Benchmark Results: Digital Audio Workstation
We used two different Digital Audio Workstation tests. First, we used DAWBench, which is about as close to an industry standard test for the DAW market as there is. Secondly, we used a ‘score’ piece written just for our purposes. Both tests are performed under REAPER. The system is equipped with a Creative Labs X-Fi Titanium PCIe card. But because Creative Labs disables their ASIO drivers under 64-bit versions of Windows, we had to use the ASIO4All drivers instead.
DAWBench is a DAW-oriented metric that tests the digital signal processing ability of the system when playing back audio on your workstation in real-time. The test is run at various latency settings (set by the number of samples). The DSP ability of a DAW system in real time is critical when working in your DAW, as you can run out of processing power. It works by applying multiple copies of a multiband compressor (specifically, Wave Arts MultiDynamics 5) to audio tracks within the software during playback until the system cannot keep up with the workload. This usually means the audio playback starts breaking up. The ‘measure’ of the test is how many copies of the compressor are applied before audio breakup occurs, and the “Samples” is ASIO latency, in samples.
DAWBench shows almost a 71% improvement across the board from Hyper-Threading. Higher latency settings benefit less from Hyper-Threading, possibly showing that the system is encountering another limitation like cache size or memory access times. A professional audio card with ASIO drivers native to it (as opposed to using ASIO4All running on top of the Creative drivers) would likely improve the performance of the system in this test.
Tom’s Score Test
This benchmark employs a piece of music consisting of two audio tracks (percussion), plus multiple tracks of software synthesizers (brass, strings, pads) and choral vocals. Reverb, compression, and equalization are applied to the tracks separately.
This test, like the 3D tests, was designed as more of a ‘real workload’ test instead of something synthetic. The results indicate how long it takes to mix the piece down to a stereo 24-bit WAV file at the highest quality settings. Two different sample rates are used in the test: CD-quality 44.1 KHz, and the highest sample rate of 192 KHz.
Like DAWBench, the score test also benefits drastically from Hyper-Threading. The audio driver issue that affects DAWBench is not likely to affect this test, as it runs purely on the CPU.