsung 840 EVO Windows 8.1 x64 ACER 1440p 144hz G15 
PowerCaseMouse
Corsair 850tx Stormtrooper MX518 
  hide details  
Reply
post #2404 of 2885
Quote:
Originally Posted by SLI_Maniac View Post
Quick OC question!

How safe is it to run 1.35 on the FSB on this board?
I ask because i thought I had hit a wall with my quad and I was running 1.4v on the vcore and 1.3 on the FSB. I just decided to try bumping up the FSB to 1.35 (which is red) and now I've been able to lower my voltages to 1.325 on the vcore and still have it stable in IBT! So how safe is 1.35 on FSB?
I believe it's red to the CPU not the board, but I might be wrong. If you tweak the GLT lanes right you can run lower FSB voltage.
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
post #2405 of 2885
So why the F&^k do companies and developers continue to claim core temps? I doubt anyone can answer this. But it seems if it is not a temp at all but a calculation done off a count down (so to speak) which also is not a standard since it is calculated per chip at factory. And this number only matters if you actually hit the 0 mark and the processor throttle back to cool itself. This is not the thermal stuff in BIOS (EIST etc) but a CPU controlled function. So by pushing until you hit 0 you wont damage anything so to speak but the CPU is gonna slow ya down.
So all in all my view of core temperature is pretty a non existant setting. Distance to TJMax still matters but temperature is inacurate and thus meaningless, and I am sure it is why so many people are confused on the subject. Reall the way the calibration is done the CPU will trip the thermals before any of the cores will since it is calibrated slightly above to ensure there are not false TJmax hits that throttle needlessly.

Thats it for me on this rant. Pizza did rep ya, should be 2 or 3 as many posts that have helped me out. Lets move to networking and server related and I will get ya back ;-).


Edited to add...
So now that my actual cores were 10C below what I thought (63 vs 53 again not that it matters), and CPU temp (which matters) was at 43C OC'd I have alot more room than I thought in the max heat area. Maybe its time to try for 3.8 now?
Edited by blkrnbw - 8/9/09 at 5:45pm
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
post #2406 of 2885
Quote:
Originally Posted by blkrnbw View Post
So why the F&^k do companies and developers continue to claim core temps? I doubt anyone can answer this. But it seems if it is not a temp at all but a calculation done off a count down (so to speak) which also is not a standard since it is calculated per chip at factory. And this number only matters if you actually hit the 0 mark and the processor throttle back to cool itself. This is not the thermal stuff in BIOS (EIST etc) but a CPU controlled function. So by pushing until you hit 0 you wont damage anything so to speak but the CPU is gonna slow ya down.
So all in all my view of core temperature is pretty a non existant setting. Distance to TJMax still matters but temperature is inacurate and thus meaningless, and I am sure it is why so many people are confused on the subject. Reall the way the calibration is done the CPU will trip the thermals before any of the cores will since it is calibrated slightly above to ensure there are not false TJmax hits that throttle needlessly.

Thats it for me on this rant. Pizza did rep ya, should be 2 or 3 as many posts that have helped me out. Lets move to networking and server related and I will get ya back ;-).


Edited to add...
So now that my actual cores were 10C below what I thought (63 vs 53 again not that it matters), and CPU temp (which matters) was at 43C OC'd I have alot more room than I thought in the max heat area. Maybe its time to try for 3.8 now?
Yea, companies tell you what you want to here not what you need to here. Also, try and imagine explaining TJmax distance to the normal PC user. BTW, TJmax is not the point in which damage starts. If you hit the TJmax, you have likely already caused some damage to your chip before you reach TJmax. Also, most chips will crap out and restart or shutoff due to instability from increased temps before ever reaching TJmax.

From all the reading I've done in Intel's data sheets, damage will likely start somewhere around 15C before TJmax is reached, but like everything else it just depends on the chip. You might have chips that start to damage at 20C. Though, this is also where stability becomes another factor. Running a chip unstable for can cause chip damage/degrading.

Just bare in mind, if you stay ~30C away from TJmax you should be OK.
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
post #2407 of 2885
I think you read that wrong. TJmax is calibrated above spec, but Tcase will hit its high mark first and take steps to cool if it is enabled in BIOS, if it cannot then TJmax is the point at which the processor will slow/error or what not. The link I posted last night on a Q6600 his TJmax was way above what it should have been, and it did not actually hit the PROCHOT until the CPU hit the 71C mark, cores were showing 80 through everest, 98 through real temp (with TJmax set to 100C). The CPU hit first so I think the calibration is indeed above so it does not hit TJ before Tc. Read that link rather interesting.

Brian
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
post #2408 of 2885
Quote:
Originally Posted by blkrnbw View Post
I think you read that wrong. TJmax is calibrated above spec, but Tcase will hit its high mark first and take steps to cool if it is enabled in BIOS, if it cannot then TJmax is the point at which the processor will slow/error or what not. The link I posted last night on a Q6600 his TJmax was way above what it should have been, and it did not actually hit the PROCHOT until the CPU hit the 71C mark, cores were showing 80 through everest, 98 through real temp (with TJmax set to 100C). The CPU hit first so I think the calibration is indeed above so it does not hit TJ before Tc. Read that link rather interesting.

Brian
Well, that's only if you're talking about using stock voltage. Once you increase the voltage the chips power consumption changes. Intel's data sheets have a graph in which power comsumption can be estimated from temperature. Though, once we start overvolting we can thru those graphs out the window. So, if the chip reaches TJmax with stock voltage, the throttling may actually save the chip, but we are not talking about stock voltages.


