Overclock.net › Forums › Graphics Cards › NVIDIA › [Official] GTX 580 Overclocking Club
New Posts  All Forums:Forum Nav:

[Official] GTX 580 Overclocking Club - Page 273

post #2721 of 3582
I'm not going to jump into the 2-3 page debade about the cards but something that hasn't been mentioned are stock volts. Of course the cards which are clocked higher from factory are at those levels if their volts are higher than normal. I don't own eVGA or SOC so don't know those volts but from reading around they appear to be high (>1.05v)?

add: also remember that the Lightning came in 2 forms (at least). The 'normal' one for US market at least with the 832 clocks and the 'other' one for Australia and elsewhere where it was set to stock clocks at 772 on core. When I asked MSI they explained there was no binning and it was simply a case that those regions requested that model of the card (btw: there is no model number change etc, just the firmware was different). Updating the 832 firmware into those cards worked fine.


On the topic of ASIC score, that's another oddity. First people were saying you want it to be higher for better OC ability then few said you want it to be lower with explanation/reason. Now if you look around you find cards that are over 100% which of course doesn't make any sense. It was said to be a number that nvidia and amd(?) puts there and the program simply reads it, but if that's the case then what about all the comments saying the next version of gpu-z will have revised it so there won't be cards showing >100% anymore?
Edited by error-id10t - 1/31/12 at 2:43pm
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
Just a 'puter
(18 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 6700K ASUS Maximus VIII Hero iGPU thanks to Nvidia 970 3.5GB Corsair Vengeance LPX CMK8GX4M2B3200C16 
Hard DriveHard DriveCoolingCooling
Samsung 950 PRO 512GB Crucial M4 256GB RX360 V3 Koolance CPU-380I 
CoolingCoolingOSMonitor
Koolance RP-452X2 Reservoir Koolance PMP-450 12V Variable Speed Pump Windows 10 Pro x64 benq XL2411Z 
KeyboardPowerCaseMouse
logitech G510 Seasonic X-760 Corsair 800D Mionix Castor 
AudioAudio
Creative Sound Blaster X-Fi Titanium HD DT 990 Premium 250Ohm 
  hide details  
Reply
post #2722 of 3582
Quote:
"The next new feature is ASIC quality, designed for NVIDIA Fermi (GF10x and GF11x GPUs) and AMD Southern Islands (HD 7800 series and above), aimed at advanced users, hardware manufacturers, and the likes. We've found the ways in which AMD and NVIDIA segregate their freshly-made GPU ASICs based on the electrical leakages the chips produce (to increase yield by allotting them in different SKUs and performance bins), and we've found ways in which ASIC quality can be quantified and displayed. Find this feature in the context menu of GPU-Z. We're working on implementing this feature on older AMD Radeon GPUs. " http://www.techpowerup.com/159098/TechPowerUp-GPU-Z-0.5.8-Released.html

"high leakage low VID"
Quote:
Originally Posted by Blameless View Post

It's true, to an extent.

I don't know too much of the details behind it but higher leakage parts tend to be able to handle more voltage and current (clock speed) before hitting their limits, but at the same time they run hotter, which under ambient cooling may limit them to lower clocks than their more efficient counterparts.

AMDs TWKR CPUs were and example of chips binned to do well at high voltages and low temperatures.
Quote:
What makes these chips special is the fact that they are considered 'high leakage' parts. In this context, high leakage means that the processors 'leak' more current than a normal Phenom II CPU and consequently run hotter, which actually turns out to be a positive for the overclocking crowd (even if it means they're more likely to die an early death). They can take on more voltage and generally tend to scale higher when cooled with extreme methods, such as liquid nitrogen or liquid helium.” http://www.tomshardware.com/news/amd-phenom-twkr-contest-overclocking,8166.html

ASIC-SoC-VLSI Design
Quote:
Development of the digital integrated circuits is challenged by higher power consumption. The combination of higher clock speeds, greater functional integration, and smaller process geometries has contributed to significant growth in power density. At 90 nm and below, leakage power management is essential in the ASIC design process. As voltages scale downward with the geometries threshold voltages must also decrease to gain the performance advantages of the new technology but leakage current increases exponentially. Thinner gate oxides have led to an increase in gate leakage current.http://asic-soc.blogspot.com.au/2008/03/leakage-power-trends.html

AMD’s Dave Baumann says:
Quote:
We scale the voltage based on leakage, so the higher leakage parts use lower voltage and the lower leakage parts use a higher voltage – what this is does narrow the entire TDP range of the product.

