I tested it just to see if independently adjusting the shader clocks made a large difference.
Note: I just used the RTHDRIBL program (running at 1920x1184 and 4XAA) as a quick way to test.
Results:
Increasing the core and keeping the shaders the same made a much larger difference. (ie. increasing the core from 515 to 650 (+135MHz) and keeping the shaders at 1500MHz gave about an additional 10fps while keeping the core at 650 and increasing the shaders from 1500MHz to 1600MHz (+100MHz) gave only about an additional 1-2fps. Memory was kept the same throughout.
Keeping the core at 515MHz and increasing the shaders to 1600MHz only gave about 1-2fps as well.
So I guess the core speed is more important or maybe there's some other bottleneck?
Anyone else done any testing?
Note: I just used the RTHDRIBL program (running at 1920x1184 and 4XAA) as a quick way to test.
Results:
Increasing the core and keeping the shaders the same made a much larger difference. (ie. increasing the core from 515 to 650 (+135MHz) and keeping the shaders at 1500MHz gave about an additional 10fps while keeping the core at 650 and increasing the shaders from 1500MHz to 1600MHz (+100MHz) gave only about an additional 1-2fps. Memory was kept the same throughout.
Keeping the core at 515MHz and increasing the shaders to 1600MHz only gave about 1-2fps as well.
So I guess the core speed is more important or maybe there's some other bottleneck?
Anyone else done any testing?
Facebook
Twitter