I’ve used two desktop systems as the test machines: An AMD Phenom 9700 2.4GHz system fitted with an ATI Radeon 3850 and 4GB of RAM An Intel Pentium Dual Core E2200 2.2GHz fitted with an NVIDIA GeForce 8400 GS and 1GB of RAM The tests There are 31 tests in all, most of which are self-explanatory: 1.Install OS - Time it takes to install the OS 2.Boot up - Average boot time to usable desktop 3.Shut down - Average shut down time 4.Move 100MB files - Move 100MB of JPEG files from one hard drive to another 5.Move 2.5GB files - Move 2.5GB of mixed size files (ranging from 1MB to 100MB) from one hard drive to another 6.Network transfer 100MB files - Move 100MB of JPEG files from test machine to NAS device 7.Network transfer 2.5GB files - Move 2.5GB of mixed size files (ranging from 1MB to 100MB) from test machine to NAS device 8.Move 100MB files under load - Move 100MB of JPEG files from one hard drive to another while ripping DVD to .ISO file 9.Move 2.5GB files under load - Move 2.5GB of mixed size files (ranging from 1MB to 100MB) from one hard drive to another while ripping DVD to .ISO file 10.Network transfer 100MB files under load - Move 100MB of JPEG files from test machine to NAS device while ripping DVD to .ISO file 11.Network transfer 2.5GB files under load - Move 2.5GB of mixed size files (ranging from 1MB to 100MB) from test machine to NAS device while ripping DVD to .ISO file 12.Compress 100MB files - Using built-in ZIP compression 13.Compress 1GB files - Using built-in ZIP compression 14.Extract 100MB files - Using built-in ZIP compression 15.Extract 1GB files - Using built-in ZIP compression 16.Compress 100MB files under load - Using built-in ZIP compression while ripping DVD to .ISO file 17.Compress 1GB files under load - Using built-in ZIP compression while ripping DVD to .ISO file 18.Extract 100MB files under load - Using built-in ZIP compression while ripping DVD to .ISO file 19.Extract 1GB files under load - Using built-in ZIP compression while ripping DVD to .ISO file 20.Install Office 2007 - Ultimate version, from DVD 21.Open 10 page Word doc - Text only 22.Open 100 page Word doc - Text and images 23.Open simple Excel doc - Basic formatting 24.Open complex Excel doc - Including formula and charts 25.Burn DVD - Win 7 beta 1 .ISO to disc using CDBurnerXP 26.Open 10 page PDF - Text only, using latest Adobe Reader 8 27.Open 100 page PDF - Text and images, using latest Adobe Reader 8 These series of tests will pitch Windows 7 build 7057, 7048 (64-bit) and 7000 32/64-bit against Windows Vista SP1 32-bit and Windows XP SP3 32-bit. The scoring The scoring system that I use seems to have confused some readers. It’s actually very simple. We run each test for each OS in turn and the time taken to complete the task is noted (average of three runs). The fastest OS is given a score of 1, the runner ups 2, 3, 4, and respectively and the slowest OS scores a 6. The scores are added up and the OS with the lowest score (that is, the one that performed the best overall) at the end is the winner. After I carried out the performance test on Widows 7 build 7048 I commented that I thought some of the results were all over the place. However, on looking at this latest build, I’ve no doubt that not only will Windows 7 will be fast. In fact it will blow away all previous Windows OSes, including the sainted Windows XP. So far it seems that on the hardware platforms that I’ve used that the 32-bit flavor of Windows 7 is faster than the Beta 1 build (significantly faster in some of the tests). Will this be true of the RC and RTM builds? Initially, I expect it to be the case unless you are throwing a lot of RAM and CPU at the OS. I suspect that the cause for this is immature drivers and that once we see an RC build vendors will put more effort into optimizing the drivers for both 32-bit and 64-bit. I’m still confident that we’re on target to see the Windows 7 RC appear sometime during April.
What about x64 vs x86 flavor for build 7057 ? That's top say Is x64 more stable faster than x86 ? I am testing x64 win 7 ~b7048 and it's so much faster and stable than Vista64sp1.
Did I interpret the benchmarks correctly - that as far as everyday usability, 7057 x86 soundly kicked 7057's ass? If so, that's food for thought.
yes the x86 versions of Win 7 have been performing better than the x64 versions. However since Server 2008 R2 is x64 only, I think they are just working the code a little more slowly than the x86 code since the kernel is the same for client and server again.