[LLVMdev] [RFC] Extend LLVM IR to express "fast-math" at a per-instruction level

[Krzysztof Parzyszek]

On 11/14/2012 9:37 PM, Chandler Carruth wrote:
>
> I'm really not trying to rehash a discussion in too much depth, but I
> have to wonder with all of this -- why not use metadata as the
> *encoding* mechanism for these flags?

I'm not familiar with the decoding algorithm, but the first thing that 
comes to my mind is that wouldn't it make decoding of an instruction a 
two-step process?  I.e.
(1) decode the instruction,
(2) update the instruction after metadata has been decoded.

-Krzysztof

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, 
hosted by The Linux Foundation