Everything is qualified at worst case anyway; all the TDP calcs and the fan settings are completed on the wors case for the product range. http://forum.beyond3d.com/showpost.php?p=1344008&postcount=29

That's what the web says anyway.
Edited by alancsalt - 1/31/12 at 3:54pm
Nyarlathotep
(17 items)
 
Cxaxukluth
(19 items)
 
Yog-Sothoth
(10 items)
 
CPUMotherboardGraphicsRAM
Intel Core i7 3970X ASUS X79 ROG Rampage IV Extreme Bios 2105 EVGA KingPin Classified 780 Ti 3GB Tri-SLI  G.Skill Trident X F3-2400C10Q-16GTX 16GB (... 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 4 SSD Seagate Barracuda 1TB ST31000524AS Lite-On IHAS124 24X DVDRW 3 x XSPC Rasa 750 RX360 Universal CPU Water Coo... 
CoolingOSMonitorPower
3 x EK-FC780 GTX Classy - Nickel Windows 7 Professional x64 3 x Dell U2412M SilverStone 1500W V2.2 x 2 
Case
Lian Li Pitstop T60 Testbench 
CPUMotherboardGraphicsRAM
i7 - 5960X Asus Rampage Extreme V Gigabyte Titan X 12GB G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... 
Hard DriveHard DriveHard DriveOptical Drive
Samsung 850 Pro Series 1TB SSD 3 x Seagate 2TB  Seagate 3TB LG 
CoolingCoolingOSMonitor
Phobya Xtreme 1260 Radiator XSPC D5 Photon 170 Reservoir/Pump Combo Microsoft Windows 7 Professional 64bit with SP1  Samsung S34E790C 34in Ultrawide Curved LED Mo... 
KeyboardPowerCaseMouse
Microsoft Wireless 3000 Silverstone ST1500 Antec Eleven Hundred Microsoft Wireless 3000 
AudioAudioAudio
Audioengine D1 Premium 24-Bit DAC Audioengine 5+ Premium Powered Speakers Black Audioengine S8 Black 
CPUMotherboardGraphicsRAM
i7 3960X Asus Rampage IV Extreme Gainward "GOOD" GTX 580 x 2 AData x 2 
Hard DriveCoolingCoolingOS
Seagate XSPC Rasa EX 360 crossflow kit x 2, XSPC Rasa E... EK GW-580 Waterblocks Win 7 Pro 
PowerCase
SilverStone Strider ST1500 InWin D Frame 
  hide details  
Reply
Nyarlathotep
(17 items)
 
Cxaxukluth
(19 items)
 
Yog-Sothoth
(10 items)
 
CPUMotherboardGraphicsRAM
Intel Core i7 3970X ASUS X79 ROG Rampage IV Extreme Bios 2105 EVGA KingPin Classified 780 Ti 3GB Tri-SLI  G.Skill Trident X F3-2400C10Q-16GTX 16GB (... 
Hard DriveHard DriveOptical DriveCooling
OCZ Vertex 4 SSD Seagate Barracuda 1TB ST31000524AS Lite-On IHAS124 24X DVDRW 3 x XSPC Rasa 750 RX360 Universal CPU Water Coo... 
CoolingOSMonitorPower
3 x EK-FC780 GTX Classy - Nickel Windows 7 Professional x64 3 x Dell U2412M SilverStone 1500W V2.2 x 2 
Case
Lian Li Pitstop T60 Testbench 
CPUMotherboardGraphicsRAM
i7 - 5960X Asus Rampage Extreme V Gigabyte Titan X 12GB G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... 
Hard DriveHard DriveHard DriveOptical Drive
Samsung 850 Pro Series 1TB SSD 3 x Seagate 2TB  Seagate 3TB LG 
CoolingCoolingOSMonitor
Phobya Xtreme 1260 Radiator XSPC D5 Photon 170 Reservoir/Pump Combo Microsoft Windows 7 Professional 64bit with SP1  Samsung S34E790C 34in Ultrawide Curved LED Mo... 
KeyboardPowerCaseMouse
Microsoft Wireless 3000 Silverstone ST1500 Antec Eleven Hundred Microsoft Wireless 3000 
AudioAudioAudio
Audioengine D1 Premium 24-Bit DAC Audioengine 5+ Premium Powered Speakers Black Audioengine S8 Black 
CPUMotherboardGraphicsRAM
i7 3960X Asus Rampage IV Extreme Gainward "GOOD" GTX 580 x 2 AData x 2 
Hard DriveCoolingCoolingOS
Seagate XSPC Rasa EX 360 crossflow kit x 2, XSPC Rasa E... EK GW-580 Waterblocks Win 7 Pro 
PowerCase
SilverStone Strider ST1500 InWin D Frame 
  hide details  
Reply
post #2723 of 3582

So someone asked about ASIC quality and my OC, well here it is.

