Benchmarks I
I'll run SiSoft SANDRA's unbuffered benchmark first to see just how much bandwidth we have on offer. Running the unbuffered benchmark generally takes timings as well as basic speed into account. Remember, I'll be running memory at 166MHz with 2-7-2-2-1T enhanced timings, 222MHz (166FSB 3:4 ratio) with 2.5-7-3-3 timings, and 222MHz with 2-7-2-2-1T enhanced timings. The CPU speed remains constant throughout at 2830MHz.
Obviously we have more bandwidth with RAM at 222MHz than at 166MHz. What's pleasing, however, is the extra synthetic bandwidth that the stricter timings of 2-7-2-2 (enhanced) give.
You can see just how well it compares to other chipset configurations, especially with these timings and basic speed.
How do our RAM speeds and timings affect Pifast. It just loves bandwidth and strict timings. Benchmark calculated to 10-million places.
Here we see that having the ability to run memory at 222MHz with strict timings manifests itself in a 4.3 second advantage over decent timings at 166MHz. That's what you get when you have access to some super-fast RAM and a platform that can readily take advantage of the bandwidth that the RAM can offer. The CPU was steady at 2830MHz. The Pifast results also bores out the advantages of stricter timings at elevated speeds.
Another activity that benefits greatly from increases in usable bandwidth is DVD-to-DivX encoding. I'm benchmarking by using the DivX5.02 Pro with a 2-pass encoding of Gone in 60 seconds and 1800kb/s bit rate. An average is calculated when the first VOB is complete (both passes).
An increase in performance of over 6% by changing memory speeds from 166MHz (2-7-2-2 enhanced) to 222MHz. That's pretty impressive given that this activity can take multiple hours to complete.
