Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Reduce memory decoding #968

Merged
merged 8 commits into from
Aug 7, 2017
Merged

Reduce memory decoding #968

merged 8 commits into from
Aug 7, 2017

Commits on Aug 7, 2017

  1. Decrease memory consumption for whole image single tile decoding. 

    We can use the same buffer for the tile decoding and the final image, and
save the intermediate buffer to transfer between those.

Effect on the decoding of MAPA (9944 x 13498 x 3 components of size byte)

Peak memory from 4.5 GB to 2.7 GB

Now:
n5: 2699708767 (heap allocation functions) malloc/new/new[], --alloc-fns, etc.
 n1: 1610689344 0x4E77E07: opj_aligned_malloc (opj_malloc.c:61) <-- final image
  n1: 1610689344 0x4E7195B: opj_alloc_tile_component_data (tcd.c:676)
   n1: 1610689344 0x4E722D2: opj_tcd_init_decode_tile (tcd.c:816)
    n1: 1610689344 0x4E4BCF1: opj_j2k_read_tile_header (j2k.c:8597)
     n1: 1610689344 0x4E4C742: opj_j2k_decode_tiles (j2k.c:10324)
      n1: 1610689344 0x4E4E20E: opj_j2k_decode (j2k.c:7826)
       n1: 1610689344 0x4E52E42: opj_jp2_decode (jp2.c:1564)
        n0: 1610689344 0x40369E: main (opj_decompress.c:1459)
 n1: 815554560 0x4E72231: opj_tcd_init_decode_tile (tcd.c:1217) <-- working memory for code blocks: 9944*13498/64/64*8192*3
  n1: 815554560 0x4E4BCF1: opj_j2k_read_tile_header (j2k.c:8597)
   n1: 815554560 0x4E4C742: opj_j2k_decode_tiles (j2k.c:10324)
    n1: 815554560 0x4E4E20E: opj_j2k_decode (j2k.c:7826)
     n1: 815554560 0x4E52E42: opj_jp2_decode (jp2.c:1564)
      n0: 815554560 0x40369E: main (opj_decompress.c:1459)
 n1: 219758391 0x4E4C0BF: opj_j2k_read_tile_header (j2k.c:4661) <-- ingestion of code stream
  n1: 219758391 0x4E4C742: opj_j2k_decode_tiles (j2k.c:10324)
   n1: 219758391 0x4E4E20E: opj_j2k_decode (j2k.c:7826)
    n1: 219758391 0x4E52E42: opj_jp2_decode (jp2.c:1564)
     n0: 219758391 0x40369E: main (opj_decompress.c:1459)
 n1: 39822000 0x4E7224F: opj_tcd_init_decode_tile (tcd.c:1224) <-- OPJ_J2K_DEFAULT_NB_SEGS*sizeof(opj_tcd_seg_t) per codeblock
  n1: 39822000 0x4E4BCF1: opj_j2k_read_tile_header (j2k.c:8597)
   n1: 39822000 0x4E4C742: opj_j2k_decode_tiles (j2k.c:10324)
    n1: 39822000 0x4E4E20E: opj_j2k_decode (j2k.c:7826)
     n1: 39822000 0x4E52E42: opj_jp2_decode (jp2.c:1564)
      n0: 39822000 0x40369E: main (opj_decompress.c:1459)
 n0: 13884472 in 49 places, all below massif's threshold (1.00%)