 

Note that I have gone as high as 959 MHz core 1918 MHz shader, 2155 MHz (1077.5 as measured in GPU-Z) @ 1.150V in 3DMark11, Heaven, and gaming, but it's not stable for 24/7 folding on -advmethods there.

 

What you're seeing in my screenshot is my second monitor (and a small portion of my primary monitor) showing my 24/7 folding stable 900 MHz core 1800 MHz shader, 2106 MHz (1053 in GPU-Z) @ 1.113V along with my maxed out fan speeds while folding LOL. I've been playing with VirtualBox but I'm annoyed with how it's causing my GPU TPF to fluctuate, so I'm going back to the slower windows SMP client. (I usually get 3:44 or 3:45 TPF and with the Linux client going and this website open it's 3:49 :( )

 

Tangent alert!

 

 

Ok back to the ASIC at 80.9%

 

New Bitmap Image.jpg

 

 

The card is a reference hardware but factory overclocked EVGA GTX 580 Superclocked.


 

Quote:
Originally Posted by error-id10t View Post

I'm not going to jump into the 2-3 page debade about the cards but something that hasn't been mentioned are stock volts. Of course the cards which are clocked higher from factory are at those levels if their volts are higher than normal. I don't own eVGA or SOC so don't know those volts but from reading around they appear to be high (>1.05v)?


 

You asked aboute stock voltages. This is my evga Superclocked showing the stock settings in the [brackets] and the overclocked settings bold-faced.

 

NVIDIA GTX 580 reference is 772 MHz core, 1544 MHz shader, 2004 MHz memory, and voltage varies by ASIC apparently.

 

EVGA GTX 580 Superclock (in a reference hardware configuration) is 797 MHz core, 1594 MHz shader, and 2025 MHz memory, and my voltage just happened to be 1.063V from the factory at stock settings.

 

nvidia_20120131_234459.png

 

As in the post above, I've been as high as 959/1918/2155 @ 1.150V but for folding 24/7 I'm at 900/1800/2106 @ 1.113V  (I swear, I used to fold at 1.150V at these frequencies but apparently I've been at 1.113V for the last month+... Oh well, stable is stable!)

 

Please use the edit button instead of double posting.

 

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #2724 of 3582
Always wondered why evga picked 797 and not just 800. 800 sounds much better.
550D
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-2700K Gigabyte Z77X-UD3H EVGA GTX 970 ACX 2.0+ SSC 4 x 4GB Samsung Green 
Hard DriveHard DriveCoolingOS
Crucial 240gb Samsung EVO 250gb Corsair H80 Win7 Ultimate 64 
MonitorKeyboardPowerCase
Qnix 2560x1440p Quickfire TK MX Brown XFX 750w Black Edition Corsair 550D 
MouseMouse Pad
Lefty Deathadder X Trac Ripper XXL 
  hide details  
Reply
550D
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-2700K Gigabyte Z77X-UD3H EVGA GTX 970 ACX 2.0+ SSC 4 x 4GB Samsung Green 
Hard DriveHard DriveCoolingOS
Crucial 240gb Samsung EVO 250gb Corsair H80 Win7 Ultimate 64 
MonitorKeyboardPowerCase
Qnix 2560x1440p Quickfire TK MX Brown XFX 750w Black Edition Corsair 550D 
MouseMouse Pad
Lefty Deathadder X Trac Ripper XXL 
  hide details  
Reply
post #2725 of 3582
Quote:
Originally Posted by lightsout View Post

Always wondered why evga picked 797 and not just 800. 800 sounds much better.

 

 

Cards would not clock like this before, there were certain settings, and odd increments, like it would set your clock to a number different than what you input, so if the straps were 605 - 612 - 619 - 625 and you manually set 615 it might read 612, but if you set it to 616 it would jump to 619. Obviously these are just off the top of my head and not actuall settings I remember, but you get the point.

