A quick running ray tracer test.
Topdown metrics shows a fairly high retirement rate limited by backend stalls of both CPU and memory.
AMD metrics show running on all cores, floating point code and not many L2 accesses or misses.
elapsed 148.090
on_cpu 0.838 # 13.40 / 16 cores
utime 1983.478
stime 1.405
nvcsw 20374 # 34.56%
nivcsw 38574 # 65.44%
inblock 0 # 0.00/sec
onblock 14056 # 94.92/sec
cpu-clock 1984937360494 # 1984.937 seconds
task-clock 1984951444204 # 1984.951 seconds
page faults 198620 # 100.063/sec
context switches 59509 # 29.980/sec
cpu migrations 2780 # 1.401/sec
major page faults 2 # 0.001/sec
minor page faults 198618 # 100.062/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 1014202012293 # 73.753 branches per 1000 inst
branch misses 6191746506 # 0.61% branch miss
conditional 609554742644 # 44.327 conditional branches per 1000 inst
indirect 84296971221 # 6.130 indirect branches per 1000 inst
cpu-cycles 7888007229708 # 3.32 GHz
instructions 13748191578115 # 1.74 IPC
slots 15779680550700 #
retiring 4822507950863 # 30.6% (48.6%)
-- ucode 2839998364 # 0.0%
-- fastpath 4819667952499 # 30.5%
frontend 584429036683 # 3.7% ( 5.9%)
-- latency 422768735268 # 2.7%
-- bandwidth 161660301415 # 1.0%
backend 4314856481849 # 27.3% (43.4%)
-- cpu 2741261117778 # 17.4%
-- memory 1573595364071 # 10.0%
speculation 209239720319 # 1.3% ( 2.1%)
-- branch mispredict 165667444230 # 1.0%
-- pipeline restart 43572276089 # 0.3%
smt-contention 5848634155814 # 37.1% ( 0.0%)
cpu-cycles 7888366991559 # 3.33 GHz
instructions 13756479664683 # 1.74 IPC
instructions 4587432604742 # 18.648 l2 access per 1000 inst
l2 hit from l1 76718057387 # 0.50% l2 miss
l2 miss from l1 223100129 #
l2 hit from l2 pf 8622365520 #
l3 hit from l2 pf 188302969 #
l3 miss from l2 pf 15496089 #
instructions 4578337251665 # 214.073 float per 1000 inst
float 512 51 # 0.000 AVX-512 per 1000 inst
float 256 664 # 0.000 AVX-256 per 1000 inst
float 128 980097688996 # 214.073 AVX-128 per 1000 inst
float MMX 0 # 0.000 MMX per 1000 inst
float scalar 4 # 0.000 scalar per 1000 inst
Intel metrics. The Intel version wasn’t as stable when running, perhaps differences between cores?
elapsed 807.771
on_cpu 0.880 # 14.07 / 16 cores
utime 11364.904
stime 2.490
nvcsw 80897 # 31.26%
nivcsw 177859 # 68.74%
inblock 1328 # 1.64/sec
onblock 7864 # 9.74/sec
cpu-clock 11367494151495 # 11367.494 seconds
task-clock 11367537895233 # 11367.538 seconds
page faults 360565 # 31.719/sec
context switches 262583 # 23.099/sec
cpu migrations 7004 # 0.616/sec
major page faults 14 # 0.001/sec
minor page faults 360551 # 31.718/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 4663954542631 # 70.318 branches per 1000 inst
branch misses 23249701772 # 0.50% branch miss
conditional 4663954574567 # 70.318 conditional branches per 1000 inst
indirect 1515855054426 # 22.855 indirect branches per 1000 inst
slots 11383013944694 #
retiring 7724310148011 # 67.9% (67.9%)
-- ucode 434825331904 # 3.8%
-- fastpath 7289484816107 # 64.0%
frontend 2693413158175 # 23.7% (23.7%)
-- latency 1649219459909 # 14.5%
-- bandwidth 1044193698266 # 9.2%
backend 741447952721 # 6.5% ( 6.5%)
-- cpu 415808685357 # 3.7%
-- memory 325639267364 # 2.9%
speculation 241057697650 # 2.1% ( 2.1%)
-- branch mispredict 230083087910 # 2.0%
-- pipeline restart 10974609740 # 0.1%
smt-contention 0 # 0.0% ( 0.0%)
cpu-cycles 6366658262732 # 2.26 GHz
instructions 12862934995320 # 2.02 IPC
l2 access 48169438096 # 6.398 l2 access per 1000 inst
l2 miss 502292780 # 1.04% l2 miss
