A speech to text process using Tensorflow to transcribe a three minute audio recording. One workload that runs in about a minute and then three times. Looks to be single-threaded
Topdown profile looks to be backend bound.
AMD metrics confirm a backend-stall heavy workload with low frontend stalls. There is not much floating point. A good amount of L2 access.
elapsed 160.615
on_cpu 0.070 # 1.12 / 16 cores
utime 169.116
stime 10.486
nvcsw 487526 # 99.62%
nivcsw 1847 # 0.38%
inblock 0 # 0.00/sec
onblock 12568 # 78.25/sec
cpu-clock 177998788938 # 177.999 seconds
task-clock 178430160190 # 178.430 seconds
page faults 235378 # 1319.160/sec
context switches 489985 # 2746.088/sec
cpu migrations 23465 # 131.508/sec
major page faults 3416 # 19.145/sec
minor page faults 231962 # 1300.016/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 49259821446 # 63.246 branches per 1000 inst
branch misses 521285591 # 1.06% branch miss
conditional 40408849370 # 51.882 conditional branches per 1000 inst
indirect 1381988941 # 1.774 indirect branches per 1000 inst
cpu-cycles 696267329926 # 0.24 GHz
instructions 772925724731 # 1.11 IPC
slots 1390055158284 #
retiring 256880226504 # 18.5% (18.7%)
-- ucode 151528830 # 0.0%
-- fastpath 256728697674 # 18.5%
frontend 58842049778 # 4.2% ( 4.3%) low
-- latency 47353735122 # 3.4%
-- bandwidth 11488314656 # 0.8%
backend 1047097332617 # 75.3% (76.2%) high
-- cpu 130578369248 # 9.4%
-- memory 916518963369 # 65.9%
speculation 11335440378 # 0.8% ( 0.8%) low
-- branch mispredict 10259441048 # 0.7%
-- pipeline restart 1075999330 # 0.1%
smt-contention 15848595211 # 1.1% ( 0.0%)
cpu-cycles 697932624106 # 0.28 GHz
instructions 777877029300 # 1.11 IPC
instructions 259176178766 # 172.445 l2 access per 1000 inst
l2 hit from l1 26157068658 # 31.31% l2 miss
l2 miss from l1 2105517564 #
l2 hit from l2 pf 6648857572 #
l3 hit from l2 pf 269934316 #
l3 miss from l2 pf 11617658376 #
instructions 259534288888 # 15.867 float per 1000 inst
float 512 65 # 0.000 AVX-512 per 1000 inst
float 256 2318481 # 0.009 AVX-256 per 1000 inst
float 128 4115648027 # 15.858 AVX-128 per 1000 inst
float MMX 0 # 0.000 MMX per 1000 inst
float scalar 4 # 0.000 scalar per 1000 inst
instructions 778907715084 #
opcache 94567228200 # 121.410 opcache per 1000 inst
opcache miss 4023676179 # 4.3% opcache miss rate
l1 dTLB miss 4315198834 # 5.540 L1 dTLB per 1000 inst
l2 dTLB miss 1044990269 # 1.342 L2 dTLB per 1000 inst
instructions 778450113458 #
icache 7298837934 # 9.376 icache per 1000 inst
icache miss 1369530995 # 18.8% icache miss rate
l1 iTLB miss 23755944 # 0.031 L1 iTLB per 1000 inst
l2 iTLB miss 0 # 0.000 L2 iTLB per 1000 inst
tlb flush 16742 # 0.000 TLB flush per 1000 inst
Intel metrics confirm both L3 and dram-bound natures of the memory-bound stalls.
elapsed 201.626
on_cpu 0.056 # 0.90 / 16 cores
utime 176.392
stime 5.582
nvcsw 468751 # 97.63%
nivcsw 11384 # 2.37%
inblock 1208 # 5.99/sec
onblock 1288 # 6.39/sec
cpu-clock 180258810774 # 180.259 seconds
task-clock 180510870708 # 180.511 seconds
page faults 222169 # 1230.779/sec
context switches 480881 # 2664.000/sec
cpu migrations 38123 # 211.195/sec
major page faults 2428 # 13.451/sec
minor page faults 219741 # 1217.328/sec
alignment faults 0 # 0.000/sec
emulation faults 0 # 0.000/sec
branches 49026825305 # 59.224 branches per 1000 inst
branch misses 354171008 # 0.72% branch miss
conditional 49026840985 # 59.224 conditional branches per 1000 inst
indirect 3105678778 # 3.752 indirect branches per 1000 inst
slots 3095032907966 #
retiring 773986641565 # 25.0% (25.0%)
-- ucode 10055304339 # 0.3%
-- fastpath 763931337226 # 24.7%
frontend 74881303981 # 2.4% ( 2.4%) low
-- latency 36919463368 # 1.2%
-- bandwidth 37961840613 # 1.2%
backend 2231130561294 # 72.1% (72.1%) high
-- cpu 1108840747474 # 35.8%
-- memory 1122289813820 # 36.3%
speculation 42072490016 # 1.4% ( 1.4%)
-- branch mispredict 39314118809 # 1.3%
-- pipeline restart 2758371207 # 0.1%
smt-contention 0 # 0.0% ( 0.0%)
cpu-cycles 618989502966 # 0.24 GHz
instructions 878893138266 # 1.42 IPC
l2 access 144912559758 # 188.820 l2 access per 1000 inst
l2 miss 98850887983 # 68.21% l2 miss
cpu-cycles 540075543736 # 42.5% memory latency
load stalls 228742926778 # 0.0% l1 bound
l1 miss 463786101064 # 47.1% l2 bound
l2 miss 209158694764 # 14.2% l3 bound
l3 miss 132645464628 # 24.6% dram bound
store_stalls 624046835 # 0.1% store bound
Process profile shows time spent on the deepspeech process.
