Hello everyone, new to CUDA and to this forum too. Learning my first steps to program a financial strategy backtesting and needing some help to design the software, any idea is more than welcomed as currently learning.
What the simulator basically does is processing a strategy, with let’s say 5 decimal variables that go 0.1 to 1.0 each, in a given period of time. What’s the best way to do this? I was thinking about splitting the date between X days (one day per block) and process them in parallel but there’s information being passed between days so they need to be serialized.
So my second attempt is processing a complete date stamp per block but with a different values combinations for each block/thread.
Does it make sense? What should be a good way to dimension the block and threads?