Process overview
elapsed 807.771
on_cpu 0.880 # 14.07 / 16 cores
utime 11364.904
stime 2.490
nvcsw 80897 # 31.26%
nivcsw 177859 # 68.74%
inblock 1328 # 1.64/sec
onblock 7864 # 9.74/sec
cpu-clock 11367494151495 # 11367.494 seconds
task-clock 11367537895233 # 11367.538 seconds
page faults 360565 # 31.719/sec
context switches 262583 # 23.099/sec
cpu migrations 7004 # 0.616/sec
major page faults 14 # 0.001/sec
minor page faults 360551 # 31.718/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 4663954542631 # 70.318 branches per 1000 inst
branch misses 23249701772 # 0.50% branch miss
conditional 4663954574567 # 70.318 conditional branches per 1000 inst
indirect 1515855054426 # 22.855 indirect branches per 1000 inst
slots 11383013944694 #
retiring 7724310148011 # 67.9% (67.9%)
-- ucode 434825331904 # 3.8%
-- fastpath 7289484816107 # 64.0%
frontend 2693413158175 # 23.7% (23.7%)
-- latency 1649219459909 # 14.5%
-- bandwidth 1044193698266 # 9.2%
backend 741447952721 # 6.5% ( 6.5%)
-- cpu 415808685357 # 3.7%
-- memory 325639267364 # 2.9%
speculation 241057697650 # 2.1% ( 2.1%)
-- branch mispredict 230083087910 # 2.0%
-- pipeline restart 10974609740 # 0.1%
smt-contention 0 # 0.0% ( 0.0%)
cpu-cycles 6366658262732 # 2.26 GHz
instructions 12862934995320 # 2.02 IPC
l2 access 48169438096 # 6.398 l2 access per 1000 inst
l2 miss 502292780 # 1.04% l2 miss
Computation block
2580699) povray cpu=8 start=53.46 finish=97.40
2580700) povray cpu=9 start=53.46 finish=53.46
2580701) povray cpu=8 start=53.46 finish=97.39
2580702) povray cpu=3 start=53.46 finish=97.40
2580703) povray cpu=13 start=53.46 finish=97.39
2580704) povray cpu=3 start=53.47 finish=97.34
2580705) povray cpu=15 start=53.72 finish=97.14
2580706) povray cpu=9 start=53.72 finish=53.96
2580707) povray cpu=0 start=53.96 finish=53.96
2580708) povray cpu=5 start=53.97 finish=97.14
2580709) povray cpu=9 start=53.97 finish=53.97
2580710) povray cpu=10 start=53.97 finish=53.97
2580711) povray cpu=9 start=54.02 finish=55.04
2580712) povray cpu=10 start=54.02 finish=54.20
2580713) povray cpu=11 start=54.02 finish=54.02
2580714) ?? cpu=0 start=54.03 finish=0.00
2580715) povray cpu=15 start=54.03 finish=54.03
2580716) povray cpu=4 start=54.03 finish=54.03
2580717) povray cpu=14 start=54.03 finish=54.03
2580718) povray cpu=13 start=54.03 finish=54.03
2580719) povray cpu=11 start=54.03 finish=54.03
2580720) povray cpu=15 start=54.03 finish=54.03
2580721) povray cpu=12 start=54.03 finish=54.03
2580722) povray cpu=13 start=54.03 finish=54.03
2580723) povray cpu=13 start=54.03 finish=54.03
2580724) povray cpu=14 start=54.03 finish=54.03
2580725) povray cpu=13 start=54.03 finish=54.03
2580726) povray cpu=13 start=54.03 finish=54.03
2580727) povray cpu=9 start=55.08 finish=55.09
2580728) povray cpu=9 start=55.13 finish=95.89
2580729) povray cpu=3 start=55.13 finish=95.84
2580730) povray cpu=13 start=55.13 finish=96.33
2580731) povray cpu=14 start=55.13 finish=96.38
2580732) povray cpu=10 start=55.13 finish=96.65
2580733) povray cpu=11 start=55.13 finish=96.66
2580734) povray cpu=15 start=55.13 finish=96.56
2580735) povray cpu=12 start=55.13 finish=96.40
2580736) povray cpu=0 start=55.13 finish=95.82
2580737) povray cpu=8 start=55.13 finish=95.96
2580738) povray cpu=11 start=55.13 finish=95.81
2580739) povray cpu=0 start=55.13 finish=96.84
2580740) povray cpu=9 start=55.13 finish=97.00
2580741) povray cpu=15 start=55.13 finish=95.82
2580742) povray cpu=1 start=55.13 finish=96.27
2580743) povray cpu=2 start=55.13 finish=95.92