Before:
n5: 4493329848 (heap allocation functions) malloc/new/new[], --alloc-fns, etc.
 n2: 1610709160 0x4E77C87: opj_aligned_malloc (opj_malloc.c:61)
  n1: 1610689344 0x4E717DB: opj_alloc_tile_component_data (tcd.c:676)
   n1: 1610689344 0x4E72152: opj_tcd_init_decode_tile (tcd.c:816)
    n1: 1610689344 0x4E4BCF1: opj_j2k_read_tile_header (j2k.c:8597)
     n1: 1610689344 0x4E4C64A: opj_j2k_decode_tiles (j2k.c:10318)
      n1: 1610689344 0x4E4E08E: opj_j2k_decode (j2k.c:7826)
       n1: 1610689344 0x4E52CC2: opj_jp2_decode (jp2.c:1564)
        n0: 1610689344 0x40369E: main (opj_decompress.c:1459)
  n0: 19816 in 2 places, all below massif's threshold (1.00%)
 n1: 1610689344 0x4E43F36: opj_j2k_update_image_data.isra.7 (j2k.c:8743)
  n1: 1610689344 0x4E4C5C1: opj_j2k_decode_tiles (j2k.c:10358)
   n1: 1610689344 0x4E4E08E: opj_j2k_decode (j2k.c:7826)
    n1: 1610689344 0x4E52CC2: opj_jp2_decode (jp2.c:1564)
     n0: 1610689344 0x40369E: main (opj_decompress.c:1459)
 n1: 815554560 0x4E720B1: opj_tcd_init_decode_tile (tcd.c:1217)
  n1: 815554560 0x4E4BCF1: opj_j2k_read_tile_header (j2k.c:8597)
   n1: 815554560 0x4E4C64A: opj_j2k_decode_tiles (j2k.c:10318)
    n1: 815554560 0x4E4E08E: opj_j2k_decode (j2k.c:7826)
     n1: 815554560 0x4E52CC2: opj_jp2_decode (jp2.c:1564)
      n0: 815554560 0x40369E: main (opj_decompress.c:1459)
 n1: 402672336 0x4E4C545: opj_j2k_decode_tiles (j2k.c:10336)
  n1: 402672336 0x4E4E08E: opj_j2k_decode (j2k.c:7826)
   n1: 402672336 0x4E52CC2: opj_jp2_decode (jp2.c:1564)
    n0: 402672336 0x40369E: main (opj_decompress.c:1459)
 n0: 53704448 in 58 places, all below massif's threshold (1.00%)
    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    793edc3 View commit details
    Browse the repository at this point in the history

  2. Fix crash on Windows due to b7594c0 

    b7594c0 may put opj_tcd_tilecomp_t->data
allocated by opj_alloc_tile_component_data() as the image->comps[].data. As
opj_alloc_tile_component_data() use opj_aligned_malloc() we must be sure to
ue opj_alined_malloc()/_free() in all places where we alloc/free
image->comps[].data.

Note: this might have some compatibility impact in case user code does itself
the allocation/free of image->comps[].data
    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    61fb5dd View commit details
    Browse the repository at this point in the history

  3. Add opj_image_data_alloc() / opj_image_data_free() 

    As bin/common/color.c used to directly call malloc()/free(), we need
to export functions dedicated to allocating/freeing image component data.
    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    f58aab9 View commit details
    Browse the repository at this point in the history

  4. Complementary fix to previous commit

    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    0c1fc05 View commit details
    Browse the repository at this point in the history

  5. opj_jp2_apply_pclr() also needs to use opj_image_data_alloc/opj_image… 

    …_data_free
    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    434ace4 View commit details
    Browse the repository at this point in the history

  6. Add documentation for magic values in the code

    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    373520d View commit details
    Browse the repository at this point in the history

  7. Decoding: do not allocate memory for the codestream of each codeblock 

    Currently we allocate at least 8192 bytes for each codeblock, and copy
the relevant parts of the codestream in that per-codeblock buffer as we
decode packets.
As the whole codestream for the tile is ingested in memory and alive
during the decoding, we can directly point to it instead of copying. But
to do that, we need an intermediate concept, a 'chunk' of code-stream segment,
given that segments may be made of data at different places in the code-stream
when quality layers are used.

With that change, the decoding of MAPA_005.jp2 goes down from the previous
improvement of 2.7 GB down to 1.9 GB.

