Overclock.net › Forums › Graphics Cards › Graphics Cards - General › But, why would higher leakage chips overclock higher on water?
New Posts  All Forums:Forum Nav:

But, why would higher leakage chips overclock higher on water? - Page 2

post #11 of 14
I don't have a source, mine is pure conjecture as well. I like to tell myself it's "educated conjecture", but that doesn't mean it necessarily is lol.
   
Z77 Bench-a-lot
(31 items)
 
CPUMotherboardGraphicsGraphics
i7 3930K @ 4.92Ghz Asus Rampage 4 Extreme EVGA GTX670 FTW 2GB  EVGA GTX780Ti Classified K|ngp|n Edition @ 1483... 
GraphicsRAMHard DriveHard Drive
EVGA GTX780Ti Classified K|ngp|n Edition @ 1483... 16GB G.Skill Ripjaws-Z @ DDR3-2248 9-10-9-26 1T Samsung 830 256GB SSD WD10EZEX 1TB HDD 
Hard DriveHard DriveHard DriveHard Drive
WD10EALX 1TB HDD Seagate Barracuda 7200.14 1TB WD3200AAKS 320GB HDD WD5000HHTZ 10krpm 500GB VelociRaptor 
Optical DriveCoolingCoolingCooling
LG/Hitachi Slim DVD-R/W Custom MB-Backside VRM-sinks (Copper) 3x NZXT RF-FX140LB 1-2krpm 98.3CFM Case Fans 2x Corsair SP120-PE Case Fans 
CoolingCoolingCoolingCooling
Heatkiller GTX680 "Hole Edition" Block + Backplate XSPC Single-Bay Res (Single 5.25 bay) EK Multioption Res X2 - Advanced 150 DD "Fill-Port Res" + Fill-Port Fitting (Red) 
CoolingCoolingCoolingCooling
2x MCP35X + Heatsinks/80x15mm Fan XSPC EX420 Alphacool NeXXos UT60 240 6x Bgears Blasters 140mm 1800rpm Fans 
CoolingCoolingCoolingOS
4x Koolance 120x25mm 2600rpm Fans Comp/Rotary Fittings (BP/XSPC/Monsoon) Primoflex Pro LRT - White (1/2x3/4) Windows 7 Professional 64bit  
MonitorMonitorMonitorMonitor
Dell P2210Hb 21.5" 1920x1080p Viewsonic VA2012WB 20.1" LCD 1680x1050 Dell U3014 Eizo 27" 1440p 10bit RGB-LED Backlit AH-IPS Panel 
MonitorKeyboardPowerCase
55" LG HDTV (their top 2014 model) Ducky Mechanical (Browns) NZXT HALE90 850W + 8 blown-up AX1200/i's NZXT Switch 810 (White) 
MouseMouse PadAudioAudio
R.A.T.3 + MX510 My Desk GRADO RS1i Headphones Grado PS1000 Headphones 
OtherOtherOtherOther
Bitfenix Hydra Pro Fan Controller 5ch 30W/ch Bitfenix Alchemy Individually-Sleeved Extension... NZXT Individually-Sleeved Front Panel Wires (Bl... Phobya Sleeved 1x3pin-to-4x3pin Fan Splitter 
OtherOtherOther
Sleeved Wires w Paracord myself (no Heatshrink) Custom Front and Top Panels w HexxMesh *In Prog... Working on Custom Pedestal Prototype 
  hide details  
Reply
   
Z77 Bench-a-lot
(31 items)
 
