UPDATED: Gears of War 4 GPU Benchmark - Ultra, High, & Scaling

By Published October 06, 2016 at 7:08 pm
  •  

We had a clerical error in our original Gears of War 4 GPU benchmark, but that's been fully rectified with this content. The error was a mix of several variables, primarily having three different folks working on the benchmarks, and working with a game that has about 40 graphics settings. We also had our custom Python script (which works perfectly) for interpreting PresentMon, a new tool to FPS capture, and that threw enough production changes into the mix that we had to unpublish the content and correct it.

All of our tests, though, were good. That's the good news. The error was in chart generation, where nVidia and AMD cards were put on the same charts using different settings, creating an unintentional misrepresentation of our data. And as a reminder, that data was valid and accurate – it just wasn't put in the right place. My apologies for that. Thankfully, we caught that early and have fixed everything.

I've been in communication with AMD and nVidia all morning, so everyone is clear on what's going on. Our 4K charts were completely accurate, but the others needed a rework. We've corrected the charts and have added several new, accurately presented tests to add some value to our original benchmark. Some of that includes, for instance, new tests that look at Ultra performance on nVidia vs AMD properly, tests that look at the 3GB vs 6GB GTX 1060, and more.e titles distributed to both PC and Xbox, generally leveraging UWP as a link.

Gears of War 4 is a DirectX 12 title. To this end, the game requires Windows 10 to play – Anniversary Edition, to be specific about what Microsoft forces users to install – and grants lower level access to the GPU via the engine. Asynchronous compute is now supported in Gears of War 4, useful for both nVidia and AMD, and dozens of graphics options make for a brilliantly complex assortment of options for PC enthusiasts. In this regard, The Coalition has done well to deliver a PC title of high flexibility, going the next step further to meticulously detail the options with CPU, GPU, and memory intensive indicators. Configure the game in an ambitious way, and it'll warn the user of a specific setting which may cause issues on the detected hardware.

That's incredible, honestly. This takes what GTA V did by adding a VRAM slider, then furthers it several steps. We cannot commend The Coalition enough for not only supporting PC players, but for doing so in a way which is so explicitly built for fine-tuning and maximizing hardware on the market.

In this benchmark of Gears of War 4, we'll test the FPS of various GPUs at Ultra and High settings (4K, 1440p, 1080p), furthering our tests by splashing in an FPS scaling chart across Low, Medium, High, and Ultra graphics. The benchmarks include the GTX 1080, 1070, 1060, RX 480, 470, and 460, and then further include last gen's GTX 980 Ti, 970, 960, and 950 with AMD's R9 Fury X, R9 390X, and R9 380X.

Gears of War 4 Insane Graphics on GTX 1080

Gears of War 4 Graphics Card Benchmark [Video]

Gears of War 4 PC Graphics Settings Explained

gow-gn-quality-sets-p1

gow-gn-quality-sets-p2

Let's start by defining some of the settings present within the game, and then defining their performance characteristics and consequential hardware requirements.

Scaled Resolution: Rather than a "normal" resolution drop-down selection, GOW scales resolution relative to the native display resolution. At 1920x1080, 200% would = 4K, for instance. If you'd rather scale resolution in game when performance flexibility presents itself, this would be the tool to use. VSR and DSR will handle scaling externally.

Vertical Sync: See our V-Sync dictionary listing for more on this. This setting synchronizes the refresh rate and the framerate, where the framerate exceeds the vertical refresh rate. If your FPS is locked to 60 in Gears of War 4, disable this setting and max-out the framerate.

Field of View: The width and distortion of the player's camera POV. Field of View (FOV) adjusts how wide the view frustum is.

Show Stats: Enables/disables FPS & frametime statistics in Gears of War 4. With praise to The Coalition, the Stats option also allows visibility of the time between frames (in ms) for both CPU and GPU, then further details CPU and GPU load level. VRAM consumption is also presented, theoretically with greater accuracy than what software tools report (e.g. GPU-Z, which can only show VRAM requested – not utilized).

TEXTURE SETTINGS

Character Texture Detail: Texture resolution for characters and character accessories/clothing. Jackets, boots, faces, skin, etc. are all affected by this setting. Most heavily impacts VRAM utilization.

