cpu: o3: single cycle default div microop latency on x86

This patch sets the default latency of the division microop to a single cycle
on x86.  This is because the division instructions DIV and IDIV have been
implemented as loops of div microops, where each microop computes a single bit
of the quotient.

diff --git a/src/cpu/o3/FuncUnitConfig.py b/src/cpu/o3/FuncUnitConfig.py
index 34c56163d02a076d1ac05b5a3b2a7dd8ba5d766c..0f5efb7767baa99a2f99ae51a662ad2ca80257e6 100644 (file)
--- a/src/cpu/o3/FuncUnitConfig.py
+++ b/src/cpu/o3/FuncUnitConfig.py
@@ -39,6 +39,7 @@
 # Authors: Kevin Lim
 
 from m5.SimObject import SimObject
+from m5.defines import buildEnv
 from m5.params import *
 from FuncUnit import *
 
@@ -49,6 +50,15 @@ class IntALU(FUDesc):
 class IntMultDiv(FUDesc):
     opList = [ OpDesc(opClass='IntMult', opLat=3),
                OpDesc(opClass='IntDiv', opLat=20, issueLat=19) ]
+
+    # DIV and IDIV instructions in x86 are implemented using a loop which
+    # issues division microops.  The latency of these microops should really be
+    # one (or a small number) cycle each since each of these computes one bit
+    # of the quotient.
+    if buildEnv['TARGET_ISA'] in ('x86'):
+        opList[1].opLat=1
+        opList[1].issueLat=1
+
     count=2
 
 class FP_ALU(FUDesc):