The HP Compaq dc7600 starts with the Pentium D 820 and adds 1GB of DDR667 memory, using a proprietary motherboard with Intel’s 945G chipset that includes a GMA350 graphics core. HP wisely sidesteps the integrated graphics and substitutes an ATI Radeon x300 SE PCI Express DVI card with 128MB. (This value-point graphics option is aimed at multimedia PCs, but would probably not satisfy a hard-core gamer.) An 80GB SATA hard drive (rated at 3Gb/s) and CDRW/DVD-ROM combo drive cap off the system hardware. Windows XP Pro with SP2 runs the show. A variety of CRTs and LCDs are available. This configuration sells for $1,242.
For comparison, I tested the HP Compaq dc7600 alongside two other computers: a home-built single-core Intel Pentium 4 640 system with HyperThreading, and a dual-core Athlon X2 4200+ system.
As with the Cyberpower Athlon system, I used the COSBI OpenSourceMark (OSMark) benchmarking tool. It's a synthetic test that doesn't rely on real-world applications, but uses a prolonged series of trials that exercise the CPU, memory, graphics, and, to a limited extent, the disk subsystems. The test is designed to provide multiple threads when more than one real or virtual processor is detected.
Preconceptions about clock speed -- as in "a higher clock speed results in a faster system" -- tend to go awry when testing dual-core or HyperThreading processors. For example, the dual-core 2.2GHz Athlon X2 test system came out ahead with a score of 1,806, while the single-core Pentium 4 640 system (clocked at 3.56GHz) placed second with a 1,266 result. The review system, the 2.8GHz dual-core HP Compaq dc7600, brought up the rear with a miserly score of 1,062. Thus, the Athlon X2 4200+ pulled the best test numbers although it was anywhere from 600MHz to 1.3GHz slower than either of the Pentium systems.
Testing For Real-World Apps
However, because the OSMark test doesn't load all subsystems, especially the hard disk, it's often a better indicator of a computer's strength with applications that mimic those same types of operation, like games or graphics. It is not necessarily a good indicator of a computer's strength with database or video rendering applications. That's where the video rendering test comes in.
For this, I used Ulead's VideoStudeo 8.0, a video editing and rendering package. Six clips, totaling 42 minutes and 39 seconds in length, were defined from 1 hour and 7 minutes of master MPEG video and rendered twice. The first time through, I had VideoStudio extract the clips from the original and render them into a single MPEG video with the same parameters (resolution, video and audio bit rate, etc.) as the original clips. This method suppresses the video display during rendering. For the second pass, I changed the parameters for the final video, causing VideoStudio to display the video as it renders.
The results of the first part of the video rendering test didn't favor the dc7600. When not displaying the clips, the single-core Pentium 4 640 rendered the six video segments in just 141 seconds. The X2 4200+ and dc7600 finished neck-and-neck at 253 and 262 seconds, respectively. This is a demonstration of raw horsepower – clock speed plus memory management. The 3.56GHz Pentium 4 640 and its HyperThreading technology muscled both the 2.2GHz Athlon and 2.8GHz Pentium dual-core processors out of the way. (Note, however, that the gap between the Athlon system and the dc7600 was nearly negligible.)
However, when I forced the video to be displayed in the second video test, things changed. In spite of its mid-range graphics card, the dc7600 required only 34 minutes, 11 seconds to render the final video. It passed the Pentium 4 640, which needed 37 minutes, 44 seconds, and literally blew by the Athlon X2 4200+’s time of 49 minutes, 54 seconds.
What happened in this last case is a demonstration of cohesive system performance. As the video data came off the hard drive and down the bus to the memory and CPU (where it was decompressed, pulled apart, stitched together, and then recompressed), it was also being pushed back up the bus to the graphics card to be displayed on the screen. This is a good indication of how the system will perform for most multitasking day-to-day applications.
Conclusion
Unlike gaming, where the graphic card’s GPU does much of the work, most typical business applications use many or all of a computer’s resources in concert with each other. (Video rendering, which causes system-wide heavy lifting, is an extreme example of that behavior.)
Based on the results of the OSMark and video testing, it's clear that AMD remains the king of gaming. However, it would appear that HP’s dc7600 is the obvious choice for cost effectiveness when desktop work gets serious.
HP Compaq dc7600 Hewlett-Packard, Inc. www.hp.com Price:$1,242 Summary:Intel's new Pentium D dual-core CPU won't speed up your favorite game, but it is ideal for serious business apps. |