Tests

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.

For you lazy people out there, the table after each diagram shows the performance difference caused by a specific difference between the old and new CPU in each pair. It's basically the same data as shown on the diagram but presented as numbers.

The "Core clock" value is obtained by comparing the 4th CPU pair, Pentium G620 and G640, which only differ by 200 MHz of clock rate. You can find their specifications above.

3D Modeling

The 7.8% core clock rate increase adds just 4% of performance, less than we have expected. However, in other pairs, the progress is even smaller: the 1.5-times cache increase gives just 2.4%; the move from DDR3-1066 to DDR3-1333 yields nothing, and HT even decreases the results, because this test group demands no more than two threads.

Final 3D Rendering

The results in this multithreading group of benchmarks are rather predictable: only HT provides a significant boost—24.2%, and the cache size yields a slight progress as well. The core clock yields a directly proportional increase, because the tasks are computational. Again, the memory switch yields nothing,

Data Compression/Decompression

The triumph for the RAM, which yields the boost comparable to that from HT. Still, the percentage for the clock rate is larger.

Audio Encoding

These computational tasks are not cache or memory-demanding, and Hyper-Threading here yields a very large boost. By the way, quad-core Intel processors gain almost the same from HT, so sometimes more threads is better than more cores. This is also proved by the smaller difference between hexacore i7-3960X and quad-core i7-2600. For the low-end CPUs, HT becomes even more important: it lets dual-core i3 processors compete with triple-core AMD CPUs.

Compiling

Do we have special audio encoding tests aimed at multithreading? No, because those results are proved by compiling: here we have an even larger, 36%, boost! Of course, the compiling tests benefit from other specifications as well, but 2%, 4%, and even 6% are nothing if compared with the HT's boost. Obviously, there is a difference between dual-core processors: without HT we have a simple dual-core CPU, with HT we get a competitor for low-end quad-core solutions.

Mathematical and Engineering Computations

Here extra cores and threads have nothing to do, so the conclusion is simple: it's good to have a dual-core with big cache and high frequency. As for this group of tests, we remember that AMD Phenom II X2 and Intel Core 2 Duo E8000 lose to the top Core 2 Quad or Phenom II X4 just because of L2 cache or core clock difference, so the modern Pentium G870 can become a challenger as well.

Write a comment below. No registration needed!