Some questions that I have:
Are there any common software packages on the 'problem' clients that are not installed on the 'non problem' clients?
Have you taken a 'non-problem' laptop and connected it to the same port as a 'problem' client? If so, what were the results?
Are all of the 'problem' clients on the same switch or is there a mixture of both 'problem' and 'non-problem' clients across all switches?
Are you using Cisco switches? If so, are you using the monitor session command to SPAN/RSPAN ports for any reason? If not configured properly, this can significantly impact the performance of the source ports and if one of these 'problem' clients is in one of these ports it could explain the performance delay.
Many people have asked about the user account in AD, but what about the machine accounts for the 'problem' clients? Are they all in the same OU or scattered amongst multiple OU's? Do they have any GPO's over them? You can blow out a machine, even give it a different name, move the user to an OU that doesn't have any GPO's applied to it and still have issues if the default location for machine accounts still has a GPO over it. You've checked the AD user accounts, check the machine accounts as well.
How many GC's (Global Catalog servers) do you have in your domain? Do you have either the PDC Emulator, Schema Master or Infrastructure Master roles assigned to a DC that's also a GC? If you go into AD Sites and Services are you able to successfully force replication between GC's? I have seen instances of the password that GC's use to replicate becoming corrupt - causing all kinds of weird issues. I haven't seen it cause random drive mappings, but theoretically it could if those drive mappings were once part of a GPO that has not received the most recent changes because of a broken replication bond and a client hits the GC with the old GPO settings during bootup.
Only XPP. To test this I used a generic XP CD with SP1. It was not a proprietary disk that came with HP or anything like that and had extra software...nothing but XPP with SP1. From that point I installed only the NIC driver, joined the domain, added the user, logged on and tried it. I did this on both a Dell and a HP...same thing
Yes, we have done that. Actually that was done sort of immediately. The VP who got the new computer who first noticed it. used to have a Toshiba Tablet. When he got his new computer, it had this problem, so we took it back and gave him his tablet back. The new HP wouldnt work in his office, but his old Tablet does. Plus we have tested that further up here.
Well, to be honest, we dont know how many people are having the problem. Like I said, it is only a couple of folders that have the problem, so unless you click on one of these folders, you wouldnt notice it. We havent gone around testing people machines because we dont want to alert them to a problem they dont know they have. So, of the ones we know about, they are all in our area, so one our switch...however, they did not all start out on the same switch.
We are using Cisco routers and switches, but I have had no part in the configuration. We have an outside consultant who manages those. Nobody in our department knows how they are configured.
All of the users are in the same OU and there are no group policies assigned. I wish it was a little more organized than this, but it is not my call.
The last question is over my head...sorry.