I've no experience of the AMLogic chipsets myself, but I seem to remember reading somewhere that their dual-core Mali is clocked higher than the quad-Mali in the RK3066 - possibly 400MHz vs 266MHz? I can't remember where I read this but, assuming it is accurate, this ought to mean that the GPU performance deficit may be less than 50%!
On the original AmLogic single-core platform, I think they used a Single Core Mali vs. the A10 which uses a Dual-Core Mali and furthermore the AmLogic was clocked at 266Mhz whereas the A10 had its Malis clocked at 400 Mhz.
As someone else mentioned, when it comes to 3D performance in particular, my guess is that the RK3066 does eat it for breakfast...
Here is some quadrant info using the Ramos W17 Pro vs the Cube U30GT. The U30GT is slightly higher resolution as well so bear that in mind. This is aggregated scores, I don't have the breakdown
My Guess is that the Cube is indeed smacking it around in the GPU tests which is why it ends up being so much higher for the Aggregate score.
Linpack Multithread testing gives the RK3066 around Mflops vs the AML chipset at about 60
Antutu GPU: 1586 for the RK3066 vs 1199 for the AmLogic chipset (which would seem to support the higher clock frequency for the Aml Mali's theory!).
I haven't used the AmLogic Dual-Core yet, my partners have used it though and the RK3066 and have said the RK3066 does indeed feel a bit faster. I don't think they did much 3D game testing. All I have to say is that NOVO 3 runs beautifully (very impressive) on the RK3066, I have no idea how it would go over on the AmLogic, it would be interesting if someone could test.
Edited by Roman2025, 26 July 2012 - 04:20 PM.