intel/perf: Hide extended metrics by default

XE architecture enables many more metrics, perhaps too many for
the average user. Reduce reported metrics to smaller subset,
known as non-extended metrics, by default. Can re-enable extended
metrics with env var INTEL_EXTENDED_METRICS=1

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21841>

diff --git a/docs/envvars.rst b/docs/envvars.rst
index 346c68b..39a3c55 100644 (file)
--- a/docs/envvars.rst
+++ b/docs/envvars.rst
@@ -566,6 +566,12 @@ Intel driver environment variables
    ``offsets``
       print offsets of instructions
 
+.. envvar:: INTEL_EXTENDED_METRICS
+
+   By default, only a standard set of gpu metrics are advertised. This
+   reduces time to collect metrics and hides infrequently used metrics.
+   To enable all metrics, set value to 1.
+
 .. envvar:: INTEL_MEASURE
 
    Collects GPU timestamps over common intervals, and generates a CSV report

diff --git a/src/intel/perf/intel_perf.c b/src/intel/perf/intel_perf.c
index c0b24f4..b5124c8 100644 (file)
--- a/src/intel/perf/intel_perf.c
+++ b/src/intel/perf/intel_perf.c
@@ -49,6 +49,7 @@
 #include "util/bitscan.h"
 #include "util/macros.h"
 #include "util/mesa-sha1.h"
+#include "util/u_debug.h"
 #include "util/u_math.h"
 
 #define FILE_DEBUG_FLAG DEBUG_PERFMON
@@ -170,12 +171,24 @@ read_sysfs_drm_device_file_uint64(struct intel_perf_config *perf,
    return read_file_uint64(buf, value);
 }
 
+static bool
+oa_config_enabled(struct intel_perf_config *perf,
+                  const struct intel_perf_query_info *query) {
+   // Hide extended metrics unless enabled with env param
+   bool is_extended_metric = strncmp(query->name, "Ext", 3) == 0;
+
+   return perf->enable_all_metrics || !is_extended_metric;
+}
+
 static void
 register_oa_config(struct intel_perf_config *perf,
                    const struct intel_device_info *devinfo,
                    const struct intel_perf_query_info *query,
                    uint64_t config_id)
 {
+   if (!oa_config_enabled(perf, query))
+      return;
+
    struct intel_perf_query_info *registered_query =
       intel_perf_append_query_info(perf, 0);
 
@@ -718,6 +731,7 @@ oa_metrics_available(struct intel_perf_config *perf, int fd,
    }
 
    perf->i915_query_supported = i915_query_perf_config_supported(perf, fd);
+   perf->enable_all_metrics = debug_get_bool_option("INTEL_EXTENDED_METRICS", false);
    perf->i915_perf_version = i915_perf_version(fd);
 
    /* TODO: We should query this from i915 */

diff --git a/src/intel/perf/intel_perf.h b/src/intel/perf/intel_perf.h
index 6a9f835..9c6e6f9 100644 (file)
--- a/src/intel/perf/intel_perf.h
+++ b/src/intel/perf/intel_perf.h
@@ -327,6 +327,9 @@ struct intel_perf_config {
    /* Whether i915 has DRM_I915_QUERY_PERF_CONFIG support. */
    bool i915_query_supported;
 
+   /* Have extended metrics been enabled */
+   bool enable_all_metrics;
+
    /* Version of the i915-perf subsystem, refer to i915_drm.h. */
    int i915_perf_version;