From: Dorota Czaplejewicz <dorota.czaplejewicz@collabora.co.uk>
To: Intel GFX discussion <intel-gfx@lists.freedesktop.org>
Subject: [PATCH i-g-t 1/2] igt/kms_setmode.c: Tweak timings check constants
Date: Wed, 30 Nov 2016 17:05:34 +0100 [thread overview]
Message-ID: <20161130170534.787fbd22@collabora.co.uk> (raw)
Add explanations of the checks and note the way the constants were derived.
Tweak the constants to reflect their purpose better.
---
tests/kms_setmode.c | 36 +++++++++++++++++++++++++++++++++---
1 file changed, 33 insertions(+), 3 deletions(-)
diff --git a/tests/kms_setmode.c b/tests/kms_setmode.c
index 4031d81..74eed88 100644
--- a/tests/kms_setmode.c
+++ b/tests/kms_setmode.c
@@ -474,18 +474,48 @@ static void check_timings(int crtc_idx, const drmModeModeInfo *kmode)
mean = igt_stats_get_mean(&stats);
stddev = igt_stats_get_std_deviation(&stats);
- accuracy = 3.090 * stddev; /* 2-tailed 99% confidence */
+
+ /* 99.7% samples fall within `accuracy` on both sides of mean in normal
+ * distribution if `accuracy = 3 * sigma`.
+ * https://en.wikipedia.org/wiki/68%E2%80%9395%E2%80%9399.7_rule
+ *
+ * The value of 99.7% was chosen to suit requirements of test cases
+ * which depend on timing, giving the lowest acceptable MTBF of 5.6s
+ * for 60Hz sampling rate.
+ */
+ accuracy = 3. * stddev;
igt_info("Expected frametime: %.0fus; measured %.1fus +- %.3fus accuracy %.2f%% [%.2f scanlines]\n",
expected, mean, stddev,
100 * accuracy / mean, accuracy / line_time(kmode));
- /* 99% accuracy within one scanline */
+ /* 99.7% samples within one scanline on each side of mean */
igt_assert_f(accuracy < line_time(kmode),
"vblank accuracy (%.3fus, %.1f%%) worse than a scanline (%.3fus)\n",
accuracy, 100 * accuracy / mean, line_time(kmode));
- igt_assert_f(fabs(mean - expected) < 2*stddev,
+ /* At least 90% of frame times fall within the one scanline on each
+ * side of expected mean.
+ *
+ * Expected scanline duration:
+ * (expected - accuracy, expected + accuracy).
+ * Assuming maximum difference allowed:
+ * expected = mean + n * sigma
+ * the scanline duration becomes:
+ * (mean - accuracy + n * sigma, mean + accuracy + n * sigma)
+ * The expected scanline captures the following number of samples
+ * from each side of expected:
+ * (erf(abs(-(accuracy/sigma) + n) / sqrt(2))
+ * + erf((accuracy/sigma) + n) / sqrt(2))) / 2
+ * = samples
+ *
+ * Solving for samples = 0.9:
+ * n = 1.718
+ *
+ * See:
+ * https://en.wikipedia.org/wiki/Standard_deviation#Rules_for_normally_distributed_data
+ */
+ igt_assert_f(fabs(mean - expected) < 1.718 * stddev,
"vblank interval differs from modeline! expected %.1fus, measured %1.fus +- %.3fus, difference %.1fus (%.1f sigma)\n",
expected, mean, stddev,
fabs(mean - expected), fabs(mean - expected) / stddev);
--
2.7.4
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
next reply other threads:[~2016-11-30 16:05 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-11-30 16:05 Dorota Czaplejewicz [this message]
2016-11-30 16:06 ` [PATCH i-g-t 2/2] igt/kms_flip.c: Fix timings check Dorota Czaplejewicz
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20161130170534.787fbd22@collabora.co.uk \
--to=dorota.czaplejewicz@collabora.co.uk \
--cc=intel-gfx@lists.freedesktop.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).