 

I wonder if the software now just reads what you set it to, even if the card sets it to the closest default...  Meaning if you set your card to 800, its actually running at 797...

 

Also the strap could be on the shader, meaning that clock wont run at 1600, it will run at 1594. or it will run at 1620 and with the core/shader linked thats what it comes to......

 

 

Just my thoughts tho, no idea how or why they go about deciding a OC setting...

 

They all seem to do it tho, the lightning is 832 why not 830 or 835.....  

 

The Money Pit
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-950 - 4.24 @ 1.40v Batch #3941B089 EVGA X58 SLI LE MSI 580 GTX Lightning - 950/1900/2200 @ 1.1v 6GB Mushkin @ 1475 - 6-8-6-24 
Hard DriveHard DriveOptical DriveCooling
120GB Vertex Plus 2 x 500GB WD Blues RAID0 Lite-On DVD/CD Custom H20 Loop 
OSMonitorPowerCase
Win 7 Ulti-64 25" I-Inc iH253 Antec TruePower New TP-750 CoolerMaster HAF 932 
  hide details  
Reply
The Money Pit
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-950 - 4.24 @ 1.40v Batch #3941B089 EVGA X58 SLI LE MSI 580 GTX Lightning - 950/1900/2200 @ 1.1v 6GB Mushkin @ 1475 - 6-8-6-24 
Hard DriveHard DriveOptical DriveCooling
120GB Vertex Plus 2 x 500GB WD Blues RAID0 Lite-On DVD/CD Custom H20 Loop 
OSMonitorPowerCase
Win 7 Ulti-64 25" I-Inc iH253 Antec TruePower New TP-750 CoolerMaster HAF 932 
  hide details  
Reply
post #2726 of 3582
Quote:
Originally Posted by 10acjed View Post



Quote:
Originally Posted by lightsout View Post

Always wondered why evga picked 797 and not just 800. 800 sounds much better.

 

 

Cards would not clock like this before, there were certain settings, and odd increments, like it would set your clock to a number different than what you input, so if the straps were 605 - 612 - 619 - 625 and you manually set 615 it might read 612, but if you set it to 616 it would jump to 619. Obviously these are just off the top of my head and not actuall settings I remember, but you get the point.

 

I wonder if the software now just reads what you set it to, even if the card sets it to the closest default...  Meaning if you set your card to 800, its actually running at 797...

 

Also the strap could be on the shader, meaning that clock wont run at 1600, it will run at 1594. or it will run at 1620 and with the core/shader linked thats what it comes to......

 

 

Just my thoughts tho, no idea how or why they go about deciding a OC setting...

 

They all seem to do it tho, the lightning is 832 why not 830 or 835.....  

 



It has to do with the base clock frequency generated by whatever clockgen circuit they're using and then the multiplier they're using to generate the final clock for the purpose. I don't know enough about the GF 110 architecture to tell you exactly why this is true, but I can tell you that there are, indeed, steps and jumps.

 

Here is an example of the "step" pattern on the Core clocks starting at the 772/1544 reference core (for the 580) and moving up to where my card crashes running even a 256x256 OC Scanner (furmark). You're pretty unlikely to have a card stable at a 2025 MHz shader clock anyhow, unless you bought one of the hand-picked Ultra Classifieds, Lightning Extremes, or other high-end flagship LN2 capable cards and are water cooling it.

 

772 / 1544

773 / 1546

776 / 1552

I hid some of these intermediary clocks where nothing interesting happened in a spoiler, but if you want the details, you can click here. (Click to show)

 

779 / 1558

781 / 1562

783 / 1566

784.5 / 1569

788.5 / 1575

790 / 1580

792.5 / 1583

793 / 1586

794.5 / 1589

 

 

795.5 / 1591

796.5 / 1593

797.5 / 1595 (Oddly, when I played through all of the settings, it wouldn't "stick" on the actual default value of my card, which is 797 / 1594  haha)

810 / 1620 (notice the large jump here from my factory OC settings to the next step up?)

