Overclock.net › Forums › Intel › Intel CPUs › [Official] Delidded Club / Guide
New Posts  All Forums:Forum Nav:

[Official] Delidded Club / Guide - Page 1127

post #11261 of 33692
Quote:
Originally Posted by chronicfx View Post

If you want my take on it as a well trained scientist, I don't think there is any difference between leaving the heatspreader and not. People are getting a placebo effect and thinking that going direct die is really making a difference. Technically the entire bottleneck to your heat dissapation would be from the die through that first layer of TIM as the area of contact is the smallest making the heat transfered per unit area the highest of any interface. So whether that heat is going directly to a cpu heatsink/waterblock or into the CPU heat spreader and then through the same TIM interface spread over a larger area thereby reducing the heat transfer requirements (heat per unit area) dramatically and then into the cpu heatsink/waterblock doesn't matter at that point. The only time it could matter is if your heatspreader does not make good contact with the die surface from being concave/convex, but you can have a concave/convex water block too and get bad results. So if you want my final feeling on direct die, it just doesn't make any scientific sense. Sorry guys you're just risking a cracked die. IMHO

I really didn't want to start a debate, but as a "well trained scientist", where is your empirical evidence? Certainly a " well trained scientist" would do some kind of testing before speculating that it "doesn't make any scientific sense"

I have done some testing, not a whole lot, as I was not expecting to get published, but just enough to convince myself. I have done several mounts both with and without the IHS and each time they are consistent with the following table:
Warning: Spoiler! (Click to show)

  • Did I get a reduction in temperature -- Absolutely, 7 to 8 degrees off my hottest core at 5.0GHz.
  • Did it allow me to lower my VCore -- Absolutely, 0.020V at 5.0GHz.
  • I want to use LN2, will going direct die help me? -- I don't think so, my testing at 0 degrees showed no improvement, direct die vs IHS.
  • If I go direct die does it mean I can run 24/7 at 5.3 GHz? -- Probably not, as frequencies go up, so does the voltage required to gate those circuits at those speeds.


Here is a chart I did for my own research. It shows the minimum VCore required to pass 10 passes of IBT for my chip. The results are from a delidded chip before going direct die, but it should give you some general idea of what kind of voltages you will need to reach the next increment in speed:
Warning: Spoiler! (Click to show)

Edited by MikeG - 2/8/13 at 2:41pm
Smoke and Mirrors
(12 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7 3770K  Z77 MPower (MS-7751) MSI R7950 Twin Frozr 3GD5/OC Intel(R) HD Graphics 4000 
RAMHard DriveCoolingCooling
Avexir MSI Z77 MPOWER Optimized DDR3 2400 OCZ Vertex 4 VTX4-25SAT3-256G 2.5" 256GB SATA I... Koolance CPU-380I Water Block XSPC EX360 Radiator  
CoolingOSPowerCase
XSPC D5 Dual Bay Reservoir/Pump Combo Windows 7 Corsair AX860i NZXT Switch 810  
  hide details  
Reply
Smoke and Mirrors
(12 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7 3770K  Z77 MPower (MS-7751) MSI R7950 Twin Frozr 3GD5/OC Intel(R) HD Graphics 4000 
RAMHard DriveCoolingCooling
Avexir MSI Z77 MPOWER Optimized DDR3 2400 OCZ Vertex 4 VTX4-25SAT3-256G 2.5" 256GB SATA I... Koolance CPU-380I Water Block XSPC EX360 Radiator  
CoolingOSPowerCase
XSPC D5 Dual Bay Reservoir/Pump Combo Windows 7 Corsair AX860i NZXT Switch 810  
  hide details  
Reply
post #11262 of 33692
Quote:
Originally Posted by MikeG View Post

I really didn't want to start a debate, but as a "well trained scientist", where is your empirical evidence? Certainly a " well trained scientist" would do some kind of testing before speculating that it "doesn't make any scientific sense"

I have done some testing, not a whole lot, as I was not expecting to get published, but just enough to convince myself. I have done several mounts both with and without the IHS and each time they are consistent with the following table:
Warning: Spoiler! (Click to show)

  • Did I get a reduction in temperature -- Absolutely, 7 to 8 degrees off my hottest core at 5.0GHz.
  • Did it allow me to lower my VCore -- Absolutely, 0.20V at 5.0GHz.
  • I want to use LN2, will going direct die help me? -- I don't think so, my testing at 0 degrees showed no improvement, direct die vs IHS.
  • If I go direct die does it mean I can run 24/7 at 5.3 GHz? -- Probably not, as frequencies go up, so does the voltage required to gate those circuits at those speeds.


