Signed-off-by: Kenneth Graunke
---
tests/shaders/fp-indirections.c | 13 +
1 file changed, 13 insertions(+)
It looks like people expose real values, 128 (Intel/Windows), 16384 (NVIDIA,
i965,
Gallium), 65536 (one NVIDIA card), and INT_MAX (AMD). So 10,000 is pretty
arbitrary, but it's conservative and speeds up my Piglit run times.
diff --git a/tests/shaders/fp-indirections.c b/tests/shaders/fp-indirections.c
index 3c84aea..540c252 100644
--- a/tests/shaders/fp-indirections.c
+++ b/tests/shaders/fp-indirections.c
@@ -296,4 +296,17 @@ piglit_init(int argc, char **argv)
get_program_i(GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB));
printf("Maximum native tex indirections: %d\n",
get_program_i(GL_MAX_PROGRAM_NATIVE_TEX_INDIRECTIONS_ARB));
+
+ /* If the GL reports more than 1 texture indirections, then we're
probably
+* running on hardware with no limitations - the driver just picked some
+* arbitrary large number to report back. The test isn't meaningful,
and
+* the run time explodes with huge limits, so just skip it.
+*
+* For reference, Mesa and NVIDIA report 16384; AMD reports 2147483647.
+* Pineview hardware (where this test is relevant) has a limit of 4.
+*/
+ if (get_program_i(GL_MAX_PROGRAM_TEX_INDIRECTIONS_ARB) > 1) {
+ printf("Skipping; the hardware doesn't appear to have real
limits.\n");
+ piglit_report_result(PIGLIT_SKIP);
+ }
}
--
2.3.4
___
Piglit mailing list
Piglit@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/piglit