Tweet
I am always hesitant to trust performance benchmarks that run within the client OS (especially Windows XP). I know from experience how much faster a computer runs with a fresh installation of Windows. Does anyone have any experience with benchmarking a stale installation of Windows (2+ years), and then retesting the same machine after a scrubbing?
-
-
If you defrag the disk and make sure you have as little as possible running in the background, then the results *should* vary only a little from a fresh install.
But often the quickest way of getting to the state of 'as little as possible running in the background' is to re-install the O/S. -
I've added and removed a good deal of software and hardware since the last full reinstall of one of my Windows XP partitions, and I've recently been checking what's loading when Windows starts. There are a good number of drivers left over from old applications and hardware, and getting rid of them seems to be difficult.Originally posted by passmark View Post
I first tried deleting using Device Manager showing hidden devices and deleting the drivers I knew were associated with long-gone hardware or applications. Most of the ones loading at boot time were in the Non-plug-and-Play devices section. Some had messages that the device was not attached; others indicated that the device was operating properly and had ...SYS driver names associated with them.
In a lot of cases, checking the ...SYS driver properties turned up software names indicating where the device came from, and Internet searches helped with some others. Unfortunately, most Internet searches turned up dozens of web sites reporting the risk of the file being associated with spyware or other malware for every site indicating what the ...SYS file belonged to and what it's function was.
Unfortunately, deleting some of the drivers resulted in Windows XP getting a BSOD for INACCESSIBLE_BOOT_DEVICE. Interestingly, all the hits I got on that text or the 0x0000007B 0x0F789E524 seemed to indicate that the cause had to be corrupted files or hardware problems; none mentioned that registry issues related to drivers loading at boot time could be a factor.
As a result, I've been working through the entries which are definitely invalid one at a time, backing up the partition so it can be restored prior to the last change if Windows XP refuses to boot.
I've found that I can disable but not delete something labelled "AMDPCI" which is left over from a previous motherboard with a different chipset. I can change a Powerquest Partition Magic boot mode driver from "boot" to "demand", but disabling or deleting it results in the BSOD even though Partition Magic was removed from Windows XP over a year ago. Symantec (which took over Partition Magic) Tech Support spent an hour working on this driver with me without being able to figure out why none of the standard approaches Symantec uses to clean up old installations had any effect on it or why Windows XP has made it essential to the boot process.
I have Windows XP Professional installed in two other partitions, a clean install of Windows XP Pro x86 with SP3 RC1 slipstreamed into it with the current hardware, and a similar clean install of Windows XP Pro x64 with its SP2 slipstreamed. Neither has any of the Device Manager Non-plug-and-play devices I've been trying to delete. It's possible that the x64 version doesn't use them and SP3 has eliminated them, but since the driver properties associate them with long-gone hardware or software, it's unlikely.
This is a long preamble to the conclusion that a well-maintained Windows install without significant hardware or software changes probably can be expected to test as well as a clean install. However, it seems to be impossible to change hardware and software without leaving some residue behind which is likely to impact test results, so if there have been changes, a clean install is probably necessary to get accurate test results for the current configuration.MikeComment
Comment