From: Andrea Di Biagio Date: Wed, 19 Jun 2019 16:10:58 +0000 (+0000) Subject: [llvm-mca][docs] clarify how the quality of the perf report is affected by the qualit... X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=daec1e09d338c0af74886ef8c71a64a3cfe1592b;p=llvm [llvm-mca][docs] clarify how the quality of the perf report is affected by the quality of the scheduling models. Differential Revision: https://reviews.llvm.org/D63556 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@363830 91177308-0d34-0410-b5e6-96231b3b80d8 --- diff --git a/docs/CommandGuide/llvm-mca.rst b/docs/CommandGuide/llvm-mca.rst index d8c3b57207c..847a043fe7a 100644 --- a/docs/CommandGuide/llvm-mca.rst +++ b/docs/CommandGuide/llvm-mca.rst @@ -38,6 +38,17 @@ Or for Intel syntax: $ clang foo.c -O2 -target x86_64-unknown-unknown -mllvm -x86-asm-syntax=intel -S -o - | llvm-mca -mcpu=btver2 +Scheduling models are not just used to compute instruction latencies and +throughput, but also to understand what processor resources are available +and how to simulate them. + +By design, the quality of the analysis conducted by :program:`llvm-mca` is +inevitably affected by the quality of the scheduling models in LLVM. + +If you see that the performance report is not accurate for a processor, +please `file a bug `_ +against the appropriate backend. + OPTIONS -------