Description of problem: This probably is NOT a kernel problem but I have no idea where the problem is so it is going here for now. I have noticed (and I have read articles in which others have commented on this) that Red Hat 9 takes a very long time to launch applications. For example, launching OpenOffice for the first time can take a minute or more on a relatively new Pentium 4 laptop. I have also noticed similarly slow performance while copying large numbers of files. Doing hdparm indicates that I am getting a decent 15-16 MB/sec, so DMA is working fine. The reason I post this as a problem is that under Mandrake 9.1 on the same machine, performance is a lot faster. About 25 seconds to launch OpenOffice (writer). Mandrake is using a slightly newer kernel but I doubt this is the problem. Any thoughts? Why is this? In many ways I would prefer to use Red Hat 9 since it seems to be a bit more stable than Mandrake 9.1, but the slow performance is a killer. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
I was discussing this problem with a friend of mine familiar with kernel development and he thought that it might actually be kernel related. Newer kernels have added optimizations allowing for better real-time response, which have to do with allocating more time to certain types of foreground I/O operations. Basically, I think what might be at issue here is foreground application priority (including giving timeslices to launching applications).
AHA! I discovered the answer to this problem!! After nosing around the Net for a while I discovered it: the problem is magicdev. Remove the magicdev package, and suddenly Open Office goes from 100 seconds to launch to 15 seconds. Clocked. Yes, that's 100 seconds down to 15 seconds. And I have read other people reporting the same thing. There is something seriously wrong with magicdev at least on some systems (mine is an Inspiron 8200). Magicdev should definitely be reexamined. This bug should be reassigned to someone who knows about magicdev.
*** This bug has been marked as a duplicate of 73661 ***
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.