New profile:

n4: 1885648469 (heap allocation functions) malloc/new/new[], --alloc-fns, etc.
 n1: 1610689344 0x4E78287: opj_aligned_malloc (opj_malloc.c:61)
  n1: 1610689344 0x4E71D7B: opj_alloc_tile_component_data (tcd.c:676)
   n1: 1610689344 0x4E7272C: opj_tcd_init_decode_tile (tcd.c:816)
    n1: 1610689344 0x4E4BDD9: opj_j2k_read_tile_header (j2k.c:8618)
     n1: 1610689344 0x4E4C8A2: opj_j2k_decode_tiles (j2k.c:10349)
      n1: 1610689344 0x4E4E36E: opj_j2k_decode (j2k.c:7847)
       n1: 1610689344 0x4E52FA2: opj_jp2_decode (jp2.c:1564)
        n0: 1610689344 0x40374E: main (opj_decompress.c:1459)
 n1: 219232541 0x4E4BBF0: opj_j2k_read_tile_header (j2k.c:4685)
  n1: 219232541 0x4E4C8A2: opj_j2k_decode_tiles (j2k.c:10349)
   n1: 219232541 0x4E4E36E: opj_j2k_decode (j2k.c:7847)
    n1: 219232541 0x4E52FA2: opj_jp2_decode (jp2.c:1564)
     n0: 219232541 0x40374E: main (opj_decompress.c:1459)
 n1: 39822000 0x4E727A9: opj_tcd_init_decode_tile (tcd.c:1219)
  n1: 39822000 0x4E4BDD9: opj_j2k_read_tile_header (j2k.c:8618)
   n1: 39822000 0x4E4C8A2: opj_j2k_decode_tiles (j2k.c:10349)
    n1: 39822000 0x4E4E36E: opj_j2k_decode (j2k.c:7847)
     n1: 39822000 0x4E52FA2: opj_jp2_decode (jp2.c:1564)
      n0: 39822000 0x40374E: main (opj_decompress.c:1459)
 n0: 15904584 in 52 places, all below massif's threshold (1.00%)
    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    ca34d13 View commit details
    Browse the repository at this point in the history

  8. Slight improvement in management of code block chunks 

    Instead of having the chunk array at the segment level, we can move it down to
the codeblock itself since segments are filled in sequential order.
Limit the number of memory allocation, and decrease slightly the memory usage.

On MAPA_005.jp2

n4: 1871312549 (heap allocation functions) malloc/new/new[], --alloc-fns, etc.
 n1: 1610689344 0x4E781E7: opj_aligned_malloc (opj_malloc.c:61)
  n1: 1610689344 0x4E71D1B: opj_alloc_tile_component_data (tcd.c:676)
   n1: 1610689344 0x4E726CF: opj_tcd_init_decode_tile (tcd.c:816)
    n1: 1610689344 0x4E4BE39: opj_j2k_read_tile_header (j2k.c:8617)
     n1: 1610689344 0x4E4C902: opj_j2k_decode_tiles (j2k.c:10348)
      n1: 1610689344 0x4E4E3CE: opj_j2k_decode (j2k.c:7846)
       n1: 1610689344 0x4E53002: opj_jp2_decode (jp2.c:1564)
        n0: 1610689344 0x40374E: main (opj_decompress.c:1459)
 n1: 219232541 0x4E4BC50: opj_j2k_read_tile_header (j2k.c:4683)
  n1: 219232541 0x4E4C902: opj_j2k_decode_tiles (j2k.c:10348)
   n1: 219232541 0x4E4E3CE: opj_j2k_decode (j2k.c:7846)
    n1: 219232541 0x4E53002: opj_jp2_decode (jp2.c:1564)
     n0: 219232541 0x40374E: main (opj_decompress.c:1459)
 n1: 23893200 0x4E72735: opj_tcd_init_decode_tile (tcd.c:1225)
  n1: 23893200 0x4E4BE39: opj_j2k_read_tile_header (j2k.c:8617)
   n1: 23893200 0x4E4C902: opj_j2k_decode_tiles (j2k.c:10348)
    n1: 23893200 0x4E4E3CE: opj_j2k_decode (j2k.c:7846)
     n1: 23893200 0x4E53002: opj_jp2_decode (jp2.c:1564)
      n0: 23893200 0x40374E: main (opj_decompress.c:1459)
 n0: 17497464 in 52 places, all below massif's threshold (1.00%)
    @rouault
    rouault committed Aug 7, 2017
    Configuration menu
    Copy the full SHA
    9211469 View commit details
    Browse the repository at this point in the history