2021.05.02 13:55 "Re: [Tiff] SIMD optimizations", by Bob Friesenhahn

- regarding performance gains, I only got a 5% improvement however on a 8000x8000 large image, lzw compressed, predictor 2, tiles of 512x512, 4 channels, interleaved, 16bit, compression ratio (uncompressed_size/compressed_size) of 1.6. That seems a bit too low to justify all the above efforts.

The benefits of any optimizations can only be evaluated given a standard set of test cases (a performance test suite) which can be replicated on any system.

CPU and system capabilities vary widely so something which substantially improves performance on one CPU may substantially diminish performance on a different CPU. Even a small difference in image size or parameters can make a big difference given the CPU (e.g. due to overflowing a cache).

The libtiff utilities are mostly not very useful for performance testing since they do something different than common usages.

In summary, it would be good for a portable performance test suite to be developed and contributed to libtiff before spending a lot of time working on optimizations.

Bob

Bob Friesenhahn
bfriesen@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/
Public Key, http://www.simplesystems.org/users/bfriesen/public-key.txt