The impetus behind this article came out of a desire to find out which of the two technologies, Avivo HD and PureVideo HD had more to offer to consumers, especially in its latest incarnation, where the entire video decoding pipeline can be hardware accelerated for most HD CODECs. The earlier articles have attempted to address these concerns but with the newer versions and finally a suitable Blu-ray /HD DVD player, we could do more comprehensive testing this time round.
Since the focus is on how Avivo HD and PureVideo HD perform in video decoding, we have gathered Blu-ray/HD DVD movies encoded with all three CODECs used nowadays, namely MPEG-2, H.264/AVC and VC-1. MPEG-2 is understandably going out of vogue, with H.264 and VC-1 the more popular ones. We have also tried to include movies that had higher bit rates than usual, in order to fully stress the hardware. The list of movies tested and the CODECs they are encoded in are listed below:
For the test system, we were aiming for a mid-range configuration that would be available to most enthusiasts. Since the graphics cards and their video decoding hardware were the 'stars' of this test, the details of the configuration weren't that important besides keeping it consistent for all the tests. The following are some of the relevant components we used:
As you shall find out shortly, we have tested quite a few graphics cards from both ATI and NVIDIA for this article. For the purpose of comparing Avivo HD against PureVideo HD, strictly as an evaluation of their merits, we have chosen to highlight the performance of the current favorites among enthusiasts now, the GeForce 8800 GT 512MB and the Radeon HD 3870 512MB. For all the NVIDIA cards tested, ForceWare 169.21 was used as the driver while the ATI cards were on Catalyst 8.1.
Our testing methodology was to record the average CPU utilization rates (hence, the lower this figure, the better) over the course of a 3 minute segment of the movies. Hardware acceleration was enabled to test the graphics cards and a reference set of results were also taken with no hardware acceleration. At least two runs of the tests were conducted, with the average taken. Windows XP's Performance Monitoring tool was used to record the CPU overhead.
A rather frustrating observation we found during the course of testing was the fact that hardware acceleration was automatically enabled by our version of CyberLink Power DVD Ultra and it could not be disabled, even when we attempted some registry editing. Since we were trying to show the advantages of hardware acceleration and needed a baseline to compare (without hardware acceleration), this posed a problem. Eventually, we got around it by using an OEM version of CyberLink Power DVD Ultra 7 that allowed us to disable hardware acceleration.