450 processes
102 deepspeech 5788.84 199.24
68 clinfo 16.53 6.33
38 vulkaninfo 1.34 0.96
4 vulkani:disk$0 0.15 0.11
6 glxinfo:gdrv0 0.09 0.09
6 glxinfo:gl0 0.09 0.09
2 llvmpipe-0 0.07 0.05
2 llvmpipe-1 0.07 0.05
2 llvmpipe-10 0.07 0.05
2 llvmpipe-11 0.07 0.05
2 llvmpipe-12 0.07 0.05
2 llvmpipe-13 0.07 0.05
2 llvmpipe-14 0.07 0.05
2 llvmpipe-15 0.07 0.05
2 llvmpipe-2 0.07 0.05
2 llvmpipe-3 0.07 0.05
2 llvmpipe-4 0.07 0.05
2 llvmpipe-5 0.07 0.05
2 llvmpipe-6 0.07 0.05
2 llvmpipe-7 0.07 0.05
2 llvmpipe-8 0.07 0.05
2 llvmpipe-9 0.07 0.05
6 php 0.06 0.09
2 glxinfo 0.06 0.03
6 clang 0.05 0.07
2 glxinfo:cs0 0.05 0.03
2 glxinfo:disk$0 0.05 0.03
2 glxinfo:sh0 0.05 0.03
2 glxinfo:shlo0 0.05 0.03
3 rocminfo 0.00 0.03
1 lspci 0.00 0.02
81 sh 0.00 0.00
12 gcc 0.00 0.00
9 gsettings 0.00 0.00
8 stat 0.00 0.00
8 systemd-detect- 0.00 0.00
6 llvm-link 0.00 0.00
5 gmain 0.00 0.00
5 phoronix-test-s 0.00 0.00
3 deepspeech-run 0.00 0.00
2 dconf worker 0.00 0.00
2 lscpu 0.00 0.00
2 uname 0.00 0.00
2 which 0.00 0.00
2 xset 0.00 0.00
1 cc 0.00 0.00
1 date 0.00 0.00
1 dirname 0.00 0.00
1 dmesg 0.00 0.00
1 dmidecode 0.00 0.00
1 grep 0.00 0.00
1 ifconfig 0.00 0.00
1 ip 0.00 0.00
1 lsmod 0.00 0.00
1 mktemp 0.00 0.00
1 ps 0.00 0.00
1 qdbus 0.00 0.00
1 readlink 0.00 0.00
1 realpath 0.00 0.00
1 sed 0.00 0.00
1 sort 0.00 0.00
1 stty 0.00 0.00
1 systemctl 0.00 0.00
1 template.sh 0.00 0.00
1 wc 0.00 0.00
1 xrandr 0.00 0.00
0 processes running
47 maximum processes
Computation blocks
906663) deepspeech-run cpu=5 start=5.99 finish=49.12
906664) deepspeech cpu=5 start=6.00 finish=49.11
906665) deepspeech cpu=4 start=6.01 finish=49.11
906666) deepspeech cpu=14 start=6.01 finish=49.11
906667) deepspeech cpu=13 start=6.01 finish=49.11
906668) deepspeech cpu=2 start=6.01 finish=49.11
906669) deepspeech cpu=9 start=6.01 finish=49.11
906670) deepspeech cpu=7 start=6.01 finish=49.11
906671) deepspeech cpu=1 start=6.01 finish=49.11
906672) deepspeech cpu=8 start=6.01 finish=49.11
906673) deepspeech cpu=12 start=6.01 finish=49.11
906674) deepspeech cpu=8 start=6.01 finish=49.11
906675) deepspeech cpu=15 start=6.01 finish=49.11
906676) deepspeech cpu=6 start=6.01 finish=49.11
906677) deepspeech cpu=10 start=6.01 finish=49.11
906678) deepspeech cpu=3 start=6.01 finish=49.11
906679) deepspeech cpu=9 start=6.01 finish=49.11
906680) deepspeech cpu=11 start=6.01 finish=49.11
906681) deepspeech cpu=7 start=6.01 finish=49.11
906682) deepspeech cpu=11 start=6.01 finish=49.11
906683) deepspeech cpu=0 start=6.01 finish=49.11
906684) deepspeech cpu=14 start=6.01 finish=49.11
906685) deepspeech cpu=6 start=6.01 finish=49.11
906686) deepspeech cpu=13 start=6.01 finish=49.11
906687) deepspeech cpu=4 start=6.01 finish=49.11
906688) deepspeech cpu=12 start=6.01 finish=49.11
906689) deepspeech cpu=13 start=6.01 finish=49.11
906690) deepspeech cpu=15 start=6.01 finish=49.11
906691) deepspeech cpu=5 start=6.01 finish=49.11
906692) deepspeech cpu=10 start=6.01 finish=49.11
906693) deepspeech cpu=1 start=6.01 finish=49.11
906694) deepspeech cpu=2 start=6.01 finish=49.11
906695) deepspeech cpu=0 start=6.01 finish=49.11
906696) deepspeech cpu=3 start=6.01 finish=49.11
906697) deepspeech cpu=11 start=6.01 finish=49.11