AMD: A-Series APUs (Trinity/Llano)
AMD's Fusion initiative sought to combine host processing and graphics resources in the same chip, ideally circumventing the need for separate CPUs and graphics cards in mainstream PCs. In many ways, this is similar to what Intel does with its Sandy and Ivy Bridge architectures, though Intel more heavily emphasizes x86 performance, while AMD's strength is its GPUs.
A disagreement with another vendor led AMD to move away from its Fusion brand and adopt Heterogeneous Systems Architecture, or HSA. The SoCs belonging to both efforts are referred to as APUs (accelerated processing units).
The first desktop APUs emerged under the code name Llano in 2011, and were manufactured using a 32 nm process. They combined AMD's Stars architecture without L3 cache and its Evergreen graphics design. Second-generation Trinity-based APUs employ the company's most modern Piledriver CPU architecture and its VLIW4 graphics configuration (the same shader arrangement found on Radeon HD 6900-series cards).
- Gaming At 1920x1080: AMD's Trinity Takes On Intel HD Graphics
- AMD Desktop Trinity Update: Now With Core i3 And A8-3870K
- AMD Trinity On The Desktop: A10, A8, And A6 Get Benchmarked!
- Better With Time? The A8-3870 And Pentium G630, One Year Later
- Professional Help: Getting The Best Overclock From AMD's A8-3870K
- AMD A8-3850 Review: Llano Rocks Entry-Level Desktops
Benchmarked AMD Fusion-Based APUs:
|A-Series||Code Name||Rev.||CPU Socket||Number of Cores||Clock Frequency||L2 Cache||iGPU||Memory Controller||TDP|
|A6-3650||Llano||B0||FM1||4||2.6 GHz||4 x 1024 KB||HD 6530D444 MHz||integrated up to DDR3-1866||100 W|
|A6-3670K||Llano||B0||FM1||4||2.7 GHz||4 x 1024 KB||HD 6530D444 MHz||integrated up to DDR3-1866||100 W|
|A8-3850||Llano||B0||FM1||4||2.9 GHz||4 x 1024 KB||HD 6550D600MHz||integrated up to DDR3-1866||100 W|
|A8-3870K||Llano||B0||FM1||4||3.0 GHz||4 x 1024 KB||HD 6550D600MHz||integrated up to DDR3-1866||100 W|
|A10-5800K||Trinity||A1||FM2||4||3.8 GHz||2 x 2048 KB||HD 7660D800 MHz||integrated up to DDR3-1866||100 W|
But i want some processors which were legendary overclockers, and representatives of their generation of CPU's, included with a nominal OC :
intel C2D E7300 : 2.66- > 3.33
Intel C2Q Q6600 : 2.4- > 3.0ghz
Intel i5-750 : 2.66 - >3.33
Its highly likely that a person has owned at least one of these CPU's. I want to know how well these compare to modern processors.
Agreed, maybe just one dual core and one quad? q9550 and e6850? not that I still own both of those or anything...
But let's do some math. Just for a rough order of magnitude I figure an average of 15% increase in performance per clock cycle, per generation (not including clock speed, number of cores, etc.). So if we start back at Conroe and work our way to present day Ivy Bridge, that's 5 new generations of processors. 1.15^5 = 2.01
Which means that an Ivy Bridge CPU at the same speed as a Conroe CPU (2006ish) is about 2x as fast per clock cycle, on average. Once you take into account faster clock speeds, number of cores, cache sizes, integrated memory controllers, etc. and more importantly what software will be used with the CPUs the real world performance difference could be almost nothing to somewhere around 10-15x as fast.
I digress. The point being, is I would like to see some more benchies Tom's! Prove me wrong!