A cross-platform GPU stress test with OpenGL and Vulkan drivers. There are 160 combinations of drivers, screen sizes and MSAA settings. This test picks 1920×1200 resolution with OpenGL and all MSAA settings. Overall system shows this is now a CPU benchmark and runes with just a few threads.
Topdown profile shows the little CPU time is spent in frontend stalls.
AMD metrics show only 1/30th of a core time spent.
elapsed 1076.862
on_cpu 0.002 # 0.03 / 16 cores
utime 20.084
stime 16.793
nvcsw 162027 # 99.31%
nivcsw 1122 # 0.69%
inblock 8 # 0.01/sec
onblock 13640 # 12.67/sec
cpu-clock 32019749507 # 32.020 seconds
task-clock 32688331876 # 32.688 seconds
page faults 476048 # 14563.239/sec
context switches 168328 # 5149.483/sec
cpu migrations 1007 # 30.806/sec
major page faults 2 # 0.061/sec
minor page faults 476046 # 14563.178/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 8777901101 # 159.914 branches per 1000 inst
branch misses 365812203 # 4.17% branch miss
conditional 5205216990 # 94.828 conditional branches per 1000 inst
indirect 507105797 # 9.238 indirect branches per 1000 inst
cpu-cycles 53961154591 # 0.00 GHz
instructions 54738144413 # 1.01 IPC
slots 108780603978 #
retiring 19232090905 # 17.7% (17.7%)
-- ucode 52330631 # 0.0%
-- fastpath 19179760274 # 17.6%
frontend 64353407460 # 59.2% (59.3%) high
-- latency 58838576598 # 54.1%
-- bandwidth 5514830862 # 5.1%
backend 21298430222 # 19.6% (19.6%)
-- cpu 2146489744 # 2.0%
-- memory 19151940478 # 17.6%
speculation 3655643434 # 3.4% ( 3.4%)
-- branch mispredict 3554005285 # 3.3%
-- pipeline restart 101638149 # 0.1%
smt-contention 240343787 # 0.2% ( 0.0%)
cpu-cycles 54536598047 # 0.00 GHz
instructions 54631269920 # 1.00 IPC
instructions 18245618756 # 37.691 l2 access per 1000 inst
l2 hit from l1 615783822 # 43.13% l2 miss
l2 miss from l1 254071074 #
l2 hit from l2 pf 29368657 #
l3 hit from l2 pf 16021795 #
l3 miss from l2 pf 26518276 #
instructions 18197223178 # 62.137 float per 1000 inst
float 512 64 # 0.000 AVX-512 per 1000 inst
float 256 674 # 0.000 AVX-256 per 1000 inst
float 128 1130722912 # 62.137 AVX-128 per 1000 inst
float MMX 0 # 0.000 MMX per 1000 inst
float scalar 0 # 0.000 scalar per 1000 inst
instructions 54655192673 #
opcache 12529239376 # 229.242 opcache per 1000 inst
opcache miss 3057310348 # 24.4% opcache miss rate
l1 dTLB miss 105870486 # 1.937 L1 dTLB per 1000 inst
l2 dTLB miss 24007702 # 0.439 L2 dTLB per 1000 inst
instructions 54611990576 #
icache 7369764703 # 134.948 icache per 1000 inst
icache miss 1623602659 # 22.0% icache miss rate
l1 iTLB miss 39060559 # 0.715 L1 iTLB per 1000 inst
l2 iTLB miss 0 # 0.000 L2 iTLB per 1000 inst
tlb flush 25177 # 0.000 TLB flush per 1000 inst
Intel metrics
elapsed 1083.609
on_cpu 0.002 # 0.03 / 16 cores
utime 15.516
stime 15.960
nvcsw 116053 # 99.64%
nivcsw 423 # 0.36%
inblock 64 # 0.06/sec
onblock 2344 # 2.16/sec
cpu-clock 27100165512 # 27.100 seconds
task-clock 27755281301 # 27.755 seconds
page faults 375809 # 13540.090/sec
context switches 121705 # 4384.931/sec
cpu migrations 1051 # 37.867/sec
major page faults 1 # 0.036/sec
minor page faults 375808 # 13540.054/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 6642555358 # 150.585 branches per 1000 inst
branch misses 110237614 # 1.66% branch miss
conditional 6642575742 # 150.585 conditional branches per 1000 inst
indirect 478025670 # 10.837 indirect branches per 1000 inst
slots 195395490248 #
retiring 59217355013 # 30.3% (30.3%)
-- ucode 6707380516 # 3.4%
-- fastpath 52509974497 # 26.9%
frontend 30497378415 # 15.6% (15.6%)
-- latency 17340657935 # 8.9%
-- bandwidth 13156720480 # 6.7%
backend 94477580844 # 48.4% (48.4%)
-- cpu 28046200975 # 14.4%
-- memory 66431379869 # 34.0%
speculation 11809887438 # 6.0% ( 6.0%)
-- branch mispredict 10842411803 # 5.5%
-- pipeline restart 967475635 # 0.5%
smt-contention 0 # 0.0% ( 0.0%)
cpu-cycles 34678455881 # 0.00 GHz
instructions 46157951024 # 1.33 IPC
l2 access 1951534064 # 45.139 l2 access per 1000 inst
l2 miss 1285026344 # 65.85% l2 miss
cpu-cycles 32708750023 # 39.3% memory latency
load stalls 6497272907 # 3.9% l1 bound
l1 miss 5216356433 # 4.1% l2 bound
l2 miss 3877407117 # 2.1% l3 bound
l3 miss 3202485247 # 9.8% dram bound
store_stalls 6343834109 # 19.4% store bound
Overall, this is a GPU test and not a CPU test.