Overclock.net › Forums › Graphics Cards › NVIDIA › ~~The GTX 670 Overclocking Master-Guide~~
New Posts  All Forums:Forum Nav:

~~The GTX 670 Overclocking Master-Guide~~ - Page 227

post #2261 of 3357
Fast question about core clock performance

Do higher clocks increase performance in a lineal way? Let's say I wanted to know what FPS I would be getting if my card could do 100Mhz more on the core, If I downclock my card 100 Mhz and get 3 FPS less, does that mean that If I were to increase them 100 Mhz I would be getting 3 FPS more?
My System
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5-3570k @ 4.5Ghz Asus Maximus V GENE Gigabyte GTX670 OC 2x4GB G.Skill Ripjaws @ 1600 Mhz 
Hard DriveHard DriveCoolingOS
OCZ Vertex 3 120 Gb WesternDigital Velociraptor 300Gb Noctua NH-C14 Windows 7 x64 
MonitorMonitorKeyboardPower
HP ZR24W  Dell Ultrasharp P1911 SteelSeries 6Gv2 SeaSonic X-760W 
CaseMouseMouse PadAudio
Silverstone TJ08-E SteelSeries Sensei RAW SteelSeries 4HD AKG K701 + FiiO E7/E9 
  hide details  
Reply
My System
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5-3570k @ 4.5Ghz Asus Maximus V GENE Gigabyte GTX670 OC 2x4GB G.Skill Ripjaws @ 1600 Mhz 
Hard DriveHard DriveCoolingOS
OCZ Vertex 3 120 Gb WesternDigital Velociraptor 300Gb Noctua NH-C14 Windows 7 x64 
MonitorMonitorKeyboardPower
HP ZR24W  Dell Ultrasharp P1911 SteelSeries 6Gv2 SeaSonic X-760W 
CaseMouseMouse PadAudio
Silverstone TJ08-E SteelSeries Sensei RAW SteelSeries 4HD AKG K701 + FiiO E7/E9 
  hide details  
Reply
post #2262 of 3357
Quote:
Originally Posted by loneknives View Post

Finally got a good and stable score ^^
thanks for the awesome guide

http://www.3dmark.com/3dm11/6355837

modded my galaxy gtx 670 gc bios, and install Accelero hybrid cooler

1.187V 1306mhz max load temp 50C

I would say that go for big.. as you have lots of headroom it seems in that way.. go for bigger wink.gif
The Beast
(19 items)
 
  
CPUMotherboardGraphicsRAM
INTEL i5 3570K (4.4GHz) GIGABYTE Z77-D3H MSI GAMING X GTX1060 6GB (2152/9400MHz) CRUCIAL BALLISTIX TACTICAL LP 2x8GB (2400MHz | ... 
Hard DriveHard DriveOptical DriveCooling
SAMSUNG 840 PRO SSD 256GB SAMSUNG HD103SJ 1TB LG DVD-RW 22X COOLERMASTER HYPER 212X 
CoolingOSMonitorKeyboard
3x AKASA APACHE 120MM WINDOWS 10 PRO 64-bit SAMSUNG 23" 2333T C-PVA DVI FULL HD + Extra 19"... A4 TECH X7 G800 
PowerCaseMouseMouse Pad
OCZ FATAL1TY 550W AEROCOOL VS-3 ADVANCE NPET V30 IR OPTICAL MOUSE NOVA WINNER II 
AudioOtherOther
ASUS XONAR U7 CREATIVE GIGAWORKS G500 AUDIO TECHNICA ATH-A500X Audiophile Monitor 
  hide details  
Reply
The Beast
(19 items)
 
