  PASS IQ1_M   cb=28 bits=4 nibble   : 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ2_S   cb=26 bits=4 nibble   : 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ2_XS  cb=26 bits=4 nibble   : 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ2_XXS cb=29 bits=2 grid:iq2xxs: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ3_S   cb=21 bits=4 nibble   : 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ3_XXS cb=27 bits=4 nibble   : 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ4_NL  cb=20 bits=4 nibble   : 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ4_XS  cb=20 bits=4 nibble   : 256x128 through repack -> kernel -> oracle fields, exactly
  PASS Q2_K    cb=22 bits=2 bitstream: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS Q3_K    cb=23 bits=3 bitstream: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS Q4_0    cb=23 bits=4 bitstream: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS Q4_K    cb=22 bits=4 bitstream: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS Q5_K    cb=22 bits=5 bitstream: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS Q6_K    cb=23 bits=6 bitstream: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS Q8_0    cb=23 bits=8 bitstream: 256x128 through repack -> kernel -> oracle fields, exactly
  PASS IQ2_XXS cb=26 bits=4 (twin spelling) nibble   : 256x128 through repack -> kernel -> oracle fields, exactly

INTEGRATION GATE PASSED
  16 type(s) compared against a decomposition that is itself checked
  against llama.cpp's C and gguf-py, so this is NOT two of ours agreeing.
