Performance regression: delta calculation for large files is unacceptably slow.
Performance regression: delta calculation for large files is unacceptably slow.: a task in LegoFlow-SWE (Harbor dataset). When computing deltas for large files (e.g., 1 GiB with 1 KiB block sizes resulting in ~1M blocks), the signature matching becomes a severe bottleneck. Profiles show that the…
The task
When computing deltas for large files (e.g., 1 GiB with 1 KiB block sizes resulting in ~1M blocks), the signature matching becomes a severe bottleneck. Profiles show that the existing lookup structure leads to delta computation times around 593 seconds, whereas a properly tuned hash table can bring this down to…
Part of Lego-X/LegoFlow-SWE.