> when will Nvidia allow to seamlessly offload Java or another GC based language to the GPU? Lots of garbage collected languages seamlessly target GPUs. This i

> when will Nvidia allow to seamlessly offload Java or another GC based language to the GPU? Lots of garbage collected languages seamlessly target GPUs. This is typically done at the library level, either within the language ecosystem (eg in Python using Cupy instead of Numpy: https://towardsdatascience.com/heres-how-to-use-cupy-to-make-numpy-700x-faster-4b920dda1f56 https://towardsdatascience.com/heres-how-to-use-cupy-to-make...) or below it (using cuBLAS as your BLAS implementation: https://developer.nvidia.com/cublas https://developer.nvidia.com/cublas) Java can do this too - something like ArrayFire is reasonably popular: https://developer.nvidia.com/arrayfire https://developer.nvidia.com/arrayfire