Test Configuration

Test Configuration – Hardware

  • Intel Core i7-4790K (reference 4.0GHz, HyperThreading and Turbo boost is locked on to 4.4GHz by the motherboard’s BIOS; DX11 CPU graphics), supplied by Intel.
  • ASUS Z97-E motherboard (Intel Z97 chipset, latest BIOS, PCIe 3.0 specification, CrossFire/SLI 8x+8x)
  • Kingston 16 GB HyperX Beast DDR3 RAM (2×8 GB, dual-channel at 2133MHz, supplied by Kingston)
  • GTX 1080, 8GB, Founder’s Edition, reference clocks supplied by Nvidia and also overclocked +190MHz core and +400MHz memory offsets.
  • 2TB Toshiba 7200 rpm HDD
  • EVGA 1000G 1000W power supply unit
  • Cooler Master 2.0 Seidon, supplied by Cooler Master
  • Onboard Realtek Audio
  • Genius SP-D150 speakers, supplied by Genius
  • Thermaltake Overseer RX-I full tower case, supplied by Thermaltake
  • ASUS 12X Blu-ray writer
  • Monoprice Crystal Pro 4K

Test Configuration – Software

  • Nvidia’s GeForce GTX 1080 Launch Drivers 368.13 and 368.16 for overclocking (they only differ in that 368.16 enabled Precision X).   High Quality, prefer maximum performance, single display. 
  • VSync is off in the control panel.
  • AA enabled as noted in games; all in-game settings are specified with 16xAF always applied
  • All results show average frame rates including minimum frame rates shown in italics on the chart next to the averages in smaller font.
  • Highest quality sound (stereo) used in all games.
  • Clean install of Windows 10 64-bit Home edition; all DX11 titles were run under DX11 render paths. Our DX12 titles are run under the DX12 render path. Latest DirectX
  • All games are patched to their latest versions at time of publication.
  • EVGA’s Precision XOC, latest non-public beta version Precision X16 released 5/16/2016.

The 26 PC Game benchmark suite & 1 synthetic test

  • Synthetic
  • Firestrike – Basic & Extreme
  • DX11* Games
  • Crysis 3
  • Metro: Last Light Redux (2014)
  • GRID: Autosport
  • Middle Earth: Shadows of Mordor
  • Alien Isolation
  • Dragon’s Age: Inquisition
  • Dying Light
  • Total War: Attila
  • Grand Theft Auto V
  • ProjectCARS
  • the Witcher 3
  • Batman: Arkham Knight
  • Mad Max
  • Fallout 4
  • Star Wars Battlefront
  • Assassin’s Creed Syndicate
  • Just Cause 3
  • Rainbow Six Siege
  • DiRT Rally
  • Far Cry Primal
  • Tom Clancy’s The Division 
  • DOOM (*OpenGL game)
  • DX12 Games
  • Ashes of the Singularity
  • Rise of the Tomb Raider 
  • Hitman

Nvidia’s Control Panel settings:NvCP1NvCP2

Overclocking with Precision XOC – EVGA Precision X16 non-public Beta

Precision XOC is still in beta and yet it offers these features which implement all capabilities of GPU Boost 3.0, including:

  • Per voltage point frequency offsets
  • Support for integrated OC scanners
  • Overvoltaging controls

Precision XOC uses an integrated scanner that we ran.  It runs FurMark, an OpenGL program, to attempt to find the highest stable overclock.running

There are some known Issues that will be fixed for public release which I also encountered

  • Temperature limits are wrongly reported as the power target limit.
  • Recovery from the overclock scanner crashes usually crashes the driver and it will stop responding but sometimes recover.
  • Too high of an end scan point will lead to hard crashes which require a reboot.
  • The scanner cannot test the theoretical max clock to find all of the available headroom yet

We definitely experienced crashes – one after another until we finally gave up on using the scanner and we finally decided to manually overclock.

UI (User Interface)

The UI is still a work in progress. Here is the stock Precision XOC when you first open it.  We have clicked on the yellow triangle on the left to open the voltage adjustment and we have moved it to 100% in the red (below left).voltage adjustment

There are additional functions as the yellow triangles navigate to different top level pages.