CPUMotherboardGraphicsGraphics
i7 3930K @ 4.92Ghz Asus Rampage 4 Extreme EVGA GTX670 FTW 2GB  EVGA GTX780Ti Classified K|ngp|n Edition @ 1483... 
GraphicsRAMHard DriveHard Drive
EVGA GTX780Ti Classified K|ngp|n Edition @ 1483... 16GB G.Skill Ripjaws-Z @ DDR3-2248 9-10-9-26 1T Samsung 830 256GB SSD WD10EZEX 1TB HDD 
Hard DriveHard DriveHard DriveHard Drive
WD10EALX 1TB HDD Seagate Barracuda 7200.14 1TB WD3200AAKS 320GB HDD WD5000HHTZ 10krpm 500GB VelociRaptor 
Optical DriveCoolingCoolingCooling
LG/Hitachi Slim DVD-R/W Custom MB-Backside VRM-sinks (Copper) 3x NZXT RF-FX140LB 1-2krpm 98.3CFM Case Fans 2x Corsair SP120-PE Case Fans 
CoolingCoolingCoolingCooling
Heatkiller GTX680 "Hole Edition" Block + Backplate XSPC Single-Bay Res (Single 5.25 bay) EK Multioption Res X2 - Advanced 150 DD "Fill-Port Res" + Fill-Port Fitting (Red) 
CoolingCoolingCoolingCooling
2x MCP35X + Heatsinks/80x15mm Fan XSPC EX420 Alphacool NeXXos UT60 240 6x Bgears Blasters 140mm 1800rpm Fans 
CoolingCoolingCoolingOS
4x Koolance 120x25mm 2600rpm Fans Comp/Rotary Fittings (BP/XSPC/Monsoon) Primoflex Pro LRT - White (1/2x3/4) Windows 7 Professional 64bit  
MonitorMonitorMonitorMonitor
Dell P2210Hb 21.5" 1920x1080p Viewsonic VA2012WB 20.1" LCD 1680x1050 Dell U3014 Eizo 27" 1440p 10bit RGB-LED Backlit AH-IPS Panel 
MonitorKeyboardPowerCase
55" LG HDTV (their top 2014 model) Ducky Mechanical (Browns) NZXT HALE90 850W + 8 blown-up AX1200/i's NZXT Switch 810 (White) 
MouseMouse PadAudioAudio
R.A.T.3 + MX510 My Desk GRADO RS1i Headphones Grado PS1000 Headphones 
OtherOtherOtherOther
Bitfenix Hydra Pro Fan Controller 5ch 30W/ch Bitfenix Alchemy Individually-Sleeved Extension... NZXT Individually-Sleeved Front Panel Wires (Bl... Phobya Sleeved 1x3pin-to-4x3pin Fan Splitter 
OtherOtherOther
Sleeved Wires w Paracord myself (no Heatshrink) Custom Front and Top Panels w HexxMesh *In Prog... Working on Custom Pedestal Prototype 
  hide details  
Reply
post #12 of 14
Thread Starter 
So I guess we've concluded

Nobody Knows ©QI
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
post #13 of 14
Quote:
Originally Posted by fateswarm View Post

So I guess we've concluded

Nobody Knows ©QI

I'm just waiting for someone to pop in and make the buzzer sound (from QI)

I also don't quite get how the leakage of a chip corresponds to better overclockability using different cooling...

For example, my 7950 has a ASIC of 92%!!, and with the stock cooler I could only push it stable to about 1100/1500. Now that its under water, I have it stable at 1300/1600. One thing I have noticed is voltage drop on the core. I need to push the voltage up from the stock 1130mV to keep it stable, but even if I push 1300mV in Afterburner, under load the core voltage would drop down to 1170mV in GPUz (with 0% power setting). Pushing the power setting up to 20% lets the core voltage drop less, it stays around 1250mv. Also, it seriously burns the VRM's. They shoot up to 80-100C. At the moment I'm trying to keep the power setting below 10% as that's the point where I see the VRM's really start heating up (voltage setting of 1270, drop to 1210mV).

Now from my understanding, the leakage of a chip should have an effect on the voltage drop on the core with full load. Higher leakage = more voltage drop? But in my case, with an ASIC of 92%, I still get huge voltage drops. Could it be that my high ASIC score is causing the big voltage drop, or is it causing the high VRM temps if I try to compensate with more current? Or is the ASIC score only useful in estimating subzero overclocking performance....

Yup......:
Quote:
Originally Posted by fateswarm View Post

So I guess we've concluded

Nobody Knows ©QI
alt.sanity
(28 items)
 
  
CPUMotherboardGraphicsGraphics
i7 4770k Asus ROG Maximus VI Hero AMD R9-290  AMD R9-290 
RAMHard DriveHard DriveHard Drive
Corsair Dominator Platinum 2133MHz 8GB Samsung 840 Pro 256GB OCZ Vector 256GB WD Green 3TB 
Hard DriveOptical DriveCoolingCooling
WD Green 3TB LG BluRay Lamptron FC5 V2 Alphacool NexXxos XT45 Quad 
CoolingCoolingCoolingCooling
Alphacool NexXxos XT45 Triple XSPC Twin D5 Bay Res  XSPC Raystorm CPU Block EK FC7970 CSQ Plexi 
CoolingCoolingOSMonitor
[7] Corsair SP120 Performance [6] Corsair AF120 Performance Windows 7 Ultimate 64bit Samsung S23A700D 
MonitorKeyboardPowerCase
LG IPS235 Corsair K90 CherryMX Red Seasonic Platinum 1000W Corsair 900D 
MouseMouse PadAudioOther
R.A.T. 7 Contagion White Razer Goliathus Extend CM Storm Sirus MDPC-X Sleeving 
  hide details  
Reply
alt.sanity
(28 items)
 