Here is a chart I did for my own research. It shows the minimum VCore required to pass 10 passes of IBT for my chip. The results are from a delidded chip before going direct die, but it should give you some general idea of what kind of voltages you will need to reach the next increment in speed:
Warning: Spoiler! (Click to show)

"As well trained scientist" I am supposed to spot the difference between what needs the time to actually be tested and what can be answered through common sense. I think the 7-8 degree diffference means your waterblock has a smoother contact area to the die than your ihs does. Try lapping your ihs and get back to me if you feel like doing "experiments".
post #11263 of 33692
Also if you are trying to tell me that an 8 degreed drop allowed you to go from 1.5v to 1.3v at 5ghz to pass IBT then youshouldn't be posting these types of results. That is utterly rediculous.
post #11264 of 33692
Quote:
Originally Posted by chronicfx View Post

Also if you are trying to tell me that an 8 degreed drop allowed you to go from 1.5v to 1.3v at 5ghz to pass IBT then youshouldn't be posting these types of results. That is utterly rediculous.

I am not trying to be mean here. I am just telling you the truth as I see it. I will also give you background on "well trained", my father has a Ph.D in Chem and has been the chairman of chemistry at a university since I was born, my older brother has a Ph.D in chem and works at Bayer in Switzerland (Companies don't normally let american chemists work in switzerland or even the rest of europe for that matter unless you have the skills), my brother and I have been running reactions and tinkering with million dollar instrumentation since middle school. Heck I can maintain them like a mechanic swapping parts when needed. I myself am a chemist at a company that made more money than procter and gamble last year by a landslide. I don't want to come off like some kid arguing with you. I have a fairly good understanding of thermodynamics and can do all of the math involved. I am not going to go through that all on here as helpful as it may be, I don't get paid for this and I just want to participate and still learn things, my expertise is not in computer science, there are guys on here that would stomp me when it comes to computers, i bet more than 60% of the guys on this forum would. I am here to learn just like you Mike, but when people come to me and say here is my data "look" and I immediately find some important issues missing, like "did you lap the underside of the heatsink and your water block before making these claims?" I am sorry that you took all of that time to produce the data, but I am not convinced and probably wouldn't be if you found the same result 10 times in a row.

Anyway Mike lets be friends and get back to normal. Thats why I say "well trained" and thats all I ever wanted to say.
Edited by chronicfx - 2/8/13 at 7:14am
post #11265 of 33692
Quote:
Originally Posted by chronicfx View Post

Also if you are trying to tell me that an 8 degreed drop allowed you to go from 1.5v to 1.3v at 5ghz to pass IBT then youshouldn't be posting these types of results. That is utterly rediculous.
I asked a simple question about going direct to die. MikeG has actually tested both and posted his results. Chronicfx, I don't know if you intended it but your posts come off as condescending and combative, and your claim that removing the IHS can't be anything but a placebo is illogical. That would be the same as claiming going from AS5 to CL Ultra on the die is nothing but a placebo.

You can very simply and easily test whether or not CLU give you temp drops vs. AS5, just like you can very simply and easily test whether going direct to die helps your temps if you have a setup that will allow it.

I have had three 3770k chips now and checked the "flatness" of the IHS on all of them. None were even remotely flat, and the worst part was they vary significantly. Their level of concavity was not the same, and one had measurably higher corners on one side than the others. Additionally the underside of the IHS that contacts the die is not flat either.

Taking those facts into account, you were exactly correct when you wrote "The only time it could matter is if your heatspreader does not make good contact with the die surface." It is quite literally physically impossible to make very good contact with any of the three Intel IHS I have inspected because they are not manufactured to any form of high standard to make them flat on the top or bottom.

