I am running an FC3 server and bought 2x1GB modules of DDR memory. When I attempt to to put both !GB modules in, I either get a Kernel panic or the system goes scyitso when booting.

If either one of the modules alone it works fine. So my guess is that it has something to do with the Linux recognising it. I do compile my own kernels and ahave high 64GB memory enabled. I have a AMD Sempron 2800 System.

Probably bad memory. Try using memtest86 to check if the memory modules are working.

You just to need set HIMEM to 4 GB.

Did you make sure initrd and ramdisk is set as built-in. Also did you make an initrd file using mkinitrd utility.

you should check both the RAM's individually and the RAM slots too, in case any of them are bad and cause of the pblm.

Both modules work fine as the system boots when I try them individually. I have also ruled out a problem with DIMM slot 2 by inserting the 1GB and a 256MB module in to number 2. Again, as long as the system RAM is below 2GB is works.

I will try re-compiling the Kernel with highmem set to 4GB and recreate my initrd.

What do you mean by this?

beats me. where? ramdisk i though i saw in menuconfig.

I did an fsck on all the disks and ran the MS memory tester on the memory (all was ok). The system will now take all the memory. But, it has no become very unstable. The PC crashes about three or four times a day and i get messsages as follows:

hdc: lost interrupt
hda: lost interrupt
hdb: lost interrupt

All the IDE devices in the PC are hard disks. I have no CD-ROM or DVD drives. I have googled the error and it appears it has something to do with DMA. But I am unsure as to howtodisable DMA and the effect this is likely to have.

Are the RAM sticks the same speed? The fact that both work individually suggests that they are OK, however, if one is rated at a higher speed than the other, that's a prime cause of instability. What are the exact make/model of your sticks?

I am unsure of the exact make. I'd need to check. However, both modules are 333Mhz, 1024GB, unbuffered.

I am using a xen kernel, so I am going to try and use an unpatched kernel from kernel.org. However, as it worked before I am not sure it will make a difference

Maybe a weird idea, but check the website of the main board manufacturer for a new version of the BIOS.

Sometimes the spec for the main boards says that the latest memory dimm's are supported, but in fact hardware manufacturers almost never support the latest combinations or maximum configuration. Main board manufacturer’s support the latest hardware as it becomes more mainstream available.

I have tried various builds of kernel but I am still having now luck. The server runs Samba and its seems more prone to a crash during logoff on windows when the roaming profile is being copied back to the server.