![]() |
|||||||||||||||
| « Back - Print | |||||||||||||||
|
|||||||||||||||
|
Page 1 - Introduction
The Intel Developer Forum is undoubtedly one of the most interesting press events of the year. In IDF, Intel not only outlines the future of their upcoming products, but also gives an idea for the path of which the whole computer industry will follow. One of the main innovations in technology that we will see take a large role in 2007 is undoubtedly the release of Quad Core architectures. In 2007, we will see Quad Core CPUs begin to make a transition into everyday servers, workstations, and desktop PCs. In the course of the next few weeks, we will be seeing Intel introduce their server Quad Core processor the Xeon DP which will be based on the “Clovertown” Core. We will also be seeing Intel introduce their Core 2 Quad QX6700 which will be based on the Kentsfield core. Intel has been able to create their Quad Core architecturse by supporting Core Dual Core on the same processor. The Kentsfield architecture, in the simplest explanation of its workings, is basically two Core 2 Duo processors packed onto the same processor. Intel’s Cloverton architecture, on the other hand, is basically two Xeon DP Woodcrest CPUs in the same processor. Both architectures connect through the front side bus, which is a simpler solution compared to when Intel will have to begin constructing a new die that will have all four processing cores on it with a shared L2 cache.
The Kentsfield and Clovertown CPUs both come with 8 megabytes of L2 Cache. This is achieved by each two cores having 4Mbytes of L2 cache. The four cores are not located on their own pieces of Silicon. Instead, Intel has opted upon placing two separate dual core cores side by side in the same processor. The two dual core cores have their own separate L2 cache.
Looking into the future, the first 8-core architecture that will be released from Intel will utilize a unified L2 cache for the cores, be based on a 45nm production process, and most importantly have all cores on the same piece of silicon. The current code name of the architecture is Yorkfield. The architecture is expected to make an appearance in the third quarter of 2007. The test platforms that Intel provided are identical to each other except for the processors. The first test system was equipped with the Core 2 Quadro QX6700, while the other test system had the Core 2 Extreme X6800, which is currently the fastest CPU that Intel has available for desktop systems. The motherboard used in both test systems was the D975XBX2, also notoriously known as “BadAxe 2”. The motherboard is based on the Intel 975X chipset. The chipset’s specifications are identical to all other 975X chipsets that are available for commerce. The video card used was a NVIDIA GeForce 7950 GX2, while for memory the system was equipped with 2 modules of 1 Gbyte Corsair XMS of DDR2-800 memory with timings of 4-4-4-12. The hard drive was a Seagate Barracuda 7200.10. The operating system was Microsoft Windows XP Professional. Windows XP Professional, as can be seen from the above photograph, recognized the Quad core processor and all of its four cores.
Page 2 - The First Benchmarks
The benchmarks shown here were conducted during the Intel Developer Forum in fall of 2006. As the test systems and their suite of test benchmarks were all chosen by Intel, we were thus unable to test the systems with the benchmarks we usually use. The benchmarks, in addition, are limited to ones that actually take advantage of the Quad core processor. We should hopefully have a sample Kentsfield processor in our labs before its launch in November, and so at that time we will be able to test how the CPU performs compared to a larger variety of Intel and AMD CPUs.
3D Mark 2006’s CPU Test shows the benefits of the quad core architecture of the Core 2 Quad QX6700. Compared to the Core 2 Extreme X6800, which is the fastest processor that is currently available for retail, the QX6700 performs 60% faster. The results of the 3D Mark 2006 benchmark, as far as the final performance score, shows that the dual-core solution performs similar to the Quad-core test system. This is most likely the result of the higher clock speed of the X6800. The final score was 8941 points (QX6700) against 8297 points (X6800). Looking at the performance tests with a little more detail, we see that the final test that gave the QX6700 a performance advantage was the CPU test, while the video performance tests had pretty much the same performance scores.
In the POV-Ray multi-threaded beta standard benchmark, the use of the Quad core QX6700 resulted in a substantial increase in performance. It held an advantage of 81% compared to the X6800 processor. As a side note, the results we obtained from the X6800 weren’t weak results. The Core 2 Extreme X6800 actually had outstanding performance, meaning that the QX6700 had to really bring up its game to beat the X6800.
In the above test, the Core 2 Quad QX6700 encoded the video 27.355% faster than the Core 2 Extreme X6800. In this case, because we were able to take advantage of all four cores, we were able to compensate for the lower clock speed of the QX6700.
The Core 2 Quad QX6700 was able to shave off 33.76% of the time the Core 2 Extreme X6800 spent in the above test. The tests were converting some clips in standard HD using Sony Vegas 7.0a. In this case, the use of the Quad Core architecture allowed for considerable reductions in the time.
Here, the Core 2 Quad QX6700 rendered the Dragon_Character_rig.max scene in 38.5% less time than the Core 2 Extreme X6800 was able to. | |||||||||||||||