linux-kernel - [PATCH 3/6] cpufreq/intel_pstate: fix ffmpeg regression

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <1367940030-19311-4-git-send-email-dirk.j.brandewie@gmail.com>
Date:	Tue,  7 May 2013 08:20:27 -0700
From:	dirk.brandewie@...il.com
To:	rjw@...k.pl, linux-kernel@...r.kernel.org, cpufreq@...r.kernel.org
Cc:	Dirk Brandewie <dirk.j.brandewie@...el.com>, stable@...r.kernel.org
Subject: [PATCH 3/6] cpufreq/intel_pstate: fix ffmpeg regression

From: Dirk Brandewie <dirk.j.brandewie@...el.com>

The ffmpeg benchmark in the phoronix test suite has threads on
multiple cores that rely on the progress on of threads on other cores
and ping pong back and forth fast enough to make the core appear less
busy than it "should" be.  If the core has been at minimum p-state for
a while bump the pstate up to kick the core to see if it is in this
ping pong state.  If the core is truly idle the p-state will be
reduced at the next sample time.  If the core makes more progress it
will send more work to the thread bringing both threads out of the
ping pong scenario and the p-state will be selected normally.

This fixes a performance regression of approximately 30%

Cc: stable@...r.kernel.org
Signed-off-by: Dirk Brandewie <dirk.j.brandewie@...el.com>
---
 drivers/cpufreq/intel_pstate.c |   10 ++--------
 1 files changed, 2 insertions(+), 8 deletions(-)

diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c
index c0c98c2..97fbac2 100644
--- a/drivers/cpufreq/intel_pstate.c
+++ b/drivers/cpufreq/intel_pstate.c
@@ -551,22 +551,16 @@ static void intel_pstate_timer_func(unsigned long __data)
 	struct cpudata *cpu = (struct cpudata *) __data;
 
 	intel_pstate_sample(cpu);
+	intel_pstate_adjust_busy_pstate(cpu);
 
-	if (!cpu->idle_mode)
-		intel_pstate_adjust_busy_pstate(cpu);
-	else
-		intel_pstate_adjust_idle_pstate(cpu);
-
-#if defined(XPERF_FIX)
 	if (cpu->pstate.current_pstate == cpu->pstate.min_pstate) {
 		cpu->min_pstate_count++;
 		if (!(cpu->min_pstate_count % 5)) {
 			intel_pstate_set_pstate(cpu, cpu->pstate.max_pstate);
-			intel_pstate_idle_mode(cpu);
 		}
 	} else
 		cpu->min_pstate_count = 0;
-#endif
+
 	intel_pstate_set_sample_time(cpu);
 }
 
-- 
1.7.7.6

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/