The .2 was a typo, he listed it as a .02 drop previously, just like your misspelling of "rediculous" was a typo.
Ivy 5.0
(14 items)
 
Ivy 4.8
(15 items)
 
 
CPUMotherboardGraphicsRAM
i7-3770k 5.0 24/7 Maximus V Formula EVGA GTX 690 G.Skill Trident X 2400 (2x8gb) 
Hard DriveOptical DriveCoolingOS
Samsung 840 Pro 512GB Pioneer BD-RW Swiftech H220 Windows 7 Ultimate 64 bit 
MonitorKeyboardPowerCase
2 x NEC 30" Logitech G15 Corsair AX1200 NZXT Switch 810 
MouseAudio
Logitech G9x Logitech G51 
  hide details  
Reply
Ivy 5.0
(14 items)
 
Ivy 4.8
(15 items)
 
 
CPUMotherboardGraphicsRAM
i7-3770k 5.0 24/7 Maximus V Formula EVGA GTX 690 G.Skill Trident X 2400 (2x8gb) 
Hard DriveOptical DriveCoolingOS
Samsung 840 Pro 512GB Pioneer BD-RW Swiftech H220 Windows 7 Ultimate 64 bit 
MonitorKeyboardPowerCase
2 x NEC 30" Logitech G15 Corsair AX1200 NZXT Switch 810 
MouseAudio
Logitech G9x Logitech G51 
  hide details  
Reply
post #11266 of 33692
Chronicfx, saw your additional post now, thanks for that. This is a great thread, and healthy debate about subjects is a good thing. Hopefully we can all learn a little that way, or in my case learn a lot. smile.gif
Ivy 5.0
(14 items)
 
Ivy 4.8
(15 items)
 
 
CPUMotherboardGraphicsRAM
i7-3770k 5.0 24/7 Maximus V Formula EVGA GTX 690 G.Skill Trident X 2400 (2x8gb) 
Hard DriveOptical DriveCoolingOS
Samsung 840 Pro 512GB Pioneer BD-RW Swiftech H220 Windows 7 Ultimate 64 bit 
MonitorKeyboardPowerCase
2 x NEC 30" Logitech G15 Corsair AX1200 NZXT Switch 810 
MouseAudio
Logitech G9x Logitech G51 
  hide details  
Reply
Ivy 5.0
(14 items)
 
Ivy 4.8
(15 items)
 
 
CPUMotherboardGraphicsRAM
i7-3770k 5.0 24/7 Maximus V Formula EVGA GTX 690 G.Skill Trident X 2400 (2x8gb) 
Hard DriveOptical DriveCoolingOS
Samsung 840 Pro 512GB Pioneer BD-RW Swiftech H220 Windows 7 Ultimate 64 bit 
MonitorKeyboardPowerCase
2 x NEC 30" Logitech G15 Corsair AX1200 NZXT Switch 810 
MouseAudio
Logitech G9x Logitech G51 
  hide details  
Reply
post #11267 of 33692
Quote:
Originally Posted by justanoldman View Post

I asked a simple question about going direct to die. MikeG has actually tested both and posted his results. Chronicfx, I don't know if you intended it but your posts come off as condescending and combative, and your claim that removing the IHS can't be anything but a placebo is illogical. That would be the same as claiming going from AS5 to CL Ultra on the die is nothing but a placebo.

You can very simply and easily test whether or not CLU give you temp drops vs. AS5, just like you can very simply and easily test whether going direct to die helps your temps if you have a setup that will allow it.

I have had three 3770k chips now and checked the "flatness" of the IHS on all of them. None were even remotely flat, and the worst part was they vary significantly. Their level of concavity was not the same, and one had measurably higher corners on one side than the others. Additionally the underside of the IHS that contacts the die is not flat either.

Taking those facts into account, you were exactly correct when you wrote "The only time it could matter is if your heatspreader does not make good contact with the die surface." It is quite literally physically impossible to make very good contact with any of the three Intel IHS I have inspected because they are not manufactured to any form of high standard to make them flat on the top or bottom.

The .2 was a typo, he listed it as a .02 drop previously, just like your misspelling of "rediculous" was a typo.

Funny enough I read a forum where someone was trying to claim that ultra was no different and it was all to do with ambients. It made no sense lol
R@WR
(16 items)
 
