"Add performance tuning for RT_CTRL" would be more accurate as this is
handled as a workaround in the spec.
On Thu, Jun 30, 2022 at 11:35:08AM +0300, Lionel Landwerlin wrote:
The recommended number of stackIDs for Ray Tracing subsystem is 512
rather than 2048 (default HW programming).
v2: Move the programming to dg2_ctx_gt_tuning_init() (Lucas)
sorry to nitpick, but as said in my reply to v2, this is not accurate. I
make a lot of mistakes on this kind of things, but this time I even
checked if it was in the context state.
v3: Move programming to general_render_compute_wa_init() (Matt)
and also disagreed with this. We need an equivalent tuning function for
registers not context saved/restored. The tuning values are all in a
single place in the spec and having them in their own functions in the
driver make it much easier to check than having to search for it in the
middle of workarounds.
This just happens to be the first and we don't need another commit later
to move things around if we can make it right from the beginning.
Lucas De Marchi
Signed-off-by: Lionel Landwerlin <lionel.g.landwer...@intel.com>
---
drivers/gpu/drm/i915/gt/intel_gt_regs.h | 4 ++++
drivers/gpu/drm/i915/gt/intel_workarounds.c | 9 +++++++++
2 files changed, 13 insertions(+)
diff --git a/drivers/gpu/drm/i915/gt/intel_gt_regs.h
b/drivers/gpu/drm/i915/gt/intel_gt_regs.h
index 07ef111947b8c..12fc87b957425 100644
--- a/drivers/gpu/drm/i915/gt/intel_gt_regs.h
+++ b/drivers/gpu/drm/i915/gt/intel_gt_regs.h
@@ -1112,6 +1112,10 @@
#define GEN12_PUSH_CONST_DEREF_HOLD_DIS REG_BIT(8)
#define RT_CTRL _MMIO(0xe530)
+#define RT_CTRL_NUMBER_OF_STACKIDS_MASK REG_GENMASK(6, 5)
+#define NUMBER_OF_STACKIDS_512 2
+#define NUMBER_OF_STACKIDS_1024 1
+#define NUMBER_OF_STACKIDS_2048 0
#define DIS_NULL_QUERY REG_BIT(10)
#define EU_PERF_CNTL1 _MMIO(0xe558)
diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c
b/drivers/gpu/drm/i915/gt/intel_workarounds.c
index 3213c593a55f4..ea674e456cd76 100644
--- a/drivers/gpu/drm/i915/gt/intel_workarounds.c
+++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c
@@ -2737,6 +2737,15 @@ general_render_compute_wa_init(struct intel_engine_cs
*engine, struct i915_wa_li
wa_write_or(wal, VDBX_MOD_CTRL, FORCE_MISS_FTLB);
wa_write_or(wal, VEBX_MOD_CTRL, FORCE_MISS_FTLB);
}
+
+ if (IS_DG2(i915)) {
+ /* Performance tuning for Ray-tracing */
+ wa_write_clr_set(wal,
+ RT_CTRL,
+ RT_CTRL_NUMBER_OF_STACKIDS_MASK,
+ REG_FIELD_PREP(RT_CTRL_NUMBER_OF_STACKIDS_MASK,
+ NUMBER_OF_STACKIDS_512));
+ }
}
static void
--
2.34.1
