LinuxQuestions.org - Mandriva 2005 - e1000 high load problem?

A few days ago we started getting problems with one of our servers:
Asus AP1600-R a 1U rack server with dual xeon, 4 x SCSI on raid 5 running Mandrake 2005

[root@lond1 ~]# lspci
00:00.0 Host bridge: ServerWorks CMIC-LE Host Bridge (GC-LE chipset) (rev 31)
00:00.1 Host bridge: ServerWorks CMIC-LE Host Bridge (GC-LE chipset)
00:00.2 Host bridge: ServerWorks CMIC-LE Host Bridge (GC-LE chipset)
00:02.0 Ethernet controller: Intel Corporation 82540EM Gigabit Ethernet Controller (rev 02)
00:03.0 VGA compatible controller: ATI Technologies Inc Rage XL (rev 27)
00:0f.0 ISA bridge: ServerWorks CSB5 South Bridge (rev 93)
00:0f.1 IDE interface: ServerWorks CSB5 IDE Controller (rev 93)
00:0f.2 USB Controller: ServerWorks OSB4/CSB5 OHCI USB Controller (rev 05)
00:0f.3 Host bridge: ServerWorks CSB5 LPC bridge
00:11.0 Host bridge: ServerWorks CIOB-X2 PCI-X I/O Bridge (rev 03)
00:11.2 Host bridge: ServerWorks CIOB-X2 PCI-X I/O Bridge (rev 03)
01:02.0 Ethernet controller: Intel Corporation 82544GC Gigabit Ethernet Controller (LOM) (rev 02)
01:05.0 SCSI storage controller: LSI Logic / Symbios Logic 53c1030 PCI-X Fusion-MPT Dual Ultra320 SCSI (rev 07)
01:05.1 SCSI storage controller: LSI Logic / Symbios Logic 53c1030 PCI-X Fusion-MPT Dual Ultra320 SCSI (rev 07)
02:03.0 RAID bus controller: Adaptec AAC-RAID (rev 01)

Under high load the web server doesn't completely deliver images or download files. Small are ok, but anything over 100K get stuck. Then we get loads of http processes which are effectively stuck. Larger mail attachments also get corrupted. The problem is intermittent, but is almost always there when the server is very busy. At these times, ssh also will sometimes log you off with a "corrupted MAC on input". We moved the mail server off to a backup machine.
The ssh error gave us a clue that perhaps this was related to a bug with the e1000 network card under high load.

... there are references on the net but this forum won't let me post them yet!

However, we are on a 100M switch - not gigabit, and changing the vm.min_free_kbytes setting to 16384 instead of the default 3831 does not fix it.

Any suggestions? The thing that is odd is that these issues seem to be for older releases of the kernel.
We are using : 2.6.11-6 smp

Any ideas would be really appreciated. Thanks