Hi there,
if you observe the crash stack, you get following:
...
2014-12-31T01:26:26.599Z cpu3:33184)0x41238681dc20:[0x41801eea98e3]TLBDoInvalidate@vmkernel#nover+0x697 stack: 0x100000000000000
2014-12-31T01:26:26.600Z cpu3:33184)0x41238681dea0:[0x41801f396b93]UserMem_HandleMapFault@<None>#<None>+0x92f stack: 0x41238681df00
2014-12-31T01:26:26.600Z cpu3:33184)0x41238681def0:[0x41801f3802fd]User_Exception@<None>#<None>+0x175 stack: 0xffc734f4
2014-12-31T01:26:26.600Z cpu3:33184)0x41238681df10:[0x41801ee6544a]Int14_PF@vmkernel#nover+0x122 stack: 0x218248d8
...
I am more inclined to believe that this is a memory mapping issue. First there is an interrupt 14 - underscore PF might be page file? Then an exception is thrown because of User Memory Mapping Fault, and eventually the Translation Lookaside Buffer that takes care of translating the Physical to Virtual memory is invalidated, causing the PSOD. How sized are your ESXi hosts in terms of memory and with what memory type? Is your system multisocketed? Are you overprovisioning the RAM? Are the hosts swapping while the memory is stressed out?
Anyways as the conversation stated, turn off or if possible outright disable all power saving features in BIOS, I'd especially take a look into some memory saving features if available (some boards can undervolt the idle modules). Set system profile to high performance, disable enhanced halt, C-states -therefore:
P-State Coordination:Hardware
TM1:Disable (if you have good cooling, this is not needed and Hardware Monitoring inside vCenter client will take care of reporting high temperature to you)
TM2:Adaptive Throttling (I have no idea what LEM throttling is)
CPU C-States:Disable (These include low-power CPU states & support stuff such as sleep and hibernation which you don't need)
Enhanced Halt State:Disable (C1E always caused me more trouble than benefit)
ACPI C2:Disable (Another C-state, disable it)
ACPI 3.0 T-States:Disable (see above)
Extended APIC: Disable (see above)