There are many post regarding nvidia issues and I hope this is not an echo. The subject is nVidia but the specifics are different in each post. This is principly regarding FC3 and a kernel issue. I hope somone can help!

My PC Specs are here.

I get kernel incompatibility issues when trying to install the driver. Nvidia-installer cannot locate a kernel appropriate so it does 'something' automatically? I don't know what, but the driver is not working properly. The X11 settings are easy to fix, my concern is in adding the right kernel to fully support the driver, otherwise I'm stuck with the software 'nv' driver.

One thing unusual about my system (coming from Mandrake) is that the install seems to be in some kind of logical partition.

I have a maxtor 40Gb ATA133 and a 120Gb SATA 150 (Reiser). I think the default install has put the boot sector on the 40Gb drive (left the remainder empty) and installed root and all the system on my Reiser partition (120Gb Maxtor).

When I boot I briefly see this information:

VolGroup00 and lvm2 I don't know what this means?

These images are a visual interpretation of what I think the install has done.

http:///www.kabuse.com/Files/qtparted01.png
http:///www.kabuse.com/Files/qtparted02.png

Can anyone interpret my install? If so, what is my next step with a kernel patch? I was hoping to be able to just download the kernel src from nvidia and compile it. This arrangement leaves me somewhat blank, especially with the strange use of disc space?

Hmm. It sounds confusing.

The NVIDIA drivers have two parts: binary and source. The binary part is the main driver which is not released under open source. The source part is the part that get built specific to your system so that the binary NVIDIA drivers and your kernel can be "glued together". So the source part is simply an interface.
The installation comes with a few pre-made interfaces for popular kernel builds (you'd think that FC3 would be one of them), so it compares your kernel with those first. Then if it can't find a match (I've never got it to), it'll search some online repositor for a match (again, it's never found one for me). If that doesn't work, it's only alternative is to build an interface for your specific kernel. Now, I haven't figured out what it needs exactly, but it searches /usr/src/linux by default and checks your kernel source for information that it needs to build the interface. But you can't just have the kernel source, you actually have to have it built. So, if you can find a FC3 package (probably called kernel-source or something) which will give you the kernerl source tree, you could install that package, which will put the kernel source in /usr/src/linux. Then you actually need to compile the kernel, as-is, but not install. Then the nvidia installer should work for your kernel.

Or you could do it the easy way:

I got that from fedorafaq.org

I hope this helps clearify things.

As far as the diskspace issue, there are better ways of doing things. A lot of distributions will make the /boot directory on a separate partition. Generally, you only need a few megs on that partition (mine's only about 32MB). So that's normal. And then it's also normal to fill up the rest of the drive with everything else (some people make /usr a separate partition as it can contain a lot of your programs and whatnot. But I don't recommend this unless you have a special need to do so). Your installation should not have put them on two separate drives though. You can still just create a new partition for the rest of hda and format it, mount it somewhere and use it (for instance, I have a separate partition where all my media files are because I have more than one distribution that will use it). If this is a fresh install, maybe you should reinstall and manually create the partitions to your liking.

-- the dudeman