822.5 / 1645 (another big jump)

823.5 / 1647

Saving space! (Click to show)

 

824.5 / 1649

826.5 / 1651

827 / 1654

828.5 / 1657

 

 

830 / 1660

832.5 / 1665 (which skips past the 832 classified setting)

835.5 / 1671

837 / 1674

Again, I hid some of the intermediate values where nothing really interesting happened but feel free to click to see them in detail. (Click to show)

 

839 / 1678

841 / 1682

844 / 1688

847 / 1694

848 / 1696

850.5 / 1701

853.5 / 1707

855 / 1710

856.5 / 1713

860.5 / 1721

864 / 1728

865 / 1730

868 / 1736

871 / 1742

872.5 / 1745

873.5 / 1747

 

 

877.5 / 1755

882.5 / 1765  (For example, when I had it at 1760, it was really at 1755.... When I switched it to 1761 it jumped up to 1765)

883.5 / 1767 (And when I set it to 1766, it jumped right to here at 1767)

886 / 1772

Hidden to save space. (Click to show)

 

888 / 1776 (cool year!)

891 / 1782

894.5 / 1789

897 / 1794

 

 

898.5 / 1797

900 / 1800 (This is where I'm 24/7 folding stable at)

902 / 1804

903.5 / 1807

 

904.5 / 1809 (EDIT: I've since moved up to this spot for my 24/7 folding, seem to be just as stable, have about 1% better performance and lower temperatures here than I had at 900 MHz. I can't explain why my temperatures are lower at a higher clock, but they are. Maybe it has something to do with the VRM finding certain frequency ranges more difficult to regulate? Maybe these tight bands of clock speeds that precede a big jump in clock speed at the next step increase are the ideal operating points?)

 

911 / 1822 (Notice the big jump again? I wonder if there's a correlation between the jump just after factory settings to OC, and this jump from my nicely stable 24/7 OC and the settings which begin to cause issues when I fold?)

918 / 1836 (Another big jump, and folding errors on -advmethods are confirmed to occur at this clock speed regardless of what voltage I choose.)

919 / 1838

920.5 / 1841

More clock speeds hidden to save space. (Click to show)

 

922.5 / 1845

924 / 1848

925.5 / 1851

928 / 1856

931.5 / 1863

934.5 / 1869

 

 

936.5 / 1873

939 / 1878

940 / 1880

945 / 1890 (this is another medium sized jump spot, and also right about where I started to see a very tiny chance of artifacts once in a while during someof the benchmarks.)

949 / 1898

Saving space! (Click to show)

 

950 / 1900

952 / 1904

954.5 / 1909

957.5 / 1915

 

 

958.5 / 1917

962 / 1924 (This was the highest setting I was able to get to pass 3DMark11 and Heaven without crashing. Anything higher would crash me, and I also saw some random artifacts in Heaven occuring more frequently. They appeared as small, bright, and golden bloom sunbursts, where there shouldn't be sunbursts.)

966 / 1932

967.5 / 1935

969 / 1938

More clock speeds hidden to save space. (Click to show)

 

972 / 1944

974.5 / 1949

975.5 / 1951

979 / 1958

981.5 / 1963

983.5 / 1967

985.5 / 1971

987 / 1974

990 / 1980

992 / 1984

994 / 1988

995.5 / 1991

997 / 1994

 

 

998 / 1996

999 / 1998

1000 / 2000

1012.5 / 2025  (This was another big jump in core frequency from the last "step" and I can't help but wonder if these big, seemingly random, steps have something to do with stability. I started getting desktop crashes at this speed; they were driver crashes while I was just sitting in a tiny 256x256 OC scanner, which is like furmark, that doesn't really stress the video card much. These clock speeds were the absolute breaking point of stability, where even at 61C, which is a very mild temperature for these cards, I started experiencing major issues.)

 

 

I'm not going to take the time to run this same test for the memory, but you get the idea; you can see that the memory would jump in similar seemingly random patterns.

 

Again, I can't help but wonder if the places just prior to these jumps are "stability spots" and just after the jump is where problems occur. EVGA set my card from the factory at a spot just prior to a big jump. My 24/7 folding overclock is just prior to a big jump. My best benchmarks and the places where I was gaming and benchmark stable, but not folding stable, were just prior to a big jump. And the point at which the card crashes is just prior to a big jump.


Edited by shad0wfax - 2/2/12 at 10:35am
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #2727 of 3582

 

 

Anyone elsebenching on air notice that their card crashing at a certain temp?

 

When I run 950/1900 and bench my temps will get up around 85 max. However when I try to go to 960 - 975 and run a bench I get a driver crash, when it recovers I notice that my max temp was 80c when it crashed, and this has happened at various different core freq and voltages but the temp seems to always peak at 80c and then crash...

