i965: Throttle to the previous frame

In order to facilitate the concurrency offered by triple buffering and to
offset the latency induced by swapping via an external process, which
may incur extra rendering itself, only throttle to the previous frame
and not the last. The second issue that mostly affects swap benchmarks,
but also can incur jitter in the throttling, is that the throttle bo is
closer to the next SwapBuffers rather than immediately after the previous
SwapBuffers. Throttling to the previous frame doubles the maximum possible
latency at the benefit of improving throughput and reducing jitter.

v2: Rename "first_post_swapbuffer" batches array to a plain
throttle_batch[] as the pluralisation was contorting the name and not
making it clear as to whether it was the first batch or first_post_swap
batch. Not least of which was that not all throttle points are SwapBuffers.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Ben Widawsky <ben@bwidawsk.net>
Cc: Kristian Høgsberg <krh@bitplanet.net>
Cc: Chad Versace <chad.versace@linux.intel.com>
Cc: Ian Romanick <idr@freedesktop.org>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>

3 files changed, 17 insertions, 11 deletions

diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index 5120372..8257fb6 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -935,8 +935,10 @@ intelDestroyContext(__DRIcontext * driContextPriv)
 
    intel_batchbuffer_free(brw);
 
-   drm_intel_bo_unreference(brw->first_post_swapbuffers_batch);
-   brw->first_post_swapbuffers_batch = NULL;
+   drm_intel_bo_unreference(brw->throttle_batch[1]);
+   drm_intel_bo_unreference(brw->throttle_batch[0]);
+   brw->throttle_batch[1] = NULL;
+   brw->throttle_batch[0] = NULL;
 
    driDestroyOptionCache(&brw->optionCache);
 
@@ -1245,11 +1247,14 @@ intel_prepare_render(struct brw_context *brw)
     * the swap, and getting our hands on that doesn't seem worth it,
     * so we just us the first batch we emitted after the last swap.
     */
-   if (brw->need_swap_throttle && brw->first_post_swapbuffers_batch) {
-      if (!brw->disable_throttling)
-         drm_intel_bo_wait_rendering(brw->first_post_swapbuffers_batch);
-      drm_intel_bo_unreference(brw->first_post_swapbuffers_batch);
-      brw->first_post_swapbuffers_batch = NULL;
+   if (brw->need_swap_throttle && brw->throttle_batch[0]) {
+      if (brw->throttle_batch[1]) {
+         if (!brw->disable_throttling)
+            drm_intel_bo_wait_rendering(brw->throttle_batch[1]);
+         drm_intel_bo_unreference(brw->throttle_batch[1]);
+      }
+      brw->throttle_batch[1] = brw->throttle_batch[0];
+      brw->throttle_batch[0] = NULL;
       brw->need_swap_throttle = false;
       /* Throttling here is more precise than the throttle ioctl, so skip it */
       brw->need_flush_throttle = false;
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index eebd7ce..8b29e2a 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -1030,7 +1030,7 @@ struct brw_context
    bool front_buffer_dirty;
 
    /** Framerate throttling: @{ */
-   drm_intel_bo *first_post_swapbuffers_batch;
+   drm_intel_bo *throttle_batch[2];
 
    /* Limit the number of outstanding SwapBuffers by waiting for an earlier
     * frame of rendering to complete. This gives a very precise cap to the
diff --git a/src/mesa/drivers/dri/i965/intel_batchbuffer.c b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
index 5ac4d18..87862cd 100644
--- a/src/mesa/drivers/dri/i965/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
@@ -168,6 +168,7 @@ static void
 brw_new_batch(struct brw_context *brw)
 {
    /* Create a new batchbuffer and reset the associated state: */
+   drm_intel_gem_bo_clear_relocs(brw->batch.bo, 0);
    intel_batchbuffer_reset(brw);
 
    /* If the kernel supports hardware contexts, then most hardware state is
@@ -289,9 +290,9 @@ _intel_batchbuffer_flush(struct brw_context *brw,
    if (brw->batch.used == 0)
       return 0;
 
-   if (brw->first_post_swapbuffers_batch == NULL) {
-      brw->first_post_swapbuffers_batch = brw->batch.bo;
-      drm_intel_bo_reference(brw->first_post_swapbuffers_batch);
+   if (brw->throttle_batch[0] == NULL) {
+      brw->throttle_batch[0] = brw->batch.bo;
+      drm_intel_bo_reference(brw->throttle_batch[0]);
    }
 
    if (unlikely(INTEL_DEBUG & DEBUG_BATCH)) {