| ||
| by MAC | September 6, 2009 | ||
| Synthetic Benchmarks: CPU & Cache Synthetic Benchmarks: CPU & CacheLavalys Everest Ultimate v5.02 Everest Ultimate is the most useful tool for any and all benchmarkers or overclockers. With the ability to pick up most voltage, temperature, and fan sensors on almost every motherboard available, Everest provides the ability to customize the outputs in a number of forms on your desktop. We selected two of Everest's seven CPU benchmarks: CPU Queen and FPU Mandel. According to Lavalys, CPU Queen simple integer benchmark focuses on the branch prediction capabilities and the misprediction penalties of the CPU. It finds the solutions for the classic "Queens problem" on a 10 by 10 sized chessboard. At the same clock speed theoretically the processor with the shorter pipeline and smaller misprediction penalties will attain higher benchmark scores. The FPU Mandel benchmark measures the double precision (also known as 64-bit) floating-point performance through the computation of several frames of the popular "Mandelbrot" fractal. Both tests consume less than 1 MB system memory, and are HyperThreading, multi-processor (SMP) and multi-core (CMP) aware.  When it comes to computational power it is pretty clear that the Core i7 processors are in a class of their own. Both the i7 800 & 900 series put up impressive numbers largely due to the fact that they are Hyper-Threading enabled. The $562 i7-870 is secondly only to a $999 i7-975 Extreme Edition in the CPU Queen test, and third behind the i7-950 in the FPU Mandel test, not a bad first showing. The non-HT i5-750 is quite competitive in CPU Queen, slightly surpassing the AMD Phenom II X4 945, but falling behind the venerable Core 2 Quad Q9550, a chip that is currently priced $25 more than the i5 model. In FPU Mandel though, the tables are turned with the $170 X4 945 surpassing both the i5-750 and Q9550. Lavalys Everest Ultimate v5.02 As part of its enthusiast favourite Cache & Memory Benchmark, Everest provides very useful and in-depth cache performance figures. For this chart, we have combined the read, write, and copy bandwidth figures to achieve an aggregate bandwidth figure for each cache stage.  On the L1 front, the i7-975 and i7-860 chips once again lead the way, but AMD's Phenom II AM3 processors put up very impressive numbers as well. The overall field is really not stretched out by much though, except when you get to the lower-clocked Core 2 models and 3-year-old Q6600. When it comes to L2 cache bandwidth, the Nehalem-based chips leave all other architectures in the dust, posting aggregate bandwidth numbers that are 60% to 100% higher than the rest. We were in for a surprise when it came to the L3 cache bandwidth, since the new Lynnfield models blew away their Bloomfield counterparts. We do know that the Uncore has been tweaked on these new chips, but seeing the i7-870 (2400Mhz Uncore) post 45% higher bandwidth numbers than the i7-975 (2666Mhz Uncore) is unusual. The bulk of the performance improvement was in the read bandwidth, while the i7-900 series appeared to still have an edge with regard to write and copy speeds. Further testing (and perhaps a new version of Everest) will be needed to settle this issue. Now we also ran an assortment of SiSoft Sandra Pro 2009.SP4 (15.124) benchmarks, however we encountered too many anomalies to be able to firmly stand by the results, especially with regard to the Core i5-750. Nevertheless, here are the charts for the curious among you.
In each case, there is no rhyme or reason to explain the clearly anomalous results, and they were repeatable 3 to 10 times in a row. As a result, we chose to disregard them just to be safe. | ||
| |
| Latest Reviews in Processors | |||||||||
|
