AMD 780G: Preview of the Best Current IGP Solution
- Posted in
- CPUs
The HD HQV benchmark is developed and provided by Silicon Optix. Our review of HD HQV is located here. Like the standard HQV Benchmark, the high-definition counterpart contains video tests and patterns that determine the quality of a product’s HD video signal processing. A set of five tests evaluate quality factors such as HD Noise Reduction, Video and Film resolution loss tests, as well a test for “jaggies”. We disable Aero capability and set the driver options to optimal settings.
Unfortunately, optimizing the drivers and PowerDVD 7.3 can and will make a difference in the tests where de-interlacing is important.
In addition, although this test is standardized to some degree, it can be subjective at times. This depends upon how strict you abide by the guideline sheet although even it differs at times with the video example tests.

In our first test, it is obvious that Image quality of the Intel G35 and NVIDIA GeForce 8200 is not up to par with the 780G with the Film Resolution test making the difference in the scores. To be honest, the 780G was on the verge of failing this particular test (subjective opinion) without optimizing the settings in the CCC panel. We also optimized quality settings in both the NVIDIA and Intel control panels to be consistent but did not notice any real changes in this test. All three boards were close to scoring 7 points in the HD noise reduction test but failed the flower section of the test while the sailboat test did show some slight improvements.
AMD is working on further enhancements in the Catalyst 8.4 drivers for the X2 processors to improve this situation and the jaggies test. With a Phenom installed, we see the advantages of the post-processing engine coming into play. Additional tuning is also available in the 8.3 drivers with the ability to turn edge enhancement off and on. We will have the standard HQV results in our roundup. The 8.3 drivers allow the 780G with a Phenom to score 122 out of 130 points and 113 with the 4850e.
Overall, we do not put much faith into this test currently as our eight person panel seemed to have a different opinion on just about every score. In fact, only three of our esteemed colleagues were in agreement with each other on test scores for each board and those were different (higher) than our published results.
Originally published at https://www.anandtech.com/show/2475 (Wayback Machine).