Beyond Compare vs WinDiff

[pb445]

For years, I "made do" with WinDiff.exe until I recently purchased Beyond Compare. Being a Microsoft freebie, WinDiff is very limited alongside Beyond Compare. However, I was wondering if anyone knows whether WinDiff uses CRC comparison, or whether it performs an actual binary comparison? I know that WinDiff will report differences in file size and time/date stamp.

Beyond Compare's default comparison criteria makes for very rapid comparisons. Comparing based on CRC or binary comparison obviously takes a lot longer, in return for the certainty that those kinds of comparisons offer.

When I used WinDiff, I found that it chugs along and "does its thing as best it can"... it's not pretty, but it gets there... eventually. Now with Beyond Compare, I have a wider range of comparison methods.

I wonder has anyone "compared" Beyond Compare's performance with WinDiff's performance over a given number and size of test files?

Reason I ask is that I have a share with hundreds of thousands of files that is to be relocated, and before I do away with "the old", I want to complete a comparison to know that "the new" is an exact duplicate. And obviously I want to use whichever utility will complete the comparison quickest and with at least CRC comparison.

Thanks for any insight anyone may be able to offer here.

PB

[Zoë]

Re: Beyond Compare vs WinDiff

I just checked and WinDiff appears to do a CRC comparison, and not a binary comparison.

Speed-wise WinDiff and BC appear to take about the same amount of time to compare files. The best I can do right now is my normal comparison set, which is 10MB in 443 files on each side. If the system doesn't have any of the files cached it takes ~9 seconds to compare the sets in both BC and WinDiff, and if they are cached it's around 1/3 second. The actual disk access is such a significant factor that I would be surprised if WinDiff and BC compared significantly differently if you use a CRC comparison in BC.

Caching, read-ahead, and disk head movement can have a significant effect on comparison speeds, so I'd suggest trying them both on a subset (say 3,000-10,000 files) to get a good idea of the performance you're looking at. Depending on the disks involved CRC comparisons can be faster than binary comparisons, but it's generally only significant for medium to large files (> 1MB), and the speed increase isn't guaranteed. I have seen cases where binary comparisons significantly outperform CRC comparisons too, which is why I'd recommend testing it first.