 

The Money Pit
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-950 - 4.24 @ 1.40v Batch #3941B089 EVGA X58 SLI LE MSI 580 GTX Lightning - 950/1900/2200 @ 1.1v 6GB Mushkin @ 1475 - 6-8-6-24 
Hard DriveHard DriveOptical DriveCooling
120GB Vertex Plus 2 x 500GB WD Blues RAID0 Lite-On DVD/CD Custom H20 Loop 
OSMonitorPowerCase
Win 7 Ulti-64 25" I-Inc iH253 Antec TruePower New TP-750 CoolerMaster HAF 932 
  hide details  
Reply
The Money Pit
(15 items)
 
  
CPUMotherboardGraphicsRAM
i7-950 - 4.24 @ 1.40v Batch #3941B089 EVGA X58 SLI LE MSI 580 GTX Lightning - 950/1900/2200 @ 1.1v 6GB Mushkin @ 1475 - 6-8-6-24 
Hard DriveHard DriveOptical DriveCooling
120GB Vertex Plus 2 x 500GB WD Blues RAID0 Lite-On DVD/CD Custom H20 Loop 
OSMonitorPowerCase
Win 7 Ulti-64 25" I-Inc iH253 Antec TruePower New TP-750 CoolerMaster HAF 932 
  hide details  
Reply
post #2728 of 3582
Quote:
Originally Posted by 10acjed View Post

 

 

Anyone elsebenching on air notice that their card crashing at a certain temp?

 

When I run 950/1900 and bench my temps will get up around 85 max. However when I try to go to 960 - 975 and run a bench I get a driver crash, when it recovers I notice that my max temp was 80c when it crashed, and this has happened at various different core freq and voltages but the temp seems to always peak at 80c and then crash...

 


Yes, this does seem to be a factor in stability but in my specific case, I'm unsure if it's due to the VRMs heating up (most likely), the fans cycling rapidly (5 second intervals) between <80%, 85%, and 100% which causes strange Amperage spikes and may be depriving the core or the VRMs of reliable clean power, or if it was purely temperature related.

 

All three are possible, but I think that VRM temperature causing drift of voltage and current to the core or current fluctuations due to the fan cycling rapidly and cascading down to the VRM, causing poor core voltage supply would be the most likely culprits.

 

I did all of my benching with the fans manually set to 100% (after flashing my BIOS to the latest fanspeed unlock firmware from my manufacturer). Once I found a setting that I was "stable" at, I would then reverse engineer a fan speed profile that kept me below 80C and I never had heat related issues.

 

However, when I began folding 24/7 on GPU -advmethods, which is about the most stressful thing you can possibly do to a Fermi card, I noticed that I was having some stability issues regardless of what voltage I chose at the higher temperatures. Keeping my core temperatures below 80C with a custom fan speed profile did seem to help a great deal and I'm 24/7 stable. If I allow the fan speed profile to revert to hardware controlled, my card will be nice and quiet, heat itself up to 85C, the fans will kick on at 85%, the card will heat itself up to 95C, the fans will cycle to 100%, the card will begin to cool down to 80C, the fans will kick down to 75% or so, and the rapid cycling of the fan from low to high speeds will continue. This cycling of temperatures (or perhaps the Amperage cycles) caused stability issues in the folding.

 

So, long story short, yes, temperature seems to be an issue and I've found that < 80C is the best place to be. My errors occurred at 85C and above.

B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
B2
(8 items)
 
  
CPUMotherboardGraphicsGraphics
Intel Core i7-5930K Haswell-E 6-Core 3.5GHz LGA... EVGA X99 Classified (151-HE-E999-KR) LGA 2011-v... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... EVGA 04G-P4-3975-KR GeForce GTX 970 SSC ACX 2.0... 
RAMHard DriveCoolingCase
G.SKILL Ripjaws 4 series 16GB (4 x 4GB) 288-Pin... SAMSUNG 850 EVO MZ-75E1T0B/AM 2.5" 1TB SATA III... CORSAIR Hydro Series H100i Extreme Performance ... Phanteks Enthoo Luxe Series PH-ES614L_BK Black ... 
  hide details  
Reply
post #2729 of 3582
What are you guys seeing for ASIC Quality with the new version of GPU-Z?

Here is mine:

528
post #2730 of 3582
Post it in another thread as well but what the heck...

495

Maximum stable OC is 950Mhz/2300Mhz on air (default is 782/2002Mhz if i am not mistaken)...
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA
Overclock.net › Forums › Graphics Cards › NVIDIA › [Official] GTX 580 Overclocking Club