[llvm-bugs] [Bug 41686] New: gather cycle counts inaccurate on Intel Hardware

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=41686

            Bug ID: 41686
           Summary: gather cycle counts inaccurate on Intel Hardware
           Product: tools
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: llvm-mca
          Assignee: unassignedbugs at nondot.org
          Reporter: jon.kennedy at intel.com
                CC: andrea.dibiagio at gmail.com, llvm-bugs at lists.llvm.org,
                    matthew.davis at sony.com

https://godbolt.org/z/yA_GLT

The above code snippet approximates the sequence of instructions (from the
inner loop) to take approx 3303 cycles for 100 iterations. This includes a
gather. 

When run on a real skylake based CPU (i7-8700K), it takes approximately ~1230
cycles.

Please update the latency/throughput numbers for the gather instruction within
llvm-mca to better reflect the real performance of the intel architectures.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20190501/dfe94a55/attachment.html>