The wider pipeline means that, theoretically, you can push twice as much data down the SIMD pipeline but as I mentioned Qualcomm lengthened the standard pipeline by also adding more stages thereby taking the data a little bit longer to complete a clock cycle. Meaning the added stages can add latency and therefore increase the amount of time it takes to push calculations through, but then again, Snapdragon can process a lot more SIMD data than a normal Cortex-A8 core. Which is why you won't always see 2x the performance in SIMD oriented benchmarks.
I hope I'm describing that well enough to make some sense.
As far as main memory access goes, that would mostly depend on different things. The amount of bandwidth to main memory and the speed of the memory itself. That includes the Mhz the RAM runs at along with it's specified latency.
Take the Droid1 and the Droid X as an example of main memory access speed. Even if both phones are clocked at 1Ghz the Droid X seems faster in normal day to day use as it uses faster LPDDR2 as compared to the original Droid's LPDDR1. I'm not sure what Snapdragon uses but I wouldn't be surprised if it's LPDDR1.
EDIT: Go here for a better description of the differences between Snapdragon and Cortex-A8.
Last edited:
