To: wbmw who wrote (256522 ) 10/7/2008 11:14:58 PM From: pgerassi Read Replies (2) | Respond to of 275872 Wbmw: Perhaps it is you that is fooled by use of just one task to simulate what happens in the real world. Those that do many tasks like Dan3 find that Opteron runs circles around Core 2 and its derivitives. And yes the funny business is well known given the history of these benchmark suites. Every time AMD or someone else does well in a benchmark, they are summarily dropped. When a benchmark does better in an older Intel CPU than a new one, it is dropped as well. The use of Excell sorting for 50% of the grade during the P4 and Athlon suites is an excellent historical example of how Intel plays funny with benchmarks to get the desired PR result. No ordinary person thought that was typical of desktop uses at that time. The previous benchmark suite showed P4 to fall well behind Athlon which is what ordinary people saw on their new PCs. In fact, even older benchmark suites showed P4 even worse wrt Athlon. There was even a class action lawsuit about the benchmarks making clains about performance of Intel based systems that were not met in real world usage. Cries of misrepresentation and outright fraud were laced through it. As for my dad's new Phenom X3 8450, its running two ATSC/NTSC tuners, encoding the output onto the hard disks, decoding two video streams to play in the rec room for him who likes old programs and in the living room for mom who like sports of many kinds, database application for knowing where to find any given DVD/VHS/OTA program, file serving, up/downloading, compiling and other miscellaneous tasks. That is a power user type load for a PC. It is typical of a HTPC use in most homes. That runs many different code streams all at the same time. This is unlike typical Intel benchmark suites that do just one code stream at a time, complete it and move on to the next single task. Given their platform limitations, it is the better way for them. Alas, not many people run their PCs as batch jobs. When the benchmarks run as a server or power user's PC does, Intel CPU platforms do poorly wrt K8/10 platforms. Its you that likes to run PCs as a batching system, one task at a time. The rest of us use it as a server doing lots of different things at the same time. Its the same with server buyers and they do test it doing 20 different things like the servers they want to replace or upgrade. Pete