any tips on how to do realtime physics simulations of supernova on consumer grade hardware? in the case of 10^53 particles in one system would it be feasible?
>>107684758Realistically.
>>107684758>any tipsyeah, expect results in a years time
>>107684758first thing: but an RTX 5090
>>107684758>any tips on how to do realtime physics simulations of supernova on consumer grade hardware? in the case of 10^53 particles in one system would it be feasible?No
>>107684758at what point do science fags finally accept that the Universe is a machine beyond the human brain to comprehend?
>>107684758>asking actual question on a website designed to contain retardsyou are better off asking in sci, fglt or programming threadthese are the only places where there is a small chance you will get an answer
>>107685427Lying to the public, at least one tip got posted prior to you. :)
>>107684758>on consumer grade hardwareWell how many decades are you willing to wait or how many thousands of 4090s and mellanox connectx-8 are you willing to buy?
>>107685758>>107684826you arent important enough to have a name hereyoure also new to /g/ little techlet
>>107684758In case you're serious, you need to lower your particle counts. If every particle is simulated, you won't have enough bits in a consumer system ram for each particle. And if you want realtime, you're going to have to sacrifice accuracy. You won't get anything realistic.
>>107685758No you are not simulating 8Peta particles in real time.
>>107685881You don't know that. Tech is getting better and better every year. I'm sure we're close to being able to do that. On my half-decade old 512gb 64-core threadripper pro workstation (not an epyc, because that's not a consumer system), I'm simulating 40^12 particles *faster* than realtime. I'm simulating them incorrectly, yes, but nowhere did op require the simulation to be physically accurate. And this is with half-decade old hardware. Imagine what you can do with today's hardware.
>>107685858I'm also trans, if that helps
>>107686053please explain further huff huff
>>107684758Houdini is capable of doing this. You can even download a free version. Of course it's not something what you'll accomplish in couple of evenings. Retards like you never do anything but just ask time wasting questions.
>>107684758it wouldn't be useful to simulate every single particle but you can compress it into meta particles that represent the average state of a cluster of a hypothetical quantity of particles which becomes a useful and doable approximation.
>>107685950software is getting worse at twice the rate hardware is getting better
>>107686433okay that makes the most sense out of all the suggestions. to what extent can you group up clusters of particles as metaparticles before losing the accuracy of the simulation altogether? is it something that can only be determined by trial and error or is there a methodology to anticipate the errors?
>>107686455well, a supernova has non-linear elements to it which you may or may not care about in your simulation and i'm guessing you want something that isn't just pretty to look at. i'd look at SPH which is commonly used for these things, not just fluids. i don't think you'll get away without manual tuning but you can get a decent close approximation to start with by supplying sane parameters and a lot of the parameters can be trivially derived, like density and pressure forces.. viscosity, not so much.
>>107684758You aren't getting anything done on consumer hardware this decade or the next one. Realistically you'd pay for supercomputer/quantum computer hours
