Ask HN: How much did your startup spend on a failed project/experiment

I have been working on a project where I optimize number of clock cycles required to access memory. All indications are that we will get no improvement compared where we were 2-3 months ago. This means our startup wasted atleast $30k. So, how much did your startup spend before realizing something was not working?