1 Comment
User's avatar
Yellow Tail Tech's avatar

Fascinating to see writeback finally get a parallelism boost. Do you think per-allocation-group parallelization might still outperform the current inode-affinity approach?

Expand full comment