  
CPUMotherboardGraphicsGraphics
i7 4770k Asus ROG Maximus VI Hero AMD R9-290  AMD R9-290 
RAMHard DriveHard DriveHard Drive
Corsair Dominator Platinum 2133MHz 8GB Samsung 840 Pro 256GB OCZ Vector 256GB WD Green 3TB 
Hard DriveOptical DriveCoolingCooling
WD Green 3TB LG BluRay Lamptron FC5 V2 Alphacool NexXxos XT45 Quad 
CoolingCoolingCoolingCooling
Alphacool NexXxos XT45 Triple XSPC Twin D5 Bay Res  XSPC Raystorm CPU Block EK FC7970 CSQ Plexi 
CoolingCoolingOSMonitor
[7] Corsair SP120 Performance [6] Corsair AF120 Performance Windows 7 Ultimate 64bit Samsung S23A700D 
MonitorKeyboardPowerCase
LG IPS235 Corsair K90 CherryMX Red Seasonic Platinum 1000W Corsair 900D 
MouseMouse PadAudioOther
R.A.T. 7 Contagion White Razer Goliathus Extend CM Storm Sirus MDPC-X Sleeving 
  hide details  
Reply
post #14 of 14
Thread Starter 
I have a loose theory about the inconsistencies some people see but it's probably off topic in this thread: It's possible that some people see differences compared to similar setups simply because of very minute differences in sitting of their cooling system on their die. I expect that even a few micrometers of deviation would spike at least parts of a die with different temperatures (e.g. imagine 3/4ths of a die being cooled better than 1/4 of it due to micrometer differences in sitting). That of course might easily explain how on main system CPUs people see enormous differences between cores sometimes.

But it may be off topic in this issue, since most of the water/air deviations here are relatively consistent.
Quote:
Originally Posted by altsanity View Post

I'm just waiting for someone to pop in and make the buzzer sound (from QI)

I also don't quite get how the leakage of a chip corresponds to better overclockability using different cooling...

For example, my 7950 has a ASIC of 92%!!, and with the stock cooler I could only push it stable to about 1100/1500. Now that its under water, I have it stable at 1300/1600. One thing I have noticed is voltage drop on the core. I need to push the voltage up from the stock 1130mV to keep it stable, but even if I push 1300mV in Afterburner, under load the core voltage would drop down to 1170mV in GPUz (with 0% power setting). Pushing the power setting up to 20% lets the core voltage drop less, it stays around 1250mv. Also, it seriously burns the VRM's. They shoot up to 80-100C. At the moment I'm trying to keep the power setting below 10% as that's the point where I see the VRM's really start heating up (voltage setting of 1270, drop to 1210mV).

Now from my understanding, the leakage of a chip should have an effect on the voltage drop on the core with full load. Higher leakage = more voltage drop? But in my case, with an ASIC of 92%, I still get huge voltage drops. Could it be that my high ASIC score is causing the big voltage drop, or is it causing the high VRM temps if I try to compensate with more current? Or is the ASIC score only useful in estimating subzero overclocking performance....

Yup......:

I think you're just getting throttled there by a system trying to save your components from overheating (when voltage drops). Voltage on higher leaking should generally be higher for the same clocks. I believe the ASIC rating was invented for automatic setting of voltage. If the system saw a low ASIC, it would use higher voltage and vise versa. I suspect its hardcoded on initial foundry testing, and it's likely to have inconsistencies with the reality of each die.



The initial text of that quote reminded me of Richard Feynman. Since atoms 'jiggle' and they jiggle less the lower the temperature then it might come to reason that they might need higher voltage to reach the same state with a lower temperature cooling solution. I've no idea if that applies to reality but it's likely.
Edited by fateswarm - 7/31/13 at 4:41am
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
PC
(9 items)
 
  
CPUMotherboardGraphicsRAM
4790K 4.6G 1.24v/1.74v, 5.1G 1.35v validation. GA-Z97X-Gaming 7 Tri-X R9 290 1100/1350 +0.012v G.Skill 2400 c10 
CoolingMonitorPowerCase
Noctua NH-D15 24EA53 IPS 76Hz OC EVGA 1000 G2 Phanteks Enthoo Pro 
Audio
SoundMagic E10 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Graphics Cards - General
Overclock.net › Forums › Graphics Cards › Graphics Cards - General › But, why would higher leakage chips overclock higher on water?