All right, now that you’re thirsty for more performance numbers, let’s see dig into our Espresso results. For a test bed, we used a Phenom II X4 955 (3.2 GHz) running at stock speeds running on an MSI DKA790GX Platinum motherboard with 2 x 1 GB of Corsair 6400C3 memory and a PC Power & Cooling Turbo-Cool 860 PSU. With the AMD ATI Radeon HD 4890 we used the Catalyst 9.5 hotfix driver, and with the GeForce GTX 280, we used the GeForce 185.85 driver.You’ve already had a preview of the first test, going from high-def to low-def MPEG-4. Here’s you can see the results more plainly. This marks a pretty obvious win for AMD, especially if you factor in the CPU usage numbers.
We ran a similar second test, increasing both the file size of the source video (another YouTube HD file) as well as the resolution of the MPEG-4 target (720x480). With the larger numbers, we have more chance for test factors to even out. Sure enough, we see AMD breezing past Nvidia again, although not by quite such a large amount this time. Our CPU-only tests confirm the level playing field. Adding in GPU acceleration, we get a 91% improvement with CUDA but a 142% improvement in Stream—a narrower gap than the first test but still a fair drubbing for Nvidia.
And the CPU numbers? Same story. I saw sustained processor utilization above 90% on both sides of the fence when running CPU-only. With CUDA enabled, sustained usage fell to the 80% to 90% range across all four cores while Stream usage plummeted to around 40%, putting most of the load on core 2 and the GPU (20-24%).
Score: AMD 2, Nvidia 0.
- Underground Stream: A GPGPU History
- Whither Avivo Video Converter?
- The “Balanced Platform”
- CyberLink Serves Up Espresso
- Let’s Pull Some Shots!
- Add Shots
- Mixed Messages
- Heavier Lifting
- Sim(ply Not Ready)HD
- In AMD’s Words
- Why Only These Codecs?
- Putting The “General” In GPGPU
- AMD On Stream’s Prospects