lxo/ChangeLog

   1 2020-11-28
   2
   3         * 532: Figured out and implemented the logic to infer mode
   4         switching for best compression under attempt 1 proposed encoding,
   5         namely with 10-bit insns, 16-bit insns, 16+16-bit insns, and
   6         32-bit insns.  10-bit insns appear in uncompressed mode, and can
   7         be followed by insns in either mode; 16-bit ones appear in
   8         compressed mode, and can remain in compressed mode, or switch to
   9         uncomprssed mode for 1 insn or for good; 16+16-bit ones appear in
  10         compressed mode, and cannot switch modes; 32-bit ones appear only
  11         in uncompressed mode, or in the single-insn slot after a 16-bit
  12         that requests it.  If we find a 16-bit insn while we're in
  13         uncompressed mode, use a 10-bit nop to tentatively switch.  Insns
  14         that can be encoded in 10-bits, but appear in compressed mode, had
  15         better be encoded in 16-bits, for that offers further subsequent
  16         encoding options, without downsides for size estimation.  Insns
  17         that can be encoded as 16+16-bit decay to 32-bit if in
  18         uncompressed mode, or if, after a sequence thereof, a later insn
  19         forces a switch to 32-bit mode without an intervening switching
  20         insn.  Still missing: the code to select what insns can be encoded
  21         in what modes.  (6:42)
  22
  23         * 532: Implemented a skeleton for compression ratio estimation,
  24         initially with the simpler mode switching of the 8-bit nop,
  25         odd-address 16-bit insns.  Next, rewrite it for all the complexity
  26         of mode switching envisioned for the "attempt 1" proposal.  (2:02)
  27
  28 2020-11-23
  29
  30         * 238: Debating various possibilities of 16-bit encoding.  (5:20)
  31
  32         * 532: Wrote a histogram python script, that breaks counts down
  33         per opcode, and within them, by operands.  (2:05)
  34
  35 2020-11-22
  36
  37         * 529: Brought up the possibilities of using 8-bit nops to switch
  38         between modes, so that 16-bit insns would be at odd addresses, so
  39         that we could use the full 16-bits; of using 2-operand insns
  40         instead of 3- for 16-bit mode so as to increase the coverage of
  41         the compact encoding.
  42         * 238: Luke moved the comment above here, where it belonged.
  43         * 529: Elaborated how using actual odd-addresses for 16-bit insns
  44         would be dealt with WRT endianness.  Prompted by luke, added it to
  45         the wiki.
  46         * Wiki: Added self to team.  (11:50)
  47
  48 2020-11-21
  49
  50         * 532: Wrote patch for binutils to print insn histogram.
  51         * Mission: Restated the proposal of adding "and users" to the
  52         mission statement, next to customers, as those we wish to enable
  53         to trust our products.  (6:48)
  54
  55 2020-11-20
  56
  57         Reposted join message to the correct list.
  58         * 238: Started looking into it, from
  59         https://libre-soc.org/openpower/sv/16_bit_compressed/
  60
  61 2020-11-19
  62
  63         Joined.