AWARE SYSTEMS
TIFF and LibTiff Mail List Archive

Thread

2021.05.02 00:36 "[Tiff] SIMD optimizations", by Larry Bank
2021.05.02 09:48 "Re: [Tiff] SIMD optimizations", by Even Rouault
2021.05.02 10:33 "Re: [Tiff] SIMD optimizations", by ZdPo Ster
2021.05.02 13:55 "Re: [Tiff] SIMD optimizations", by Bob Friesenhahn
2021.05.02 14:09 "Re: [Tiff] SIMD optimizations", by Even Rouault
2021.05.02 15:35 "Re: [Tiff] SIMD optimizations", by Larry Bank
2021.05.02 17:20 "Re: [Tiff] SIMD optimizations", by Even Rouault
2021.05.02 17:45 "Re: [Tiff] SIMD optimizations", by Larry Bank
2021.05.03 15:58 "Re: [Tiff] SIMD optimizations", by Even Rouault
2021.05.03 17:28 "Re: [Tiff] SIMD optimizations", by Larry Bank
2021.05.03 17:32 "Re: [Tiff] SIMD optimizations", by Even Rouault
2021.05.03 19:16 "Re: [Tiff] SIMD optimizations", by Jeff Breidenbach
2021.05.03 20:40 "Re: [Tiff] SIMD optimizations", by Bob Friesenhahn
2021.05.03 21:09 "Re: [Tiff] SIMD optimizations", by Even Rouault
2021.05.03 21:45 "Re: [Tiff] SIMD optimizations", by Bob Friesenhahn
2021.05.03 22:08 "Re: [Tiff] SIMD optimizations", by Even Rouault
2021.05.04 07:44 "Re: [Tiff] SIMD optimizations", by Roger Leigh
2021.05.04 07:50 "Re: [Tiff] SIMD optimizations", by Roger Leigh
2021.06.06 21:15 "Re: [Tiff] SIMD optimizations", by Roger Leigh
2021.05.03 21:44 "Re: [Tiff] SIMD optimizations", by Larry Bank
2021.05.03 21:55 "Re: [Tiff] SIMD optimizations", by Bob Friesenhahn
2021.05.03 22:50 "Re: [Tiff] SIMD optimizations", by Akira Urushibata
2021.05.04 17:34 "Re: [Tiff] SIMD optimizations", by Steve Underwood
2021.05.04 17:44 "Re: [Tiff] SIMD optimizations", by Larry Bank
2021.05.04 18:07 "Re: [Tiff] SIMD optimizations", by Toby Thain
2021.05.04 20:59 "Re: [Tiff] SIMD optimizations", by Michael Shiels
2021.05.04 21:02 "Re: [Tiff] SIMD optimizations", by Larry Bank
2021.05.05 01:14 "Re: [Tiff] SIMD optimizations", by Steve Underwood

2021.05.04 18:07 "Re: [Tiff] SIMD optimizations", by Toby Thain

I brought up this issue years ago. The libtiff G4 encoder is needlessly inefficient in both its encoding of the output bits and its searching the current and reference lines for runs of bits. There isn't just a tweak to fix it, the encoder needs to be tossed out and replaced. A faster encoder will use a bit more memory, but at this point I don't think anyone will mind using an additional 8 bytes per pixel width (current + reference line runs as uint32_t). I've been recently reminded on this mailing list that my time is better spent optimizing more important libraries like libpng. Are there any good reasons to change my mind? I still think that libtiff G4 encoding is used widely in scanned document generation (embedded in scanners and on PCs/mobile), so it should be improved.

I use G4 for every bilevel scanned document I ship, so yes. However, the existing encoder is still pretty darn fast :)

--Toby