diff --git a/stackcollapse-perf.pl b/stackcollapse-perf.pl
index 3ff39bf..43566ec 100755
--- a/stackcollapse-perf.pl
+++ b/stackcollapse-perf.pl
@@ -88,6 +88,7 @@ sub remember_stack {
 
 my $show_inline = 0;
 my $show_context = 0;
+my $aggregate_comm = 0;   # replace '/<suffix>' in input process names with '/all'
 
 my $srcline_in_input = 0; # if there are extra lines with source location (perf script -F+srcline)
 GetOptions('inline' => \$show_inline,
@@ -99,7 +100,8 @@ sub remember_stack {
            'all' => \$annotate_all,
            'tid' => \$include_tid,
            'addrs' => \$include_addrs,
-           'event-filter=s' => \$event_filter)
+           'event-filter=s' => \$event_filter,
+           'aggregate-comm' => \$aggregate_comm)
 or die <<USAGE_END;
 USAGE: $0 [options] infile > outfile\n
 	--pid		# include PID with process names [1]
@@ -111,7 +113,8 @@ sub remember_stack {
 	--context	# adds source context to --inline
 	--srcline	# parses output of 'perf script -F+srcline' and adds source context
 	--addrs		# include raw addresses where symbols can't be found
-	--event-filter=EVENT	# event name filter\n
+	--event-filter=EVENT	# event name filter
+	--aggregate-comm        # replace '/<suffix>' in input process names with '/all'\n
 [1] perf script must emit both PID and TIDs for these to work; eg, Linux < 4.1:
 	perf script -f comm,pid,tid,cpu,time,event,ip,sym,dso,trace
     for Linux >= 4.1:
@@ -256,6 +259,12 @@ sub inline {
 			$tid = $pid;
 			$pid = "?";
 		}
+		if ($aggregate_comm) {
+			# Replace per-cpu suffix in process names with '/all'.
+			# This allows aggregation across multiple CPUs.
+			# ksoftirqd/156 -> ksoftirqd/all
+			$comm =~ s/\/.+/\/all/g;
+		}
 
 		if (/:\s*(\d+)*\s+(\S+):\s*$/) {
 			$period = $1;