World Texture Detail: As above, but for world textures. If desired, you could reduce character texture resolution in favor of greater world detail, or vice versa. Includes ground textures (dirt, mostly) and environment textures.

VISUAL SETTINGS

Anti-Aliasing Quality: Quality in taps (undisclosed) for AA, e.g. 4x, 8x, 16x, except presented as low/med/high/ultra. Represents number of samples per pixel to determine periphery pixel color and reduce jagged edges.

Temporal AA Sharpening: Hardens lines when using temporal (frame-to-frame) anti-aliasing. Multiple frames are analyzed to apply movement-specific AA which is not achieved through normal methods (e.g. MSAA). Sharpening strengthens the lines to reduce potential “blurring” of the result.

Foliage Draw Distance: Distance within the camera frustum at which the foliage is drawn. Lowering this setting will help cope with lower VRAM and GPU power, but will also reduce distance at which foliage remains visible.

World Level of Detail: LOD scaling impacts overall quality of environment meshes. Lower LOD for World Detail results in less geometrically complex objects.

Character Level of Detail: As above, but for characters.

Motion Blur: Blurring effect applied during movement to convey higher speed movement.

LIGHTING AND SHADOWS

Bloom Quality: Bloom is the effect that creates the “shimmering” seen near anything that might be hot or steaming. In real life, for instance, the equivalent would be to look out over a hot, empty parking lot mid-summer. If you see “shimmering” air above the lot, that's “bloom” in games.

Screen Space Shadow Quality: Quality of shadows as cast from objects within the z-buffer (our understanding – not 100% confirmed).

Ambient Occlusion Intensity: Strength of AO as applied to dictate light interactions with objects of varying translucency and with varying material types.

ADVANCED VISUALS

Post-Process Quality: Quality of effects applied after the render and texturing portion of the GPU pipeline.

Screen Space Reflections: Real-time scene reflection when looking at reflective surfaces and bodies of water. Reflects live objects and static objects. Severely punishes performance when maxed at Insane.

Environment Reflections: Detail of world and scene objects within shiny or reflective surfaces.

Particle Spawn Rate: Density of particles as dictated by maximum spawn rate.

REAL-TIME CINEMATICS

Depth of Field: As with photography. DOF impacts “bokeh” effect around objects in focus. Can be set to “insane” for severe GPU punishment. This is a filtration effect.

Sub-Surface Scattering: SSS impacts light as it is traced beneath the surface of semi-translucent objects, like thin skin near ears or fingers.

Test Methodology – Important (!)

Gears of War 4 is a DirectX 12 game that runs through the Microsoft Store. To this end, it is effectively impossible to use benchmark applications which shim their capture between the API and game engine. FRAPS is out, and so too are several of the other tools we normally use. Thankfully, with the help of GN's Carter Harris, we've been able to program a custom Python script to extract performance data from PresentMon. Our algorithm accomplishes what we've done in the past for FRAPS capture, primarily spitting out an overall AVG FPS (by juxtaposing frame number with onPresent ms delay), then further spitting out 1% low FPS and 0.1% low FPS. Learn more about what those metrics mean in our “What Are 1% & 0.1% Lows?” content.

This moves us past limits we had encountered previously with Intel & Microsoft's PresentMon endeavor, and gets us on track to measure performance similarly to previously established methods. PresentMon, for the unfamiliar, is an open source utility created by Intel and Microsoft to resolve the issue of performance tracking within lower level APIs. The tool hooks into Windows directly, granting accurate capture of both On Display and On Present metrics. At GN, we are using On Present (to be explained in a future article), which more closely matches the method that FRAPS has long used to capture framerate. On Display has merits, but we do not believe those to be overly useful for today's specific benchmark.

This benchmark uses the built-in “Benchmark” option, our own measurement tools, and the game's options menus. The resolution is tested at 4K, 2560x1440 (1440p), and 1920x1080 (1080p) with Ultra and High settings, mostly. Until a point at which we've got more time to really entrench in the settings (which, thanks to CitizenCon & travel, may be a while), we are using the Ultra and High presets as our primary configurations.

