On client DG2 platforms, optimal performance is achieved with the
hardware's default "age based" thread execution setting. However on
ATS-M, switching this to "round robin after dependencies" provides
better performance. We'll add a new "tuning" feature flag to the ATS-M
device info to enable/disable this setting.
Bspec: 68331
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Matt Atwood <matthew.s.atwood@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220826212718.409948-1-matthew.d.roper@intel.com
#define GEN12_DISABLE_TDL_PUSH REG_BIT(9)
#define GEN11_DIS_PICK_2ND_EU REG_BIT(7)
#define GEN12_DISABLE_HDR_PAST_PAYLOAD_HOLD_FIX REG_BIT(4)
+#define THREAD_EX_ARB_MODE REG_GENMASK(3, 2)
+#define THREAD_EX_ARB_MODE_RR_AFTER_DEP REG_FIELD_PREP(THREAD_EX_ARB_MODE, 0x2)
#define HSW_ROW_CHICKEN3 _MMIO(0xe49c)
#define HSW_ROW_CHICKEN3_L3_GLOBAL_ATOMICS_DISABLE (1 << 6)
0 /* write-only, so skip validation */,
true);
}
+
+ /*
+ * This tuning setting proves beneficial only on ATS-M designs; the
+ * default "age based" setting is optimal on regular DG2 and other
+ * platforms.
+ */
+ if (INTEL_INFO(i915)->tuning_thread_rr_after_dep)
+ wa_masked_field_set(wal, GEN9_ROW_CHICKEN4, THREAD_EX_ARB_MODE,
+ THREAD_EX_ARB_MODE_RR_AFTER_DEP);
}
/*
DG2_FEATURES,
.display = { 0 },
.require_force_probe = 1,
+ .tuning_thread_rr_after_dep = 1,
};
#define XE_HPC_FEATURES \
func(has_runtime_pm); \
func(has_snoop); \
func(has_coherent_ggtt); \
+ func(tuning_thread_rr_after_dep); \
func(unfenced_needs_alignment); \
func(hws_needs_physical);