spconv2.x版本相关问题

Breaking changes in Spconv 2.x for spconv 1.x users

1. spconv.xxx move to spconv.pytorch.xxx, change all import spconv to import spconv.pytorch as spconv and from spconv.xxx import to from spconv.pytorch.xxx import.
2. use_hash and fused_bn in Sparse Convolution is removed, we only use hash table in 2.x.
3. x.features = F.relu(x.features) now raise error. use x = x.replace_feature(F.relu(x.features)) instead.
4. weight layout has been changed to RSKC (native algorithm) or KRSC (implicit gemm), no longer RSCK (spconv 1.x). RS is kernel size, C is input channel, K is output channel. 权重布局已更改为 RSKC（本机算法）或 KRSC（隐式 gemm），不再是 RSCK（spconv 1.x）。RS是内核大小，C是输入通道，K是输出通道。
5. all util ops are removed (pillar scatter/nms/rbbox_iou…)
6. VoxelGenerator has been replaced by spconv.pytorch.utils.PointToVoxel (torch API) or Point2VoxelGPU[1-4]d/Point2VoxelCPU[1-4]d (tv.Tensor API).
7. spconv < 2.1 don’t support CPU. spconv 2.1+ support cpu for debug usage.
8. test spconv 1.x model in spconv 2.x: Firstly set environment variable before run program, Then set all algo in conv/pool to ConvAlgo.Native. Linux: export SPCONV_FILTER_HWIO="1", Windows powershell: $Env:SPCONV_FILTER_HWIO = "1". WARNING test spconv 1.x model don’t support implicit gemm algorithm.**