  
CPUMotherboardGraphicsRAM
INTEL i5 3570K (4.4GHz) GIGABYTE Z77-D3H MSI GAMING X GTX1060 6GB (2152/9400MHz) CRUCIAL BALLISTIX TACTICAL LP 2x8GB (2400MHz | ... 
Hard DriveHard DriveOptical DriveCooling
SAMSUNG 840 PRO SSD 256GB SAMSUNG HD103SJ 1TB LG DVD-RW 22X COOLERMASTER HYPER 212X 
CoolingOSMonitorKeyboard
3x AKASA APACHE 120MM WINDOWS 10 PRO 64-bit SAMSUNG 23" 2333T C-PVA DVI FULL HD + Extra 19"... A4 TECH X7 G800 
PowerCaseMouseMouse Pad
OCZ FATAL1TY 550W AEROCOOL VS-3 ADVANCE NPET V30 IR OPTICAL MOUSE NOVA WINNER II 
AudioOtherOther
ASUS XONAR U7 CREATIVE GIGAWORKS G500 AUDIO TECHNICA ATH-A500X Audiophile Monitor 
  hide details  
Reply
post #2263 of 3357
Quote:
Originally Posted by mastercommander View Post

Well sorta broke in this thread but my 7950 at 1200mhz gets 10800p score and max temp at 67* with twinfrozr 3 and for less money than a Gtx 670 biggrin.gif
and your point is what? everyone knows amd wins benchmarks with new drivers.
i have physx. the only people who say physx aint squat are those running amd tongue.gif
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  
Reply
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  
Reply
post #2264 of 3357
Here's my best 3DMark11 score:
http://www.3dmark.com/3dm11/6367698
core@1306, mem@3442 (with 1.21V)
post #2265 of 3357
Quote:
Originally Posted by kevindd992002 View Post

Are you sure that MemtestCL doesn't produce any false positives?

It doesn't. However I posted this before actually getting my GTX 670 ;-) 'Unfortunately' GDDR5 has error correction which kind of defeats the purpose of using memtestcl. It may seem stable at a certain clock speed while producing a ton of memory errors which the memory would correct... up to a point though. These are errors which you don't see and neither will memtestcl. So more can actually be less as error correction also reduces the speed (each error results in a repeated read), however from my own testing performance was always higher with higher mem clocks. The key problem I saw is that it may seem stable at +500 offset after 3 hours of Haven runs/folding, and then all of a sudden (meaning 2 days later while you're playing your favorite game) it would get a burst of errors which won't be corrected and you'll get artifacts/crash. Determining truly stable GDDR5 overclocks turns out a super PITA...
post #2266 of 3357
Now I noticed something weird a short while ago, and want to try to solve before I start flashing my bios. Was seeing how well the 670 I received from rma would oc, noticed that while core and men would reflect changed values that the gpu usage stayed at 0 for that card. Also noticed that sometimes the core would not go past 1006, it down clocked itself at the start of valley. Ya think my board is the issue? Sli is enabled, tried to run bioshock to see if I got a similar result and it was really really choppy
SKYnet
(17 items)
 
  
CPUMotherboardGraphicsGraphics
Intel I7 4930K @ 4.6ghz ASUS Rampage IV Extreme EVGA GTX 970 EVGA GTX 970 
RAMHard DriveHard DriveOptical Drive
G. Skill Ripjaws X 4x4gb 1.5tb WD Caviar Green SATA Samsung 840 EVO 250gb SSD HP USB DVD 
CoolingCoolingOSMonitor
Antec Kuhler 650 ThermalRight True Spirit 140 Power Windows 10 Ultimate 64 bit Samsung 240 HD TOC 
KeyboardPowerCaseMouse
Razor Lycosa PC Power and Cooling Silencer 910 NZXT Switch 810 Matte Black It clicks.............. 
Mouse Pad
IKEA $1 pad 
  hide details  
Reply
SKYnet
(17 items)
 
  
CPUMotherboardGraphicsGraphics
Intel I7 4930K @ 4.6ghz ASUS Rampage IV Extreme EVGA GTX 970 EVGA GTX 970 
RAMHard DriveHard DriveOptical Drive
G. Skill Ripjaws X 4x4gb 1.5tb WD Caviar Green SATA Samsung 840 EVO 250gb SSD HP USB DVD 
CoolingCoolingOSMonitor
Antec Kuhler 650 ThermalRight True Spirit 140 Power Windows 10 Ultimate 64 bit Samsung 240 HD TOC 
KeyboardPowerCaseMouse
Razor Lycosa PC Power and Cooling Silencer 910 NZXT Switch 810 Matte Black It clicks.............. 
Mouse Pad
IKEA $1 pad 
  hide details  
Reply
post #2267 of 3357
Quote:
Originally Posted by Br3ach View Post