The latest Gears-ready drivers were used from both companies. NVidia's 373.02 drivers were used and AMD's 16.9.2 drivers were used. The 16.10.1 update contains the same GOW4 game optimizations as 16.9.2. Performance has not changed between driver packages.

Here's a look at our test bench:

Video Cards Tested

We've got a few more cards we'd like to test, too, including the 3GB GTX 1060 and 4GB RX 480 (update: 3GB GTX 1060 now included). We know, for instance, that some graphics settings in GOW4 are simply incompatible with the GTX 1060 3GB as a result of its sub-4GB VRAM capacity. We also suspect that VRAM limitations will choke the card's low performance, as shown in some of our other VRAM testing in this article.

GN Test Bench 2015 Name Courtesy Of Cost
Video Card This is what we're testing! - -
CPU Intel i7-5930K CPU iBUYPOWER   
$580
Memory Corsair Dominator 32GB 3200MHz Corsair $210
Motherboard EVGA X99 Classified GamersNexus $365
Power Supply NZXT 1200W HALE90 V2 NZXT $300
SSD HyperX Savage SSD Kingston Tech. $130
Case Top Deck Tech Station GamersNexus $250
CPU Cooler NZXT Kraken X41 CLC NZXT $110

In the meantime, though, this is what we're starting with. You may view our 3GB GTX 1060 vs. 6GB GTX 1060 content for more on those performance metrics.

Gears of War 4 Settings Scaling: Ultra vs. High, Medium, & Low Performance

This first chart shows performance scaling across multiple settings. We averaged the scaling between a few devices to generate the below output, with the goal of hopefully filling gaps where we weren't able to test every single card at every single graphics setting. It's just a time management question. With this, though, you can at least use the percentages to extrapolate relative performance gain or loss from various settings.

gow4-pc-benchmark-scaling

“High” is used as a baseline (100% performance) in this chart. Ultra sees an approximate 35% hit to performance versus “high” (the presets), with Medium improving performance by ~16%, and Low pushing us up by nearly 60% in framerate.

But let's move on to the charts that detail performance with greater insight.

Gears of War 4 Benchmark at 4K – GTX 1080 vs. 1070, RX 480, R9 390X, 980 Ti

gow4-pc-bench-4khigh

This is a look at Gears of War 4 at 4K. This chart is the same as our original publication run, and has remained accurate since initial publication.

At 4K/high, Gears of War 4 is handled at approximately 70FPS by the EVGA 1080 FTW Hybrid, with the GTX 1070 SC falling 34% behind at 53.3FPS AVG. The last-generation 980 Ti still outclasses Pascal's 1070 at 57FPS AVG, though does exhibit a wider FPS range overall. Low performance is improved somewhat globally on Pascal architecture over Maxwell.

As for AMD, the company is headed up by its Fury X at 50FPS AVG, followed by MSI's R9 390X at 43FPS AVG and the RX 480 Gaming X ($280) at ~38FPS average. The performance gap between the R9 Fury X and RX 480 8GB & MSI 390X 8GB cards is entirely resultant of inconsistent frametimes on AMD hardware, which often exhibit a range several times larger than what we're used to testing. This inconsistency means that the low values, as presented for AMD, are going to be all over the map. One test, the 0.1% and 1% lows might be 6 and 8; the next test, they could be 14 and 23. And the test after that, maybe back to 6 and 11. The nVidia devices, meanwhile, are generally within a range of 3~6FPS of each test pass, making for consistent and reliable numbers.

Here's a sample of raw data from the 4K test with 3 test passes on the Fury X:

AMD R9 Fury X 1050MHz

AVG FPS 1% LOW 0.1% LOW
50.4 30.8 23.8
50.4 35.9 31.6
49.6 37 32.3

Fairly consistent in the averages, with a >1FPS range, but the low performance is tied at the hip to memory limitations.

Gears of War 4 1440p Benchmark – GTX 970 vs. 1080, 1070, 1060, RX 480, RX 470

gow4-pc-bench-1440high3

At 1440p with High settings, the GTX 1080 FTW Hybrid from EVGA is now sitting at 131.7FPS AVG with tightly timed lows, including exceptionally consistent low frametime performance – as shown in this raw data – with the 980 Ti trailing. There's a noteworthy gap between Pascal and Maxwell performance when it comes to the low values, something which has been improved upon by nVidia for this generation of GPUs. 

