A set of C-language polyhedral benchmarks. There are three quick running tests.

Topdown profile is similarly sparse.

AMD metrics show less than 1/4 of core is used
elapsed 67.382
on_cpu 0.015 # 0.24 / 16 cores
utime 14.928
stime 1.044
nvcsw 2036 # 88.91%
nivcsw 254 # 11.09%
inblock 0 # 0.00/sec
onblock 12816 # 190.20/sec
cpu-clock 15999961726 # 16.000 seconds
task-clock 16003329029 # 16.003 seconds
page faults 299228 # 18697.860/sec
context switches 2446 # 152.843/sec
cpu migrations 262 # 16.372/sec
major page faults 2 # 0.125/sec
minor page faults 299226 # 18697.735/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 16223912644 # 137.618 branches per 1000 inst
branch misses 106135332 # 0.65% branch miss
conditional 15504802149 # 131.519 conditional branches per 1000 inst
indirect 46318603 # 0.393 indirect branches per 1000 inst
cpu-cycles 56802673451 # 0.05 GHz
instructions 116563864786 # 2.05 IPC
slots 115810833066 #
retiring 34742752204 # 30.0% (30.0%)
-- ucode 10738983 # 0.0%
-- fastpath 34732013221 # 30.0%
frontend 8382785080 # 7.2% ( 7.2%)
-- latency 5697719298 # 4.9%
-- bandwidth 2685065782 # 2.3%
backend 72008995376 # 62.2% (62.2%)
-- cpu 45273184687 # 39.1%
-- memory 26735810689 # 23.1%
speculation 645574399 # 0.6% ( 0.6%) low
-- branch mispredict 638080901 # 0.6%
-- pipeline restart 7493498 # 0.0%
smt-contention 30442664 # 0.0% ( 0.0%)
cpu-cycles 56803940933 # 0.05 GHz
instructions 116378467062 # 2.05 IPC
instructions 39369231008 # 276.827 l2 access per 1000 inst
l2 hit from l1 7276325564 # 5.84% l2 miss
l2 miss from l1 361027290 #
l2 hit from l2 pf 3346812187 #
l3 hit from l2 pf 266835347 #
l3 miss from l2 pf 8491441 #
instructions 39274623265 # 241.655 float per 1000 inst
float 512 55 # 0.000 AVX-512 per 1000 inst
float 256 538 # 0.000 AVX-256 per 1000 inst
float 128 9490912630 # 241.655 AVX-128 per 1000 inst
float MMX 0 # 0.000 MMX per 1000 inst
float scalar 0 # 0.000 scalar per 1000 inst
instructions 117774240339 #
opcache 4186166650 # 35.544 opcache per 1000 inst
opcache miss 717023170 # 17.1% opcache miss rate
l1 dTLB miss 18215253201 # 154.662 L1 dTLB per 1000 inst
l2 dTLB miss 105778786 # 0.898 L2 dTLB per 1000 inst
instructions 117784192075 #
icache 1513163017 # 12.847 icache per 1000 inst
icache miss 174490262 # 11.5% icache miss rate
l1 iTLB miss 8467310 # 0.072 L1 iTLB per 1000 inst
l2 iTLB miss 0 # 0.000 L2 iTLB per 1000 inst
tlb flush 17414 # 0.000 TLB flush per 1000 inst
Intel metrics are similarly quick, in this case showing backend bound nature in L2
elapsed 79.629
on_cpu 0.023 # 0.37 / 16 cores
utime 28.562
stime 0.560
nvcsw 1831 # 92.10%
nivcsw 157 # 7.90%
inblock 10216 # 128.30/sec
onblock 1576 # 19.79/sec
cpu-clock 29150212553 # 29.150 seconds
task-clock 29153359668 # 29.153 seconds
page faults 236174 # 8101.090/sec
context switches 2204 # 75.600/sec
cpu migrations 178 # 6.106/sec
major page faults 40 # 1.372/sec
minor page faults 236134 # 8099.718/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 14991982033 # 153.812 branches per 1000 inst
branch misses 22893515 # 0.15% branch miss
conditional 14991992401 # 153.813 conditional branches per 1000 inst
indirect 21024144 # 0.216 indirect branches per 1000 inst
slots 1796147433314 #
retiring 182199830495 # 10.1% (10.1%) low
-- ucode 21793493705 # 1.2%
-- fastpath 160406336790 # 8.9%
frontend 11742967638 # 0.7% ( 0.7%) low
-- latency 9138659764 # 0.5%
-- bandwidth 2604307874 # 0.1%
backend 1601029301507 # 89.1% (89.1%) high
-- cpu 452951150880 # 25.2%
-- memory 1148078150627 # 63.9%
speculation 3731819750 # 0.2% ( 0.2%) low
-- branch mispredict 2344944033 # 0.1%
-- pipeline restart 1386875717 # 0.1%
smt-contention 0 # 0.0% ( 0.0%)
cpu-cycles 259085483965 # 0.10 GHz
instructions 171145304834 # 0.66 IPC low
l2 access 53673106299 # 313.859 l2 access per 1000 inst
l2 miss 4743715242 # 8.84% l2 miss
cpu-cycles 261475873410 # 66.0% memory latency
load stalls 172553812136 # 0.0% l1 bound
l1 miss 173873149616 # 58.8% l2 bound
l2 miss 20114165538 # 4.8% l3 bound
l3 miss 7575873524 # 2.9% dram bound
store_stalls 114286947 # 0.0% store bound
