llvm dev - Nov 2015 - Way to specify instruction latency in itinerary scheduling model

Does anybody know how to specify instruction latency in the itinerary
scheduling model? For some reason no matter what I do I get a latency of 1.

-- 
Rail
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20151111/161bd6cb/attachment.html>

----- Original Message -----
> From: "Rail Shafigulin via llvm-dev" <llvm-dev at
lists.llvm.org>
> To: "llvm-dev" <llvm-dev at lists.llvm.org>
> Sent: Wednesday, November 11, 2015 6:34:03 PM
> Subject: [llvm-dev] Way to specify instruction latency in itinerary
scheduling model
> 
> Does anybody know how to specify instruction latency in the itinerary
> scheduling model? For some reason no matter what I do I get a
> latency of 1.
> 
I suspect this is because you're specifying the pipeline structure but not
the latencies. For example:

  InstrItinData<IIC_FPDivD,      [InstrStage<1, [A2_FU]>],
                                 [72, 0, 0]>,

The '[InstrStage<1, [A2_FU]>]' specifies the pipeline structure
(or, here, the first stage of the pipeline), and the '[72, 0, 0]'
specifies the operand latencies (output on cycle 72, reads on cycle 0).

 -Hal
> 
> --
> 
> Rail
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
> 
-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory

>
> I suspect this is because you're specifying the pipeline structure but
not
> the latencies. For example:
>
>   InstrItinData<IIC_FPDivD,      [InstrStage<1, [A2_FU]>],
>                                  [72, 0, 0]>,
>
> The '[InstrStage<1, [A2_FU]>]' specifies the pipeline
structure (or, here,
> the first stage of the pipeline), and the '[72, 0, 0]' specifies
the
> operand latencies (output on cycle 72, reads on cycle 0).
>
>  -Hal
>
> --
> Hal Finkel
> Assistant Computational Scientist
> Leadership Computing Facility
> Argonne National Laboratory
>

I see. Thank you.
-- 
R
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20151112/084a1a06/attachment.html>