#10750: silly assembly for comparing Doubles -------------------------------------+------------------------------------- Reporter: rwbarton | Owner: Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 7.11 (CodeGen) | Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Type of failure: Runtime | Unknown/Multiple performance bug | Test Case: Blocked By: | Blocking: Related Tickets: #10137, #10677 | Differential Revisions: -------------------------------------+------------------------------------- Comment (by nomeata): Weird. With your example, compiled with `./inplace/bin/ghc-stage2 -fforce-recomp -O -ddump-asm T10677.hs`, I get assembly like this: {{{ ==================== Asm code ==================== .text .align 8 .long S1vB_srt-(Zero.f_info)+0 .long 0 .quad 4294967301 .quad 0 .quad 12884901903 .globl Zero.f_info .type Zero.f_info, @object Zero.f_info: _c1vs: leaq -32(%rbp),%rax cmpq %r15,%rax jb _c1vt _c1vu: movq $block_c1vm_info,-8(%rbp) movq %r14,%rax movl $GHC.Classes.$fOrdDouble_closure,%r14d movq $stg_ap_pp_info,-32(%rbp) movq %rax,-24(%rbp) movq $Zero.f3_closure+1,-16(%rbp) addq $-32,%rbp jmp GHC.Classes.>_info .text .align 8 .quad 0 .quad 32 block_c1vm_info: _c1vm: andl $7,%ebx cmpq $1,%rbx jne _c1vq _c1vp: movl $Zero.f2_closure+1,%ebx addq $8,%rbp jmp *(%rbp) _c1vq: movl $Zero.f1_closure+1,%ebx addq $8,%rbp jmp *(%rbp) _c1vt: movl $Zero.f_closure,%ebx jmp *-8(%r13) .size Zero.f_info, .-Zero.f_info }}} so unless I am reading this wrong, it did not inline specialize or `>`. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/10750#comment:4 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler