Simeon H.K. Fitch

> Mb you just need to try some more cats ;) For me I'm fine with cats (I do need to use it more). I just know that not all...

@echeipesh Great followup. > Java-like API around performance and IO edges ^---- Would also make Python bindings a heck of a lot easier.

PS: I hope I've not come across as anti-`cats`, `fs2`, `IO` or whatever.... I think all that performance heavy stuff should use whatever abstraction lends to the best 1) maintainability,...

Long shot, but this issue made me think of https://github.com/locationtech/geotrellis/issues/2358 . I also have a vague memory of the time domain resolution having a big impact on memory usage, but...

@pomadchin Something tells me [this](https://github.com/locationtech/geotrellis/blob/422e6d5c08096d41e30499b0d4dd82c58215a912/store/src/main/scala/geotrellis/store/index/zcurve/Z3.scala#L126) was the killer op. That and the `toSeq`.... and the recursion... I thought GT used SFCurve? Does this op do basically the same thing [as...

@echeipesh To what degree does this get addressed with CoGs? Our number 1 or 2 performance problem right now is S3 I/O...

Possible pathway: http://nd4j.org/gpu_native_backends c.f.: https://github.com/locationtech/geotrellis/issues/39#issuecomment-336527088

@pomadchin Kryo *is* faster at the serialization part, once the requisite codec classes are constructed. IOW, the `newInstance` call here is very expensive in comparison to the actual serialization: https://github.com/locationtech/geotrellis/blob/f8af6e6d72749b8fb175a0c44cbc7d027a940686/spark/src/main/scala/geotrellis/spark/util/KryoSerializer.scala#L46...

Turns out that the `Pool` class listed in the Kryo docs doesn't exist in the version of Kryo that Spark uses, and in an initial test using an updated version...

@pomadchin Thanks for the head's up. It's been a long time coming, and definitely needed. Not sure when I'll get to an RF build tho. :/