Rawr V2
(16 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770k Z77 MPower (MS-7751) Asus Radeon R9 290 Corsair Dominator DHX Pro 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 3 SSD Seagate Barracuda LG Super Multi XSPC Raystorm laing ddc 3.25, ex 240 Swiftech X... 
OSMonitorKeyboardPower
Windows 8.1 Pro LG 24inch led Backlit Corsair K60 XFX 850WPro Fully Modular 
CaseMouseMouse PadAudio
Bitfenix Ronin Steel Series Sensei Razer Goliathus Steel Series Siberius V2 
CPUMotherboardGraphicsRAM
3770k Msi z77 mpower EVGA GTX 980 Corsair Dominator 
Hard DriveHard DriveCoolingOS
WD Blue OCZ Vertex 3 SSD Custom Water Cooling, Darkside 6 port reservoir... Windows 8.1 
MonitorKeyboardPowerCase
LG 24 inch 1080P CM Quickfire Rapid, custom keys from WASD XFX Pro Fully modular 850W Gold Corsair Air 540 
MouseMouse PadAudioAudio
Corsair M65 Razer Goliathus Roccat Kave True 5.1 Headset Asus Xonar DX 7.1 Sound card 
  hide details  
Reply
R@WR
(16 items)
 
Rawr V2
(16 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770k Z77 MPower (MS-7751) Asus Radeon R9 290 Corsair Dominator DHX Pro 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 3 SSD Seagate Barracuda LG Super Multi XSPC Raystorm laing ddc 3.25, ex 240 Swiftech X... 
OSMonitorKeyboardPower
Windows 8.1 Pro LG 24inch led Backlit Corsair K60 XFX 850WPro Fully Modular 
CaseMouseMouse PadAudio
Bitfenix Ronin Steel Series Sensei Razer Goliathus Steel Series Siberius V2 
CPUMotherboardGraphicsRAM
3770k Msi z77 mpower EVGA GTX 980 Corsair Dominator 
Hard DriveHard DriveCoolingOS
WD Blue OCZ Vertex 3 SSD Custom Water Cooling, Darkside 6 port reservoir... Windows 8.1 
MonitorKeyboardPowerCase
LG 24 inch 1080P CM Quickfire Rapid, custom keys from WASD XFX Pro Fully modular 850W Gold Corsair Air 540 
MouseMouse PadAudioAudio
Corsair M65 Razer Goliathus Roccat Kave True 5.1 Headset Asus Xonar DX 7.1 Sound card 
  hide details  
Reply
post #11268 of 33692
Reset, and start from the beginning since we are all friends here:
I figured out that my IHS is not flat, and learned from one source that even lapping it does not make it completely flat while being used due to the significant force caused by clamping it down on the mobo. Taking those two pieces of information into account, I wondered if going direct to die would be better simply because you eliminate the interference of the IHS which apparently cannot be made completely flat on the top and bottom while being used and clamped down. Unless, of course you could find a way to lap it while in your mobo.

It is only logical that removing the IHS which is never perfecly flat, and removing one of two layers of TIM would lower your temps. Lowering them how much is the question, and then the other questions are what are the risks of direct die contact and are they worth the temp drop. For his setup, MikeG has answered those questions.

Has anyone else tried direct to die so we could compare results?
If I am being dumb with these questions, that's OK, then I will learn something by figuring that out.
Ivy 5.0
(14 items)
 
Ivy 4.8
(15 items)
 
 
CPUMotherboardGraphicsRAM
i7-3770k 5.0 24/7 Maximus V Formula EVGA GTX 690 G.Skill Trident X 2400 (2x8gb) 
Hard DriveOptical DriveCoolingOS
Samsung 840 Pro 512GB Pioneer BD-RW Swiftech H220 Windows 7 Ultimate 64 bit 
MonitorKeyboardPowerCase
2 x NEC 30" Logitech G15 Corsair AX1200 NZXT Switch 810 
MouseAudio
Logitech G9x Logitech G51 
  hide details  
Reply
Ivy 5.0
(14 items)
 
Ivy 4.8
(15 items)
 
 
CPUMotherboardGraphicsRAM
i7-3770k 5.0 24/7 Maximus V Formula EVGA GTX 690 G.Skill Trident X 2400 (2x8gb) 
Hard DriveOptical DriveCoolingOS
Samsung 840 Pro 512GB Pioneer BD-RW Swiftech H220 Windows 7 Ultimate 64 bit 
MonitorKeyboardPowerCase
2 x NEC 30" Logitech G15 Corsair AX1200 NZXT Switch 810 
MouseAudio
Logitech G9x Logitech G51 
  hide details  
Reply
post #11269 of 33692
I think what he is trying to say is that all things being equal, if the mouting surfaces all contact the same area of die and waterblcok with no air gaps or varying pressures then the IHS removal would make very little if any at all difference and, as a guy who builds race engines for a living and has a pretty solid grasp of conductivity and thermal dynamics, I totally agree. (Assuming the conductivity of the IHS is sufficient and it is not acting as an insulator, which it may well be with that nickel coating it has on it)

However - my limited experience of the IHS is that they are significantly concave / convex so these things will never be equal and in theory, direct die contact from the primary cooler is likely going to produce better thermal conductivity unless you can perfectly lap the die on both sides and even then, does the mechanical strain of clampdown then warp it further? Ha anyone tested it when its installed, as opposed to on a bench?