GTX 1080 FTW Hybrid 1860MHz
AVG FPS 1% LOW 0.1% LOW
131.5 103 94.4
133.2 103.7 94
130.3 104.2 94.3

The GTX 1070 and 1060 both handle the game well at 1440p, with the AMD R9 Fury X operating at nearly 89FPS average, just above the GTX 970 SSC and GTX 1060 Gaming X at 75FPS average – effectively identical, again, other than Pascal pushing higher lows. Next, the MSI RX 480 Gaming X at 1316MHz is operating competitively to the GTX 970 and GTX 1060, with a 73.8FPS AVG, though its low performance is just as variable as we've seen in other tests. We saw at least one visible stutter in framerate output in the test runs, but AMD has a reasonable foundation with its average FPS to potentially improve performance in the lows in the future. 

The GTX 960, 950, and RX 460 are all low enough that you'd want to run at 1080p or dip settings, but few owners of these cards are likely to also run 1440p displays, anyway.

Gears of War 4 1440p Benchmark - Ultra Settings

gow4-pc-bench-1440ultra 2

As for 1440p at Ultra, we're seeing a performance output of about 100FPS on the GTX 1080, with reasonable lows that are just below 60FPS, 67.6FPS on the R9 Fury X, and 62FPS on the MSI R9 390X. Remember that the low values for AMD are a little deceptive here, since the variance is so great that some test passes will make the values appear higher than they are during certain runs or in extended gameplay. The GTX 1060 has a little bit of trouble with stutters at 1440p, but could be made playable with some drops down to “High” from “Ultra.” The RX 480 is in the same boat, and could be made more fluid with some fine-tuning.

Gears of War 4 1080p Benchmark – RX 460 vs. RX 470, 480, GTX 1060, 960, 970

gow4-pc-bench-1080high3

With 1920x1080 properly configured to “High” settings across the board, we're still hitting a bit of a CPU bottleneck with the high-end devices as limited at approximately 155FPS AVG with our 5930K at 3.8GHz. The GTX 1080, Old Man 980 Ti, and GTX 1070 are all peaking around 150, showing no issues whatsoever with Gears of War 4 at 1080p with High settings. The low frametime performance is reasonably tightly timed, with each card dipping at its lowest to 105FPS at the 0.1% metric. This plants NVIDIA close to 120Hz displays, though Gears of War 4 isn't exactly a game where that type of FPS throughput is necessary. 

AMD R9 Fury X 1050MHz
AVG FPS 1% LOW 0.1% LOW
124.3 77.4 25.1
125.6 96.8 59.5
123.3 104.1 96.2
GTX 1060 Gaming X 1809MHz
AVG FPS 1% LOW 0.1% LOW
115.9 89.9 79.2
115.1 91.7 85
116 90.9 83.4

Moving down the hierarchy, the AMD R9 Fury X is now pushing significantly better performance than what we saw with Ultra, hitting 124FPS AVG and lows of 92.8FPS and 60.3FPS. That number is deceptive, though, and it's for the same reasons we discussed originally with our Ultra tests. We are still seeing critical variance between test passes. Here's a quick look at some of the raw data (above) from the Fury X run on High settings, where we're seeing – as we saw before with Ultra – a dip to 25FPS for the absolute lowest output, and another 0.1% low output of 96.2FPS. That's a range of about 70FPS for the absolute lowest performance, which manifests itself occasionally – at least once every minute or two – as a stutter or hitch in framerate fluidity. If you look at the averages, though, you'll see that AMD's R9 Fury X remains fairly consistent in its average output. 

Compared to nVidia's devices, particularly Pascal, this is a major differentiating factor between the vendors. The good news is that AMD's average is high, and we're already in discussion with AMD on what might be causing those low dips. We suspect that a future driver push or game patch could further improve these low metrics. 

As for the 1080p results, the MSI GTX 1060 is effectively tied with the EVGA GTX 970 SSC. The GTX 1060 and 970 SSC have more consistent frametimes between tests, despite the R9 390X's overall performance between the 970 and 1060 cards. The R9 380X Nitro runs at about 83FPS AVG, and the RX 460 hangs on to remain playable, though would do better with settings reductions.

Gears of War 4 Benchmark - 1080p Ultra

gow4-pc-bench-1080ultra 2

Just to quickly show Gears of War 4 at 1080p with Ultra settings, here's a look at performance on both AMD and nVidia with the Ultra preset. We're seeing 143FPS AVG output on the GTX 1080 FTW Hybrid, 98FPS AVG output on the R9 Fury X, and 93FPS AVG on the R9 390X. The difference, again, is in the low performance where you're seeing some stuttering. Some of this can be seen in our review of Gears of War 4, already published on the channel. Those stutters do appear during gameplay, but the averages are good – it's just a matter of AMD and the Coalition working to stabilize the low-end.

Ultra is sustainable at 1080p on the R9 380X, RX 470, GTX 970 and up. The AMD cards at the low-end will need some settings adjustments to account for those drops, though.

Why Do the Low Values Look Like That?

The most critical takeaway here, and this will continue to 1080p, is that Gears of War 4 benchmarks with a fairly high variance in framerates on AMD hardware. This leads to results which, at the low end, can look somewhat inconsistent despite relatively consistent averages. In gameplay, the performance measurements manifest themselves in the form of “stutters” and frame drops. Play can feel choppy at times, and that's when there's a brief drop and hard hit to the low recorded framerates. Not every single test pass exhibits this behavior, but you'll run into the issue at least once every minute (or so) when playing.

GN has reached out to AMD for help in researching this issue. AMD has informed us that the team is investigating.

Note that 16.9.2 and 16.10.1 drivers both exhibit the same performance in Gears of War 4, which we've validated directly with AMD. We will have to look to future driver or Gears of War updates for resolution of these lower frametimes under certain conditions. Our present hypothesis is that, outside of VRAM limitations with some configurations, one of the couple dozen options within the “Ultra” preset is taxing AMD hardware exceptionally hard, because the impact is lessened at “High.” Again, this will require further research.

Best Graphics Cards for Gears of War 4

gears-of-war-4-slider-screen

The Coalition has done a remarkable job of offering PC enthusiasts a 40-item listing of graphics options, including built-in FPS monitoring, and further deserves praise for effort spent on detailing which options stress which system resources. We give The Coalition major praise for that.

Performance on the whole is generally high, with average outputs making most cards we tested reasonable pick-ups for "High" graphics settings at 1080p, and most (1060 and up, RX 480 and up) cards supporting 1440p/high. Moving to Ultra can tax the GTX 1060's and RX 480's low performance to a point of visible stuttering during gameplay, but a mix of ultra/high can be supported as a middle-ground on each device.

AMD's frametime performance seems disproportionately impacted to nVidia's, reflected in our raw data and low values above, though nVidia does generally seem to exhibit stronger performance in this particular title. We are yet unsure if driver updates or game patches would be the most direct path to resolution. GN has reached out to AMD to collaborate on researching the spurious performance metrics.

Performance is more widely varied in Gears of War 4 (release: October 11) than in some other games we've tested, like Black Ops III, but is playable on the whole by the greater majority of the modern graphics market.

Side note: Multi-GPU support is not yet prepared, but is intended to ship eventually. We will revisit once that's available.

Editorial: Steve “Lelldorianx” Burke
Video: Andrew “ColossalCake” Coleman
Sr. Test Technician: Mike “Budekai?” Gaglione
Test Technician: Carter Harris

Last modified on October 06, 2016 at 7:08 pm
Steve Burke

Steve started GamersNexus back when it was just a cool name, and now it's grown into an expansive website with an overwhelming amount of features. He recalls his first difficult decision with GN's direction: "I didn't know whether or not I wanted 'Gamers' to have a possessive apostrophe -- I mean, grammatically it should, but I didn't like it in the name. It was ugly. I also had people who were typing apostrophes into the address bar - sigh. It made sense to just leave it as 'Gamers.'"

First world problems, Steve. First world problems.

  VigLink badge