This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Doubt regarding counting number of issue groups within loop scheduled by SMS
- From: Meena <meena at acmet dot com>
- To: gcc-help <gcc-help at gcc dot gnu dot org>
- Date: Wed, 06 Jan 2010 15:08:23 +0530
- Subject: Doubt regarding counting number of issue groups within loop scheduled by SMS
- Reply-to: meena at acmet dot com
Hi,
Consider the following snip from gcc internals
---snip---
The second Haifa scheduling pass, for targets that can multiple issue,
sets the mode of an insn to TImode when it is believed that the
instruction begins an issue group. That is, when the instruction cannot
issue simultaneously with the previous. This may be relied on by later
passes, in particular machine-dependent reorg.
---snip---
As per the above snip, haifa scheduling pass sets the mode of an insn to
TImode when insn begins an issue group.
I am trying to port GCC for a DSP architecture that can multiple issue.
But if the loop is optimized using SMS (using -fmodulo-sched option)
then it does not sets the mode of an insn (within loop) to TImode.
I need this information to compute the number of issue groups within a
loop to do some target specific optimization.
How can we set the mode of an insn (within loop scheduled by SMS) to
TImode that begins an issue group?
'Or'
Is there any other way to count the number of issue groups within a
loop?
Thanks in advance,
Meena