  • Over-voltage is accessed by clicking the area left of the sliders near the lightning bolt icon.
  • Monitoring can be accessed on the 3rd page and can be expanded by double clicking on the monitoring graphs

All scanner pages use blue and green which mean:

  • Blue is the default V/F curve
  • Green is the overclock settings curve
  • You can adjust the curves by clicking the area above the blue section

Hit apply to apply any changes you wish to make or default to cancel.

Linear

scanner-linear

  • Linear mode active voltage points are denoted by the green bars at the top of the columns
  • You can assign a curve by selecting the voltage column and a desired frequency along the Y-axis
  • Hitting the default button will clear your VF curve for all modes

Basic

This can be used to help set your maximum overclock.scanner-basic

Basic uses the same overclocking method available to Maxwell by assigning a standard offset to all voltage points and if you assign a curve by selecting a frequency for any voltage and an offset curve is set automatically – just hit apply to set or default to clear it.

Manual Modescanner-linear-manual-run

This didn’t work for us very well as we had too many crashes with the scan mode which should otherwise be able to give the user a good idea of the GPU’s overclocking capabilities.  We did run with Vsync off which is critical to success but this is what we got:scanner-linear-manual-run-RESULT

So we finally set everything manually. Here is Precision XOC at default running looped Heaven 2.0, and notice that the core clock will throttle below 1733MHz when it gets too warm.

Stock everything
Next all we did next was to set the Power Target and the Temperature Target to maximum, and we continued to loop Heaven:only power - temp

Now we see a boost well above 1733MHz.  Everything else remains at stock and this is how we run our basic testing of the stock GTX 1080.

 

Now we overclocked to our highest stable frequency.  Notice that the Power Target is reduced from the maximum 120% down to 110% as this same overclock was not completely stable with a maximum Power Target.  Also, notice how the Heaven scores continue to increase sequentially with each change that we made.full oc no voltage
We achieved a final stable overclock of +190 MHz to the core which settled in above 1950MHz with GPU Boost while Heaven looped continually.  No matter what we tried – maximum voltage, fan at 100%, a cold room – we could not achieve a +200MHz offset.

 

Our GTX 1080’s GDDR5X memory also took a final offset +400MHz to achieve a 5400MHz final stable memory clock.  We noticed that there is a memory hole with instability around +500MHz offset, and that we could then push it even higher to +600MHz, but we lost some performance compared with a +400MHz offset.

 

So ultimately, the final stable overclock for BTR’s GTX 1080 sample is +190MHz core offset leading to an average boost over 1950MHz and a +400MHz memory offset resulting in a 5400MHz memory clock.  This is with a power target of +110% (not 120% which is unstable) and a temperature target of 91C.  However, there were 2 problematic games out of 26 games that required an adjustment downward – GTA V (which has always been extremely sensitive to all overclocking) and Ashes of the Singularity (which has been somewhat problematic for Nvidia cards) which required lowering the offset by 20MHz to +170MHz offset.

 

Perhaps once Precision X is finalized, our GTX 1080 overclock will go even higher with more stable fine-tuning of voltage, but this overclock is what we have now.  Let’s see the performance results of 26 games with our games at stock GTX 1080 speeds compared with our highest stable overclock.

3 COMMENTS

  1. Good article, thank you. I like the amount of detail included regarding how you did your overclock, given tool limitations. I will say, that I would have liked to see the minimum frame rates encountered, and even timeline graphs so we could see the frequency of any framerate dips.

    Otherwise, thanks again. Looking forward to your benchmarks of the 1070 when available. Not sure I’m ready to spend $900CAD on this card.

  2. Thank-you. Actually minimum frame rates are generally posted on the charts for DX11 games. They are the smaller font numbers in italics next to the bold averages. Eventually, we will add FCAT testing and charts to our benches.

  3. The frequency to voltage curve hints are pure gibberish in what is the outcome. Linear does nothing, clicking dragging and what not, nothing happens. Manual you have a complete mess quickly clicking on each column, no interpretation if you select 4-5 points, it would take a lot of labor to adjust any kind of profile and if it fails then what?

    The article has no examples of the curves and the outcome and what will cause a higher voltage and a lower one. It is so easy just to crash a 1080Ti by one click and apply. On a 1440p monitor the graphs are almost illegible, I can’t help wonder how they would look at 4K.

    In short precision is the worst of the Nvidia OCing tools I’ve used, garbage.

Comments are closed.