1; This testcase checks to make sure that the assembler can handle floating 2; point constants in IEEE hex format. This also checks that the disassembler, 3; when presented with a FP constant that cannot be represented exactly in 4; exponential form, outputs it correctly in hex format. This is a distillation 5; of the bug that was causing the Olden Health benchmark to output incorrect 6; results! 7; 8; RUN: opt -constprop -S > %t.1 < %s 9; RUN: llvm-as < %s | llvm-dis | llvm-as | opt -constprop | \ 10; RUN: llvm-dis > %t.2 11; RUN: diff %t.1 %t.2 12 13define double @test() { 14 %tmp = fmul double 7.200000e+101, 0x427F4000 ; <double> [#uses=1] 15 ret double %tmp 16} 17