DOOM GPU Benchmark – Poor Performance on R9 300 Series

By Published May 14, 2016 at 12:46 pm
  •  

Following our GTX 1080 coverage of DOOM – and preempting the eventual review – we spent the time to execute GPU benchmarks of id Software's DOOM. The new FPS boasts high-fidelity visuals and fast-paced, Quake-era gameplay mechanics. Histrionic explosions dot Doom's hellscape, overblown only by its omnipresent red tint and magma flows. The game is heavy on particle effects and post-processing, performing much of its crunching toward the back of the GPU pipeline (after geometry and rasterization).

Geometry isn't particularly complex, with the game's indoor settings comprised almost entirely of labyrinthine corridors and rooms. Framerate fluctuates heavily; the more lighting effects and particle simulation in the camera frustum, the greater the swings in FPS as players emerge into or depart from lava-filled chambers and other areas of post-FX interest.

In this Doom graphics card benchmark, we test the framerate (FPS) of various GPUs in the new Doom “4” game, including the GTX 980 Ti, 980, 970, Fury X, 390X, 380X, and more. We'll briefly define game graphics settings first; game graphics definitions include brief discussion on TSSAA, directional occlusion quality, shadows, and more.

Note: Doom will soon add support for Vulkan. It's not here yet, but we've been told to expect Vulkan support within a few weeks of launch. All current tests were executed with OpenGL. We will revisit for Vulkan once the API is enabled.

DOOM At Max Graphics Settings – 4K@60FPS on GTX 980 Ti

DOOM Video Card Benchmark Video

DOOM Graphics Settings Explained

doom-graphics-settings

Doom's graphics settings are all fairly familiar, though a few new or rare additions crop-up in the menus. The graphics settings are divided into “Video” and “Advanced” tabs, the former containing more basic resolution, chromatic aberration, and V-Sync controls. All settings can be seen in the above menu screenshot. A break-down of key items is located below.

V-Sync: As always. Disable this if you want your FPS to exceed your refresh rate.

Anti-Aliasing: Options include the below –

  • FXAA

  • TAA

  • SMAA

  • TAA, FXAA, & SMAA (1TX)

  • TSSAA (8TX)

FXAA offers a low resource consumption solution – recently showcased in Skyrim – but can appear blurry to some players.

Temporal Anti-Aliasing (TAA) reduces temporal artifacts by sampling over time (ergo, “temporal”). TAA will look from one frame to the next to determine which edges get aliased and where, dependent upon movement and updates.

Sub-pixel Morphological Anti-Aliasing (SMAA) is a post-processing AA technique that fuses the temporal approach with the morphological approach, resulting in greater performance in the pipeline and generally high image quality.

As for multi-tap TSSAA (Temporal Super-Sampling Anti-Aliasing), this is a GPU-agnostic equivalent to nVidia's TXAA. TSSAA temporally analyzes frames to determine anti-aliasing as applied to objects in motion, and does so supersampled (8-tap, in this case) to increase accuracy.

Chromatic Aberration: As we discussed in our Witcher 3 graphics optimization guide, Chromatic Aberration stak”es its origins in photography, but has trickled into game graphics. Quoted from our Witcher guide: Chromatic Aberration gives the world the appearance of looking through a lens influenced by light. In real-world photography, this occurs when a camera has difficulty focusing on a specific wavelength of light, creating 'vibration' in the photo. We found chromatic aberration to have almost no visual impact to gameplay, and no impact to FPS whatsoever.”

Resolution Scale: Render scale of graphics elements in game. Performance impact parallels global resolution reductions. Generally, you want this at 100%.

Player Self-Shadow: Toggles your own shadow.

Directional Occlusion Quality: Per-pixel occlusion calculation derived from world space coordinates and normals within the GBuffer. Samples points in the world to approximate real-time global illumination and occlusion of light. Toggle if you're OK with sacrificing lighting effects and global illumination / light occlusion in favor of performance.

Decal Quality: Those blood splatters clinging to every surface are called “decals,” and they consume a marginal amount of VRAM on the video card. If you're experiencing VRAM issues (or just want to tone-down the blood), check this option first. Anisotropic filtering of decals will help apply decals to surfaces based upon camera viewing angle, and will not noticeably impact FPS.

HDR Bloom: High-dynamic range lighting bloom. This is particularly noticeable when near lava/magma flows. The “heat noise” above lava/magma is created using bloom effects. If this annoys you – or if you'd rather have the performance – disable it.

Depth of Field: Like in photography, more accentuated DOF will sharply focus on the subject (whatever you're looking at), while blurring background elements. Disable this if you'd rather have a “flat” look to game graphics.

Test Methodology

We tested using our updated 2015 Multi-GPU test bench, detailed in the table below. Our thanks to supporting hardware vendors for supplying some of the test components.

The latest AMD drivers (16.5.2 with Doom support) were used for testing. NVidia's unreleased 368.146 drivers were used. Game settings were configured to "Ultra" with anti-aliasing set only to FXAA (no multi-tap). Resolution settings were configured to 1080p, 1440p, and 4K. We performed some additional low-end GPU testing with “Medium” settings at 1080p. Tests were conducted in the Foundry, level 2. The test path is in the video further above (pending upload).

Each scenario was tested for 30 seconds identically, then repeated three times for parity, per setting, per card – that's a maximum of 9 to 12 test passes per device.

FPS fluctuates heavily in the game. We chose one of the more intensive interior settings. Generally, card performance should be slightly higher than what's shown here – we didn't figure out a definitive “worst case,” but found a middle-of-the-road benchmark course in terms of intensiveness.

