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

Krzysztof Parzyszek kparzysz at codeaurora.org
Wed Nov 14 12:43:03 PST 2012

On 11/14/2012 2:28 PM, Michael Ilseman wrote:
> I think I missed what problem we're trying to solve here.

I'm guessing that the actual encoding may change.  If we know that there 
are certain unused bits, we don't need to store them.  If the bits are 
used, we have to keep them.  I think someone was working on some sort of 
bitcode compression scheme, and in that context the difference may be 


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

More information about the llvm-dev mailing list