factor/basis/compiler/cfg/value-numbering/misc/misc.factor

! Copyright (C) 2010 Slava Pestov.
! See http://factorcode.org/license.txt for BSD license.
USING: accessors compiler.cfg.instructions
compiler.cfg.value-numbering.graph
compiler.cfg.value-numbering.rewrite cpu.architecture kernel ;
IN: compiler.cfg.value-numbering.misc

M: ##replace rewrite
    [ loc>> ] [ src>> vreg>insn ] bi
    dup literal-insn? [
        insn>literal dup immediate-store?
        [ swap ##replace-imm new-insn ] [ 2drop f ] if
    ] [ 2drop f ] if ;
compiler.cfg: fuse ##load-integer/##load-reference into ##replace to form ##replace-imm 2010-05-01 03:04:31 -04:00			`! Copyright (C) 2010 Slava Pestov.`
			`! See http://factorcode.org/license.txt for BSD license.`
compiler: cleanup usings. 2014-12-13 19:10:21 -05:00			`USING: accessors compiler.cfg.instructions`
compiler.cfg: fuse ##load-integer/##load-reference into ##replace to form ##replace-imm 2010-05-01 03:04:31 -04:00			`compiler.cfg.value-numbering.graph`
compiler: cleanup usings. 2014-12-13 19:10:21 -05:00			`compiler.cfg.value-numbering.rewrite cpu.architecture kernel ;`
compiler.cfg: fuse ##load-integer/##load-reference into ##replace to form ##replace-imm 2010-05-01 03:04:31 -04:00			`IN: compiler.cfg.value-numbering.misc`

			`M: ##replace rewrite`
compiler: new "binary literal area" at the end of a word's machine code stores constant floats and SIMD vectors; this allows ##load-reference/##load-memory fusion to be performed on x86-64, with a RIP-relative address reaching the data; also simplifies VM since custom relocation types used by the previous 32-bit-only optimization are no longer needed 2010-05-03 17:23:03 -04:00			`[ loc>> ] [ src>> vreg>insn ] bi`
			`dup literal-insn? [`
			`insn>literal dup immediate-store?`
Now that #foo and ##foo are symbols we can remove a bunch of \ 2011-11-13 01:04:26 -05:00			`[ swap ##replace-imm new-insn ] [ 2drop f ] if`
compiler: new "binary literal area" at the end of a word's machine code stores constant floats and SIMD vectors; this allows ##load-reference/##load-memory fusion to be performed on x86-64, with a RIP-relative address reaching the data; also simplifies VM since custom relocation types used by the previous 32-bit-only optimization are no longer needed 2010-05-03 17:23:03 -04:00			`] [ 2drop f ] if ;`