Fashionable neural networks are rising not solely in measurement and complexity but additionally in inference time. One of the crucial efficient compression strategies — channel pruning — combats this development by eradicating channels from convolutional weights to scale back useful resource consumption. Nonetheless, eradicating channels is non-trivial for multi-branch segments of a mannequin, which may introduce additional reminiscence copies at inference time. These copies incur improve latency — a lot so, that the pruned mannequin is even slower than the unique, unpruned mannequin. As a workaround, present pruning works constrain sure channels to be pruned collectively. This totally eliminates inference-time reminiscence copies, however as we present, these constraints considerably impair accuracy. To resolve each challenges, our perception is to allow unconstrained pruning by reordering channels to reduce reminiscence copies. Utilizing this perception, we design a generic algorithm UCPE to prune fashions with any pruning sample. Critically, by eradicating constraints from present pruning heuristics, we enhance ImageNet top-1 accuracy for post-training pruning by 2.1 factors on common — benefiting pruned DenseNet (+16.9), EfficientNetV2 (+7.9), and ResNet (+6.2). Moreover, our UCPE algorithm reduces latency by as much as 52.8% in comparison with naive unconstrained pruning, practically totally eliminating reminiscence copies at inference time.