[PATCH v1 05/13] perf jevents: Add software prefetch (swpf) metric group for AMD

From: Ian Rogers
Date: Wed Feb 28 2024 - 19:17:04 EST


Add metrics that give the utility of software prefetches on zen2, zen3
and zen4.

Signed-off-by: Ian Rogers <irogers@xxxxxxxxxx>
---
tools/perf/pmu-events/amd_metrics.py | 95 ++++++++++++++++++++++++++++
1 file changed, 95 insertions(+)

diff --git a/tools/perf/pmu-events/amd_metrics.py b/tools/perf/pmu-events/amd_metrics.py
index f8d5ba2861ee..52451bcb4bbf 100755
--- a/tools/perf/pmu-events/amd_metrics.py
+++ b/tools/perf/pmu-events/amd_metrics.py
@@ -125,6 +125,100 @@ def AmdBr():
description="breakdown of retired branch instructions")


+def AmdSwpf() -> Optional[MetricGroup]:
+ """Returns a MetricGroup representing AMD software prefetch metrics."""
+ if zen_model <= 1:
+ return None
+
+ swp_ld = Event("ls_dispatch.ld_dispatch")
+ swp_t0 = Event("ls_pref_instr_disp.prefetch")
+ swp_w = Event("ls_pref_instr_disp.prefetch_w") # Missing on Zen1
+ swp_nt = Event("ls_pref_instr_disp.prefetch_nta")
+ swp_mab = Event("ls_inef_sw_pref.mab_mch_cnt")
+ swp_l2 = Event("ls_sw_pf_dc_fills.local_l2",
+ "ls_sw_pf_dc_fills.lcl_l2",
+ "ls_sw_pf_dc_fill.ls_mabresp_lcl_l2")
+ swp_lc = Event("ls_sw_pf_dc_fills.local_ccx",
+ "ls_sw_pf_dc_fills.int_cache",
+ "ls_sw_pf_dc_fill.ls_mabresp_lcl_cache")
+ swp_lm = Event("ls_sw_pf_dc_fills.dram_io_near",
+ "ls_sw_pf_dc_fills.mem_io_local",
+ "ls_sw_pf_dc_fill.ls_mabresp_lcl_dram")
+ swp_rc = Event("ls_sw_pf_dc_fills.far_cache",
+ "ls_sw_pf_dc_fills.ext_cache_remote",
+ "ls_sw_pf_dc_fill.ls_mabresp_rmt_cache")
+ swp_rm = Event("ls_sw_pf_dc_fills.dram_io_far",
+ "ls_sw_pf_dc_fills.mem_io_remote",
+ "ls_sw_pf_dc_fill.ls_mabresp_rmt_dram")
+
+ # All the swpf that were satisfied beyond L1D are good.
+ all_pf = swp_t0 + swp_w + swp_nt
+ good_pf = swp_l2 + swp_lc + swp_lm + swp_rc + swp_rm
+ bad_pf = max(all_pf - good_pf, 0)
+
+ loc_pf = swp_l2 + swp_lc + swp_lm
+ rem_pf = swp_rc + swp_rm
+
+ req_pend = max(0, bad_pf - swp_mab)
+
+ r1 = d_ratio(ins, all_pf)
+ r2 = d_ratio(swp_ld, all_pf)
+ r3 = d_ratio(swp_t0, interval_sec)
+ r4 = d_ratio(swp_w, interval_sec)
+ r5 = d_ratio(swp_nt, interval_sec)
+ overview = MetricGroup("swpf_overview", [
+ Metric("swpf_ov_insn_bt_swpf", "Insn between SWPF", r1, "insns"),
+ Metric("swpf_ov_loads_bt_swpf", "Loads between SWPF", r2, "loads"),
+ Metric("swpf_ov_rate_prefetch_t0_t1_t2", "Rate prefetch TO_T1_T2", r3,
+ "insns/sec"),
+ Metric("swpf_ov_rate_prefetch_w", "Rate prefetch W", r4, "insns/sec"),
+ Metric("swpf_ov_rate_preftech_nta", "Rate prefetch NTA", r5, "insns/sec"),
+ ])
+
+ r1 = d_ratio(swp_mab, all_pf)
+ r2 = d_ratio(req_pend, all_pf)
+ usefulness_bad = MetricGroup("swpf_usefulness_bad", [
+ Metric("swpf_use_bad_hit_l1", "Usefulness bad hit L1", r1, "100%"),
+ Metric("swpf_use_bad_req_pend", "Usefulness bad req pending", r2, "100%"),
+ ])
+
+ r1 = d_ratio(good_pf, all_pf)
+ usefulness_good = MetricGroup("swpf_usefulness_good", [
+ Metric("swpf_use_good_other_src", "Usefulness good other src", r1,
+ "100%"),
+ ])
+
+ usefulness = MetricGroup("swpf_usefulness", [
+ usefulness_bad,
+ usefulness_good,
+ ])
+
+ r1 = d_ratio(swp_l2, good_pf)
+ r2 = d_ratio(swp_lc, good_pf)
+ r3 = d_ratio(swp_lm, good_pf)
+ data_src_local = MetricGroup("swpf_data_src_local", [
+ Metric("swpf_data_src_local_l2", "Data source local l2", r1, "100%"),
+ Metric("swpf_data_src_local_ccx_l3_loc_ccx",
+ "Data source local ccx l3 loc ccx", r2, "100%"),
+ Metric("swpf_data_src_local_memory_or_io",
+ "Data source local memory or IO", r3, "100%"),
+ ])
+
+ r1 = d_ratio(swp_rc, good_pf)
+ r2 = d_ratio(swp_rm, good_pf)
+ data_src_remote = MetricGroup("swpf_data_src_remote", [
+ Metric("swpf_data_src_remote_cache", "Data source remote cache", r1,
+ "100%"),
+ Metric("swpf_data_src_remote_memory_or_io",
+ "Data source remote memory or IO", r2, "100%"),
+ ])
+
+ data_src = MetricGroup("swpf_data_src", [data_src_local, data_src_remote])
+
+ return MetricGroup("swpf", [overview, usefulness, data_src],
+ description="Sofware prefetch breakdown (CCX L3 = L3 of current thread, Loc CCX = CCX cache on some socket)")
+
+
def AmdUpc() -> Metric:
ops = Event("ex_ret_ops", "ex_ret_cops")
upc = d_ratio(ops, smt_cycles)
@@ -162,6 +256,7 @@ def Rapl() -> MetricGroup:

all_metrics = MetricGroup("", [
AmdBr(),
+ AmdSwpf(),
AmdUpc(),
Idle(),
Rapl(),
--
2.44.0.278.ge034bb2e1d-goog