Our test method is briefly described here. The scores on diagrams are relative to that of our reference testbed that always scores 100 points. As of 2011, it's based on the AMD Athlon II X4 620 CPU, 8GB of RAM, and Palit's NVIDIA GeForce GTX 570 1280MB. The detailed (absolute) results are traditionally provided in this summary.
That "Bridges" perform well in these benchmarks is a well-known fact, so let's focus on two things: firstly, AMD's new architecture (other things being equal) is comparable to that of the 1st Gen Core, while the previous architecture is slower. Actually, you can overclock Llano to 3.5 GHz and achieve similar performance, but that would be close to the limit. In turn, the A10-5800K and FX-4170 work in the standard mode, leaving a chance to squeeze even more performance out of them.
Final 3D Rendering
In this group of tests, the module-based APUs lack resources. Fewer vector units cause A10-5800K and FX-4170 to fall behind A8-3870K, while the lower vector unit efficiency puts them behind Intel's counterparts. However, the high clock rates negate some of that loss and allow A10-5800K and FX-4170 to outperform the previous-generation Core i3. In turn, the comparison of FX-4170 and A10-5800K reveals the progress of the new core microarchitecture: Trinity outperforms Zambezi in spite of lower clock rates.
FX-4170 is out of reach due to its enormous total cache. The bronze medal goes to A10-5800K that fits between Core i3-2100 and Core i3-3240. A8-3870K is certainly an outsider, and the gap is significant. As you see, the progress of the new, module microarchitecture in this group of tests is outstanding. Let's take a closer look.
Compression with 7-Zip provides multi-threading integer load. Modules handle it better than standard cores do, and they also have higher clock rates. Compression in WinRAR requires two threads, which means Llano only uses the half of resources, cache included. Decompression tests are single-threaded, so the A10-5800K, with higher clock rates and 2 MB of cache per thread wins over A8-3870K with its 1 MB cache per thread. As you see, the difference is critical.
A8-3870K strikes back! With each of its four physical cores. However, A10-5800K is right on the heels, outperforming all other processors, including the similar FX-4170.
Again, the true quad-core A8-3870K is the best, outperforming FX-4170 aka Cache King. A10-5800K shares the third place with the best modern Core i3 processor—certainly a good result.
Mathematical and Engineering Computations
Like in the data compression tests, A8-3870K is a complete outsider while A10-5800K slightly outperforms the FX-4170 even despite the latter's higher clock rates. As for the Intel Core processors, they are out of reach.
Write a comment below. No registration needed!