On 08/06/2018 22:35, Chris Wilson wrote:
guc also uses timer-based sampling and cannot reliably hit our accuracy
requirements for the test, so skip.
We also have an open ticket asking from GuC FW to implement something to
allow us to keep non-sampling/accurate mode. From that angle it is good
to have these few tests failing as a reminder that needs to be done. So
I'd rather we don't sweep it under the rug.
Regards,
Tvrtko
Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
Cc: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
---
tests/perf_pmu.c | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/tests/perf_pmu.c b/tests/perf_pmu.c
index 4570f926d..61e83bf7b 100644
--- a/tests/perf_pmu.c
+++ b/tests/perf_pmu.c
@@ -1507,6 +1507,17 @@ static void __rearm_spin_batch(igt_spin_t *spin)
#define assert_within(x, ref, tolerance) \
__assert_within(x, ref, tolerance, tolerance)
+static bool uses_timer_sampling(int gem_fd)
+{
+ if (!gem_has_execlists(gem_fd))
+ return true;
+
+ if (gem_has_guc_submission(gem_fd))
+ return true;
+
+ return false;
+}
+
static void
accuracy(int gem_fd, const struct intel_execution_engine2 *e,
unsigned long target_busy_pct)
@@ -1524,7 +1535,7 @@ accuracy(int gem_fd, const struct intel_execution_engine2 *e,
int fd;
/* Sampling platforms cannot reach the high accuracy criteria. */
- igt_require(gem_has_execlists(gem_fd));
+ igt_require(!uses_timer_sampling(gem_fd));
while (idle_us < 2500) {
busy_us *= 2;
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