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

Average FPS, 1% low, and 0.1% low times are measured. We do not measure maximum or minimum FPS results as we consider these numbers to be pure outliers. Instead, we take an average of the lowest 1% of results (1% low) to show real-world, noticeable dips; we then take an average of the lowest 0.1% of results for severe spikes.

Graphics Cards Tested

Doom Graphics Benchmark – 4K, Ultra – GTX 980 Ti vs. Fury X, 980, 970, 960, 390X

Although Doom is well-optimized and highly performant at the rapidly aging 1080p resolution, 4K still forces enough pixels down the pipe to cause some cards to choke.

doom-pc-benchmark-4k-ultra

The GTX 980 Ti (reference) maintains a 45FPS AVG FPS with tightly timed 37FPS & 35.67FPS lows. Dropping to “medium” would make for framerates approaching 60, but the current generation of Maxwell- and Fiji-class cards are incapable of 4K at Ultra settings in Doom.

Speaking of Fiji, the Fury X (which we've finally secured permanently) pushes 41.67FPS with less tightly timed lows (31.33 / 26). As with the 980 Ti, a settings reduction to “medium” would make Doom more playable.

The performance gap between the 980 Ti and Fury X is roughly 7.7%.

Notice that the Fury X outperforms the GTX 980 at 4K, but loses its lead as resolution declines. We talked about this in our original review of the Fury X, where we saw similar scaling.

Doom Graphics Benchmark – 1440p, Ultra – GTX 980 Ti vs. Fury X, 980, 970, 960, 390X

As is the trend with the current-gen Maxwell and Fiji/Hawaii/Tonga cards, 1440p is significantly more sustainable on single-GPU hardware than 4K. This generation made 1440p resolutions viable, and we suspect that the next generation will do the same for 4K (and VR) resolutions.

But we're not there yet; let's look at what we've got for now:

doom-pc-benchmark-1440-ultra

The GTX 980 Ti tops the chart (and that's a reference clock, too) with 86.33FPS AVG, 67.33FPS 1% lows, and 60.00FPS 0.1% lows. ~17% behind the GTX 980 Ti rests the GTX 980 with 72.67FPS AVG, flanked fast by the Fury X (19.5% slower than GTX 980 Ti; ~2.3% slower than GTX 980). All cards from the GTX 970 and up are able to sustain playable 1440p framerates, including the R9 Fury X. The R9 390X is more-or-less there, and would hit 60FPS with some slight settings tweaks.

Doom Graphics Benchmark – 1080p, Ultra – GTX 980 Ti vs. Fury X, 980, 970, 960, 390X

Despite the gradual move away from 1080p (very, very gradual, we might add), it's still far-and-away the most popular resolution right now.

doom-pc-benchmark-1080-ultra

At 1080p, Ultra, Doom proves playable on almost everything. The 380X and up (or GTX 960 and up) both sustain ~60FPS or greater averages. The GTX 950 isn't quite cut-out for Ultra (and neither would be the 270X). See notes below for info on medium settings.

The GTX 980 Ti pushes ~130FPS with the i7-5930K. AMD's R9 Fury X has lost some of its lead by the time we get to 1080p, but still holds fast against nVidia's best. Below the high-end cards, the near-60 FPS hitters include the R9 390X, GTX 960, and R9 290X.

AMD's R9 390X and 290X get obliterated by the GTX 970 – a performance gap of about ~39% – and the 290X is even getting beaten by the GTX 960. That, obviously, should never be the case with a properly optimized game and driver set. We're not sure who's at fault here, but it's clear that Doom doesn't play well with certain AMD architectures. We retested several times and validated on another (identical) machine, but still saw the issues. Some web searching shows that users of the 390X and 290X have experienced performance anomalies and framerates which coincide with our findings. The GTX 970 far-and-away outperforms the R9 390X right now, and performs effectively as well as the Fury X (for 1080p, anyway).

The only reason the 390X holds its own at the higher resolutions – at least, against nVidia devices – is because of the raw pixel throughput. AMD tends to do well in such scenarios, but that's just a brute force victory. It seems that a driver update or game optimization update is in order to maximize AMD's abysmal performance on its 390X and 290X. The Fury X isn't exactly where it should be, either, but so it goes.

Notes on 1080p, Medium

Just some quick notes:

  1080 Medium FXAA    
  AVG FPS 1% LOW 0.1% LOW
EVGA GTX 960 4GB 1342MHz 76.00 62.00 57.00
ASUS GTX 950 1329MHz 52.00 39.67 39.00

For anyone on the GTX 950, it looks like “medium” is more-or-less playable at 1080p. High falls between medium and ultra, offering minimal gains against Ultra.

Conclusion: Best Video Cards for Doom

The R9 390X and R9 290X are getting destroyed at 1080p by a GTX 970, and the 290X is even outperformed by the GTX 960 – something which wouldn't make any sense given an optimized scenario. It seems that AMD struggles with heavy shadowing and some lighting/particle effects, which are ever-present in the Foundry level (and several other levels) that we benchmarked. There's something unoptimized in either the drivers or game code that's affecting AMD's older 300 and 200 series architectures.

For 4K gaming with Doom, you'll have to step down to high/medium settings and would do best with one of these cards:

For 1440p, The following cards are ideal for “Ultra” settings:

We'd throw the R9 390X into that list, but it's obviously under-performing right now.

And at 1080p, it looks like something as simple as a GTX 960 or R9 380X will run the game just fine, mostly. If you want more sustained lows and guaranteed high performance through the heaviest load scenarios, the GTX 970 would be well-suited for that. Or a slight settings reduction – either works.

We'll update this as patches roll-out, Computex notwithstanding.

Editorial, Test Lead: Steve “Lelldorianx” Burke
Test Technician: Andie “Draguelian” Burke

Last modified on May 14, 2016 at 12:46 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.

We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.

Advertisement:

  VigLink badge