HD5770 - The real cause of bottleneck

ATi 5770 Bottleneck Investigation | AlienBabelTech

Okay everyone till now thinks AMD made a big mistake with the HD5770 series with a 128-bit interface, well I feel they shud have gone for 1200SPs and 256-bit both but leaving my thoughts alone....I came across this review which showed what is actually slowing down the HD5770 is the CORE over memory, this review speaks all!!!

*alienbabeltech.com/main/wp-content/uploads/2009/10/Text1.png



*alienbabeltech.com/main/wp-content/uploads/2009/10/Graph1.png

Commentary

Well isn’t this interesting? The 5770 is primarily limited by its core overall, not by its memory!

Underclocking the core 20% results in a 11.20% performance loss overall, while doing the same to the memory results in a smaller 8.46% performance hit. This makes the 5770 a reasonably balanced card, with a small bias towards its core. This is unlike 8800 Ultra for example, which showed a massive reliance on its core across the board.

The results actually make sense given a 5770 is basically a 5870 chopped in half, and the 5870 was proven to be primarily limited by its core too. If you shrink the processing power and bandwidth in proportion to each other like the 5770 does (i.e. cutting both in half from the 5870), you’ll end up with very similar performance characteristics.

Also the performance hit increases when moving from 2xAA and 4xAA with the core (10.58% to 12.45%). We expect this to happen with the memory (and it does) but since the core is also affected, that tells us the ROPs are also bottlenecking the equation to some degree.

As for the individual games, we see Call of Duty 5, Wolfenstein, Call of Duty 4 and Fear 2 being massively more limited by the core than the memory, while the opposite applies solely to Bioshock. The rest of the games are affected reasonably equally by both clocks.

Click to expand...