4 ms·
> Btw the performance of emulated float operations on rp2040 Vs the teensy is a factor of ~100x wow, I was going to ask whether a mult to an int before doing t
by jerrysievert 4y ago
> Btw the performance of emulated float operations on rp2040 Vs the teensy is a factor of ~100x
wow, I was going to ask whether a mult to an int before doing the DSP, then a div back to float would be faster than a softfpu, but I think that answers the question pretty well.
but, if you're looking at a teensy4, it might also be worthwhile to look at a daisy seed: https://www.electro-smith.com/daisy/daisy https://www.electro-smith.com/daisy/daisy (they also sell specific hardware if you're looking to go eurorack with it)
- runnerup 4y agoIt's still plenty fast for most real-world applications. I did a lot of work with the Cortex M3, which similarly lacks floating point. With many compilers you can even still use float() directly and the compiler will convert the floating point operations to integer math, somewhat/mostly transparently. There were a few hiccups, but for the most part you didn't have to think about it.