Skip to content

Commit

Permalink
Rewrite a terrible comment about the machine model.
Browse files Browse the repository at this point in the history
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@202576 91177308-0d34-0410-b5e6-96231b3b80d8
  • Loading branch information
atrick committed Mar 1, 2014
1 parent 06f2e69 commit abad354
Showing 1 changed file with 10 additions and 6 deletions.
16 changes: 10 additions & 6 deletions include/llvm/MC/MCSchedule.h
Original file line number Diff line number Diff line change
Expand Up @@ -32,12 +32,16 @@ struct MCProcResourceDesc {

// Number of resources that may be buffered.
//
// Buffered resources (BufferSize > 0 || BufferSize == -1) may be consumed at
// some indeterminate cycle after dispatch (e.g. for instructions that may
// issue out-of-order). Unbuffered resources (BufferSize == 0) always consume
// their resource some fixed number of cycles after dispatch (e.g. for
// instruction interlocking that may stall the pipeline). If BufferSize==1,
// the latency between producer and consumer is modeled as a stall.
// Buffered resources (BufferSize != 0) may be consumed at some indeterminate
// cycle after dispatch. This should be used for out-of-order cpus when
// instructions that use this resource can be buffered in a reservaton
// station.
//
// Unbuffered resources (BufferSize == 0) always consume their resource some
// fixed number of cycles after dispatch. If a resource is unbuffered, then
// the scheduler will avoid scheduling instructions with conflicting resources
// in the same cycle. This is for in-order cpus, or the in-order portion of
// an out-of-order cpus.
int BufferSize;

bool operator==(const MCProcResourceDesc &Other) const {
Expand Down

0 comments on commit abad354

Please sign in to comment.