The last time we looked at the impact of different audio solutions on gaming performance, the results were a wash. We weren't expecting much of a difference this time around, so to make things interesting, we busted out the test methods we introduced in Scott's article, Inside the second: A new look at game benchmarking. In addition to measuring frame rates, we've measured individual frame times, which provide a much better picture of overall smoothness.
Rather than burying you under a deluge of frame-time graphs, we'll stick to a couple of results for each game: average frames per second and 99th percentile frame times. The FPS figures should be familiar to anyone who's read a PC hardware review in the last decade. The 99th percentile numbers refer to the time below which 99% of all frames are rendered. Lower frame times translate to higher frame rates and smoother gameplay.
Finding PC games with audio options that extend beyond simple volume controls proved rather difficult, so our gaming tests are limited to Battlefield 3 and DiRT Showdown. Let's start with BF3, which features an "Enhanced" audio mode that provides surround-sound virtualization for stereo output. Our testing was conducted with headphones, so we tested each card with standard stereo output and with BF3's Enhanced mode enabled. Since the Xonar DGX and DX both support virtualization via Dolby Headphone, we tested those configurations, as well.
To make the graphs easier to read, we've colored-coded the results. The Xonar DGX and DSX both appear in dark blue, while the results for the older DX are painted a lighter shade. We've grayed out the bars for the Realtek integrated audio.
Not much to see here, folks. Less than a single frame per second separates the various solutions in Battlefield 3. Our 99th percentile frame time results are just as conclusive, with only a 0.6-millisecond gap between the fastest and slowest configurations.
Next, we'll tackle the latest chapter in the DiRT franchise. This game offers standard software audio in addition to a Rapture3D mode with its own virtualization magic. We tested both modes on each config, and we ran another set of Dolby Headphone tests on the cards that support it.
With a very different game, we see largely the same results. The various configurations are tightly packed within a span of 2.7 FPS and 1.3 milliseconds, depending on the metric.
Gaming performance has long been gated by one's graphics card and, to a lesser extent, the CPU. It's been a long time since sound cards had any major impact on the equation.
We measure input latency with Audacity, the free audio editor used by our own podcast producer to piece together the various audio streams we generate during recording sessions. The latency test follows these instructions in the official Audacity manual.
All of the Xonars have longer input latencies than the Realtek codec. The DGX is a little slower than the DSX, but the difference there is just 18 milliseconds. The gap between the quickest Xonars and the Realtek implementation is much larger, at 50 milliseconds. Hearing a difference between the delays might be difficult for end users, though.
System power draw was measured first at idle and then in Battlefield 3 using the same configurations as in our gaming performance tests.
As one might expect, adding a sound card will increase system power consumption. However, the difference doesn't amount to much. The Realtek config saves only a few watts at idle and less than 10W under load. Among the Xonars, the DSX is the most power-efficient, particularly when playing Battlefield 3.
|Gigabyte's Z170X-Gaming G1 motherboard reviewed||1|
|Star Wars Battlefront video review||24|
|Club 3D active adapters convert DisplayPort 1.2 to HDMI 2.0||11|
|Phanteks' Power Splitter lets two systems run on one PSU||31|
|Just Cause 3 system requirements won't blow up your wallet||21|
|Biostar's GeForce Gaming GTX 950 glows a fiery red||17|
|Asus updates Zenbook UX305 with a Skylake Core M CPU||44|
|Shuttle XPC Nano's svelte body is clad in black and gold||19|
|AMD ends driver support for non-GCN Radeon cards||77|