90 day warranty is a big turnoff for a 3-4k USD item.
is there not a way to build in protection circuits to the device. i know the fpga will not self throttle (afaik) without specific bitstream support (maybe i have that wrong?) but surely there are ways to monitor critical sections of the cards (VRMs, FPGA, support chips, whatever) to downclock the device. i mean gpus are more or less bulletproof in that regard, why cant your card do the same.
seems the only thing outside of your control is ESD damage. that i can see as most miners probably ignore it.
at least offer an extended warranty for an added price.
180 days i could handle, 90 days, not so much. sounds almost like youre deliberately running beyond spec and cooling abilities and dont expect it to last.
no disrespect intended but self preservation seems like it should be built into high performance parts from day one.
EDIT btw not all miners abuse their rigs. my rigs lead a better life than my daily driver.. constant power (no cycling), better cooling, more checks for maintenance (temps, dust, cable conditions) etc.
We will be building protections into the shell. HOWEVER, that's our code. We can't control other's code or what code you use. If you run some code that has no protections built in and you damage your board; do you think we should warranty that?
Obviously trying to push 300 watts through a device designed to handle 225W is not going to end well. Another example, running the devices at the upper limits of their thermal design.