However, even then, is the DIE itself actually perfectly FLAT? Has anyone checked? I didnt.

Maybe the IHS is supposed to be concave INSIDE to follow the surface of the die, and we are all making it worse. ROFLOL
Maybe being copper its just the thermal cyces that eventually lead to it being concave / convex?
(I left the inside of mine as standard)
Edited by Stu-Crossfire - 2/8/13 at 7:48am
post #11270 of 33692
Quote:
Originally Posted by justanoldman View Post

Reset, and start from the beginning since we are all friends here:
I figured out that my IHS is not flat, and learned from one source that even lapping it does not make it completely flat while being used due to the significant force caused by clamping it down on the mobo. Taking those two pieces of information into account, I wondered if going direct to die would be better simply because you eliminate the interference of the IHS which apparently cannot be made completely flat on the top and bottom while being used and clamped down. Unless, of course you could find a way to lap it while in your mobo.

It is only logical that removing the IHS which is never perfecly flat, and removing one of two layers of TIM would lower your temps. Lowering them how much is the question, and then the other questions are what are the risks of direct die contact and are they worth the temp drop. For his setup, MikeG has answered those questions.

Has anyone else tried direct to die so we could compare results?
If I am being dumb with these questions, that's OK, then I will learn something by figuring that out.

Questions are never dumb and anyone who says they are instead of trying to help shouldn't be on this forum. As for testing im too afraid to damage my chip my temps are great the way they are and its a good over locker
R@WR
(16 items)
 
Rawr V2
(16 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770k Z77 MPower (MS-7751) Asus Radeon R9 290 Corsair Dominator DHX Pro 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 3 SSD Seagate Barracuda LG Super Multi XSPC Raystorm laing ddc 3.25, ex 240 Swiftech X... 
OSMonitorKeyboardPower
Windows 8.1 Pro LG 24inch led Backlit Corsair K60 XFX 850WPro Fully Modular 
CaseMouseMouse PadAudio
Bitfenix Ronin Steel Series Sensei Razer Goliathus Steel Series Siberius V2 
CPUMotherboardGraphicsRAM
3770k Msi z77 mpower EVGA GTX 980 Corsair Dominator 
Hard DriveHard DriveCoolingOS
WD Blue OCZ Vertex 3 SSD Custom Water Cooling, Darkside 6 port reservoir... Windows 8.1 
MonitorKeyboardPowerCase
LG 24 inch 1080P CM Quickfire Rapid, custom keys from WASD XFX Pro Fully modular 850W Gold Corsair Air 540 
MouseMouse PadAudioAudio
Corsair M65 Razer Goliathus Roccat Kave True 5.1 Headset Asus Xonar DX 7.1 Sound card 
  hide details  
Reply
R@WR
(16 items)
 
Rawr V2
(16 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core i7 3770k Z77 MPower (MS-7751) Asus Radeon R9 290 Corsair Dominator DHX Pro 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 3 SSD Seagate Barracuda LG Super Multi XSPC Raystorm laing ddc 3.25, ex 240 Swiftech X... 
OSMonitorKeyboardPower
Windows 8.1 Pro LG 24inch led Backlit Corsair K60 XFX 850WPro Fully Modular 
CaseMouseMouse PadAudio
Bitfenix Ronin Steel Series Sensei Razer Goliathus Steel Series Siberius V2 
CPUMotherboardGraphicsRAM
3770k Msi z77 mpower EVGA GTX 980 Corsair Dominator 
Hard DriveHard DriveCoolingOS
WD Blue OCZ Vertex 3 SSD Custom Water Cooling, Darkside 6 port reservoir... Windows 8.1 
MonitorKeyboardPowerCase
LG 24 inch 1080P CM Quickfire Rapid, custom keys from WASD XFX Pro Fully modular 850W Gold Corsair Air 540 
MouseMouse PadAudioAudio
Corsair M65 Razer Goliathus Roccat Kave True 5.1 Headset Asus Xonar DX 7.1 Sound card 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › [Official] Delidded Club / Guide