I was thinking of experimenting with something and I was curious to here if any of you had any experience with it. I'm relatively new to overclocking and undervolting but it has proven to be far simpler than I thought it was using the frequency voltage curve in MSI afterburner. Now it seems like you pick either a target voltage for your GPU to run at for the thermal properties, or pick a target frequency for your desired performance if you have thermal headroom and adjust the voltage from there. However, in most guides I've seen, people focus on this high voltage/frequency target exclusively, even though the shape of the curve leading up to that target does seem to have a notable affect on performance. My idea: can we test the maximum stable frequency at every voltage up to our target voltage to get the most efficient curve? Ie, start undervolted to the absolute minimal voltage, with everything to the right of that set below that. This will be a massive underclock obviously, but we can adjust the frequency up at this minimal voltage until it becomes unstable, bump the frequency down until it's stable and then move on to the next lowest voltage. We repeat this with the next lowest voltage, using the frequency at the first voltage that we determined in the previous step. That way we know any instability arising is due to the the furthest point to the right only, not anything else. We can repeat this for all voltages up to a target voltage, which we choose based on thermal performance. This gives us a curve that allows us to run as efficiently as possible at every possible voltage, and ensures that we are getting stability at every point in the curve. Obviously this is extremely time consuming as you have to go through and adjust the curve and stability test at every voltage point, but hopefully the results would be worth it. My next question is: how far is it safe to underclock? Ie, am I able to start this process at the very lowest voltage on the MSI afterburner graph and go all the way up to the target voltage, or is it necessary that I pick some higher voltage starting point and assume a stable clock rate for everything to the left of it? I'd like to hear your input. If it's safe to underclock all the way down so that it's stable at minimal voltage, I'll be happy to put in the time to test and see what results of can get, and get back to you about it. It is also possible that this curve sculpting has extremely diminishing returns and this will not be worth the time at all, but seeing as in sffpc we like to push our components to the absolute power/thermal limit, I'd like to complete this, if only for science!