Quote:
To determine a processor's case temperature specification based on the thermal profile, it is necessary to accurately measure processor power dissipation. Intel has developed a methodology for accurate power measurement that correlates to Intel test temperature and voltage conditions. Refer to the appropriate Thermal and Mechanical Design Guidelines (see Section 1.2).
Quote:
In the event of a catastrophic cooling failure, the processor will automatically shut down when the silicon has reached a temperature approximately 20°C above the maximum Tc. Assertion of THERMTRIP# (Thermal Trip) indicates the processor junction temperature has reached a level beyond where permanent silicon damage may occur. Upon assertion of THERMTRIP#, the processor will shut off its internal clocks (thus, halting program execution) in an attempt to reduce the processor junction temperature.

Edited by PizzaMan - 8/9/09 at 10:31pm
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
E8400, E8600, 980BE T2RS+, P5E3, TA890FXE 8800GTS's,GTX's,GS,Ultra's,9600GT Lots of D9's and some crap DDR3 
Hard DrivePowerCase
Seagate 250GB + 80GB WD for benching TT 875w, Corsair 650TX Custom bench station 
  hide details  
Reply
post #2409 of 2885
True dat for not stock voltage, but we are not talking about stock cooling either. We are essentially talking catastrophic failure. I see this as your buddy while drunk pulled your heat sink off or similar

If you play be all the rules TjMax on Q6600 is 90C TcMax is 71C. TJmax should not hit 90C prior to TcMax hitting 71 this is due to calibration at factory ensuring Tjmax is always lower than Tc.

Even at stock voltage if you pull a heat sink you may not have a processor. They have come along way with this (anyone remember the Tom's artcile with the torched AMD?). and under any normal conditions this should not be done (removing the heat sink that is). Now with increased voltages if the thermal protections are enables via BIOS you should not run into this situation, a heat sink alone should bleed enough off to where the temp is not a immediate spike and be able to deal with the emergancy. Now with overclocking most people (myself included) turn off the safety mechanisms within BIOS that watch processor temps and take actions prior to Tc or Tj. So yes if OC'ing and running unprotected and kick the heat sink off the processor there is a chance there is going to be some permanent damage. Lets compare this is to someone finding a $10.00 hooker on the street and having unprotected sex, there is a damn good chance there is going to be some damage. maybe not life threatening but damage none the less. In both cases we know the risk and if we still make the decision to go ahead and do it, if something bad comes out of it we can't say we didn't know.
This is why after OC'in and testing I slow everything back down. The machine stays on 24x7 and gets used pretty often for testing stuff at work, I would be less paranoid with air cooling than water really. Air you still have backup fans in case that blow over sink. Water you get a steam generator inside your computer, no where for heat to go. I run EIST, TM1 and TM2 and C1E etc under normal conditions and have a rule in Nvidia monitor that starts a shutdown in CPU exceed 65C to keep safe. Not sure if anyone realizes this but I was watching my "kill-a-watt" (little box that goes between PC plug and wall socket that shows what you are drawing from socket.) But when I blue screened the machine was still drawing the same amount of power as when running. So in the event of a blue screen it does not all of a sudden stop drawing power but keeps sucking up what it was getting. I am going to measure temps during this kind of situation someday. What does thiis mean? Well nothing more than if you overheat and OS crashes your still adding heat.

Quote regarding BIOS safety measures
The power monitor continuously tracks the die temperature. If the temperature reaches the maximum allowed value, a throttle mechanism is initiated. A multi-level tracking algorithm is implemented. Throttling starts with the more efficient dynamic voltage scaling policy and if not sufficient, the power monitor algorithm continues lowering the frequency. If an extreme cooling malfunction occurs, an Out of Spec notification will be initiated, requesting controlled shutdown. Lastly, the CPU can initiate a thermal shutdown and turn off the system.

Thermal monitor function is not expected to be activated under these normal operation conditions. The thermal monitor mechanism ensures that the CPU will never exceed the CPU-specified parameters and guarantees functionality and reliability at any time.

End Quote

Lastly The graph you mentioned is from the thermal spec I believe, this is used to determine fan speed on the heat sink in relation to power draw and heat generated As temp goes up. Manufacturers use this as a base line for heat dissipation.

I gotta say it is nice to talk to someone that knows their ****.
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
my b&^$h.
(13 items)
 
  
CPUMotherboardGraphicsRAM
Q6600 G0 750i FTW 9600's SLI 8GB 
Hard DriveOptical DriveOSMonitor
too many just one XP 64 Viewsonic 22" 
KeyboardPowerCaseMouse
click clack Antec True 550 + 270 watt secondary for vid Zalman GS1000 white, with whickers 
Mouse Pad
more of a cage but can be called a pad 
  hide details  
Reply
post #2410 of 2885
Quote:
Originally Posted by PizzaMan View Post
I believe it's red to the CPU not the board, but I might be wrong. If you tweak the GLT lanes right you can run lower FSB voltage.
If its specific to the chip do you happen to know of any spec sheets for 45nm quads max FSB?
NotSoSLi
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Asus Maximus Hero VII EVGA GTX 980ti G.Skill 16GB DDR3 1600 
Hard DriveOSMonitorKeyboard
Samsung 840 EVO Windows 8.1 x64 ACER 1440p 144hz G15 
PowerCaseMouse
Corsair 850tx Stormtrooper MX518 
  hide details  
Reply
NotSoSLi
(13 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 4790k Asus Maximus Hero VII EVGA GTX 980ti G.Skill 16GB DDR3 1600 
Hard DriveOSMonitorKeyboard
Samsung 840 EVO Windows 8.1 x64 ACER 1440p 144hz G15 
PowerCaseMouse
Corsair 850tx Stormtrooper MX518 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel Motherboards
Overclock.net › Forums › Intel › Intel Motherboards › EVGA 750i FTW Overclock Thread