[PATCH] Add HLE target feature

Michael Liao michael.liao at intel.com
Thu Feb 28 17:20:52 PST 2013


Hi Nadav

On Thu, 2013-02-28 at 17:13 -0800, Nadav Rotem wrote:
> 
> On Feb 27, 2013, at 7:36 PM, Hal Finkel <hfinkel at anl.gov> wrote:
> 
> >  I am not sure that Michael has made a good argument that *this* set
> > of patches is the right approach (aside from saying that it mirrors
> > the interface presented by gcc), but HLE seems like a fairly limited
> > domain and there may not be too many alternatives. I agree that
> > modeling the behavior should allow for additional optimization, and
> > this will be important.
> 
> At the moment there is no concrete plan to add TM specific
> optimizations.  How about implementing HLE support as a set
> x86-specific intrinsics ? 

The reason why X86-specific intrinsic is not used for X86 HLE support is
that it will introduce significant interface change due to the number of
atomic instructions. Developer will make non-small modification to
change from previous atomic builtins provided by clang/llvm to X86
intrinsic-based HLE interface. That's why GCC community propose to add a
target hint for all atomic builtins. In addition, LLVM treat intrinsics
differenet from atomic insns, we will lose chances of optimizations.

In the other side, HLE support doesn't rely on TM support directly. Even
though I put example showing HLE could be implemented upon TM-only.
That's should be a future work once TM is within LLVM roadmap. Before
that, if a target needs provide HLE interface, it could provides them
through their native code generation.

Yours
- Michael





More information about the llvm-commits mailing list