search for: fextend

...[(set (v4f16 VF16:$dst), (fround (v4f32 VF32:$src)))] and 'CONV_v4f32_v4f16'. What adjustments do I need to make to the TD descriptions to make these work again? I know that FP16 is not hugely common on the mainstream platforms, but it is vital to ours. I did notice that 'fextend' was replaced by 'fpextend', is there some new ISD node type I should use for 'fround'? Thanks, MartinO -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20170123/7335c...

...re defined. " class S_HF__HF< string asmstr> : Intrinsic <[llvm_float_ty ], [llvm_float_ty ], [IntrNoMem], !strconcat(asmstr, "_f16")>; def :Pat<( f16 (int_my_math_f16 f16:$src)), (F2Hsr (FEXTsr f16:$src) )>; “ where FEXTsr is implementing the fextend type profile, F2Hsr is implementing as the float to half conversion . “int_my_math_f16” is implementing the “S_HF__HF” profile above. I am just trying to (1) convert the $src from f16 to f32 using FEXTsr. (2) use the F2Hsr to convert the f32 back to f16. for testing. however, I always got the...

...> > I think it's just so that we have a way to actually write out the > symbol into the assembly file. What does gcc do? > > -Eli > > It emits the high bits literally. The consequence is that UTF-8-encoded identifiers come out in UTF-8: scshunt at natural-flavours:~$ gcc -fextended-identifiers -std=c99 -x c -c -o test.o - int i\u03bb; scshunt at natural-flavours:~$ nm test.o 00000004 C iλ scshunt at natural-flavours:~$ As you can see, the nm output includes the literal lambda. Sean -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://...

Why is it that high (>127) bytes in symbol names get mangled by LLVM into _XX_, where XX is the hex representation of the character? Is this required by ELF or some similar standard? This behavior is inconsistent with GCC. Sean -------------- next part -------------- An HTML attachment was scrubbed... URL:

On Fri, Mar 30, 2012 at 12:12 PM, Sean Hunt <scshunt at csclub.uwaterloo.ca> wrote: > Why is it that high (>127) bytes in symbol names get mangled by LLVM into > _XX_, where XX is the hex representation of the character? Is this required > by ELF or some similar standard? This behavior is inconsistent with GCC. I think it's just so that we have a way to actually write out

...that we have a way to actually write out the >> symbol into the assembly file.  What does gcc do? >> >> -Eli >> > > It emits the high bits literally. The consequence is that UTF-8-encoded > identifiers come out in UTF-8: > > scshunt at natural-flavours:~$ gcc -fextended-identifiers -std=c99 -x c -c -o > test.o - > int i\u03bb; > scshunt at natural-flavours:~$ nm test.o > 00000004 C iλ > scshunt at natural-flavours:~$ > > As you can see, the nm output includes the literal lambda. Okay... then we should probably support that as well. Might n...