Results: Cinebench R10 and Cinebench R11.5Cinebench R10
Cinebench R10 uses an older “scanline” method of rendering a scene when using multiple cores and processors. As such it divides the scene horizontally into as many stripes as cores are available, and lets the processors render what they are given. This is not exactly equitable, as some parts of the scene are more complex than others, and some cores are left idle longer than desired by finishing their portions before the other cores. It further divides the scene as needed, but again at the cost of processor downtime.
AMD has always done well with this benchmark. The 1100T matches the single threaded performance of the i7 860, and takes a big jump ahead when using all 6 cores vs. the 4 core/8 thread design of the i7. The dual and triple cores bring up the rear.
This latest version of Cinebench changes around the rendering a bit. Instead of scanlines, it renders tiles and again divides the scene. This tends to decrease the amount of downtime for each processor.
There is no single threaded test for this, so it maximizes threads depending on the processor. We see again the 6 core AMD parts taking a meaningful lead over the i7 part with 4 cores and 8 threads. While the i7 is clocked lower, it is still in the general ballpark in terms of price.