It doesn't. However I posted this before actually getting my GTX 670 ;-) 'Unfortunately' GDDR5 has error correction which kind of defeats the purpose of using memtestcl. It may seem stable at a certain clock speed while producing a ton of memory errors which the memory would correct... up to a point though. These are errors which you don't see and neither will memtestcl. So more can actually be less as error correction also reduces the speed (each error results in a repeated read), however from my own testing performance was always higher with higher mem clocks. The key problem I saw is that it may seem stable at +500 offset after 3 hours of Haven runs/folding, and then all of a sudden (meaning 2 days later while you're playing your favorite game) it would get a burst of errors which won't be corrected and you'll get artifacts/crash. Determining truly stable GDDR5 overclocks turns out a super PITA...

That's where benchmarks come into place. When overclocking your card's memory, sometimes less overclock produces a higher benchmark score than a higher overclock. That is a proof that error correction is working.
post #2268 of 3357
Quote:
Originally Posted by kevindd992002 View Post

Not really sure on that but why don't you want to use the shroud?

Not sure if it would fit on the bottom card since I a using a PCI sound card.. but for the cost of 2 of these I might go with a full watercooling instead biggrin.gif
Nexus
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 4790K Gigabyte Z87 G1 Sniper M5  2x Zotac 980Ti (Reference Design) Corsair Dominator Platinum 1866Mhz 4x4Gb (16Gb)... 
Hard DriveHard DriveHard DriveHard Drive
Crucial M550 256Gb Corsair Force GT 240Gb Seagate 3TB 2x Samsung Spinpoint F3 1TB 
CoolingOSMonitorKeyboard
Swiftec H220X w/ GT-AP15  Windows 8.1 Pro x64 Samsung 32EH5000 32" HD Direct lit LED TV Ducky Shine 3 [Cherry MX Blue , White LED] 
PowerCaseMouseMouse Pad
Coolermaster V1000 Corsair 350D Logitech G603 Puretrack Talent 
Audio
Logitech Z5500 5.1 Digital Speaker 
  hide details  
Reply
Nexus
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 4790K Gigabyte Z87 G1 Sniper M5  2x Zotac 980Ti (Reference Design) Corsair Dominator Platinum 1866Mhz 4x4Gb (16Gb)... 
Hard DriveHard DriveHard DriveHard Drive
Crucial M550 256Gb Corsair Force GT 240Gb Seagate 3TB 2x Samsung Spinpoint F3 1TB 
CoolingOSMonitorKeyboard
Swiftec H220X w/ GT-AP15  Windows 8.1 Pro x64 Samsung 32EH5000 32" HD Direct lit LED TV Ducky Shine 3 [Cherry MX Blue , White LED] 
PowerCaseMouseMouse Pad
Coolermaster V1000 Corsair 350D Logitech G603 Puretrack Talent 
Audio
Logitech Z5500 5.1 Digital Speaker 
  hide details  
Reply
post #2269 of 3357
Theoretically yes, but as I mentioned based on my testing higher mem clock speeds always resulted in higher performance, see below:
Code:
Mem Offset   FPS     Score   Notes
0       39.8    1666    -
50      40      1676    -
100     40.5    1695    -
150     40.9    1710    -
200     41.2    1725    Artifacts
250     41.5    1738    -
300     41.9    1752    Artifacts
350     42.2    1767    Artifacts
400     42.3    1772    -
450     42.8    1791    Artifacts
500     43.1    1801    -
550     43.3    1814    -
600     24.1    1006    Artifacts then FPS drop
post #2270 of 3357
What do you guys consider a safe power target %?

I got my modded BIOS at it can go up to 150% but my guess is that it's way too much.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › ~~The GTX 670 Overclocking Master-Guide~~