Deepstream-app latency increasing

• Hardware Platform: JetsonNano
• DeepStream Version: 5.0
• JetPack Version (valid for Jetson only): 32.4.2-20200408182156
• TensorRT Version: 7.1.0-1+cuda10.2

When running deepstream-app with my own Yolov3_tiny based model, input is RTSP stream 30FPS, even the interval was set to 10 in primary gie config, and the PERF statistic says running at 40FPS, the output displayed by egl is slower than real video speed, and the video latency increasing time to time, just 4 seconds at starting, and then 30s after a while.
How can I speed up the display output? And reduce/constant the latency? Thank you.

Hi,
Please share the config file for reference. Thanks.

Hi,
Please check my reply here.
In summary, you can expect to see a growing latency if nvinfer takes longer to process that the time of one frame. There are several ways to optimize your pipeline in order to avoid this:

  1. Adding is-live=true to your source
  2. Adding sync=false to your sink
  3. Adding leaky queues before the bottleneck elements: queue leaky=2 max-size-buffers=1
  4. Set interval=n in nvinfer to skip inference in some buffers

Hi @DaneLLL and @miguel.taylor,
Thank you for your replies.

My config file as below, I attached my pipeline graph. Now I can get good performance and latency, but I have to drop the decoded frame interleave (keep - drop - keep - drop -…), and the nvinfer interval = 6. Without decoder drops, and nvinfer interval < 6, the slow output & latency increasing is still presented.
You can see in the pipeline, I added a leaky queue in front of nvinfer, and added videorate infront of h264enc.

With leaky queue in front of nvinfer and no decoded frame dropping, the slow output & high latency are still there. So it looks like the leaky queue doesn’t matter?

This is my config file:

[application]
enable-perf-measurement=1
perf-measurement-interval-sec=5
#gie-kitti-output-dir=streamscl

[tiled-display]
enable=1
rows=1
columns=1
width=1280
height=960
gpu-id=0
#(0): nvbuf-mem-default - Default memory allocated, specific to particular platform
#(1): nvbuf-mem-cuda-pinned - Allocate Pinned/Host cuda memory, applicable for Tesla
#(2): nvbuf-mem-cuda-device - Allocate Device cuda memory, applicable for Tesla
#(3): nvbuf-mem-cuda-unified - Allocate Unified cuda memory, applicable for Tesla
#(4): nvbuf-mem-surface-array - Allocate Surface Array memory, applicable for Jetson
nvbuf-memory-type=0

[source0]
enable=1
#Type - 1=CameraV4L2 2=URI 3=MultiURI
type=2
#uri=file://../../samples/streams/sample_1080p_h264.mp4
uri=rtsp://User:5h1e67665@10.100.30.199:554/video1
num-sources=1
gpu-id=0
# (0): memtype_device   - Memory type Device
# (1): memtype_pinned   - Memory type Host Pinned
# (2): memtype_unified  - Memory type Unified
cudadec-memtype=0
drop-frame-interval=2

[sink0]
enable=1
#Type - 1=FakeSink 2=EglSink 3=File 4=Rtsp
type=4
sync=0
source-id=0
gpu-id=0
nvbuf-memory-type=0
# Codec 1=H264, 2=H265, 3=MPEG4
codec=1
bitrate=8000000
udp-port=5400
rtsp-port=8554
iframeinterval=30

[osd]
enable=1
gpu-id=0
border-width=2
text-size=0
text-color=1;1;1;1;
text-bg-color=0.3;0.3;0.3;1
font=Serif
show-clock=0
clock-x-offset=800
clock-y-offset=820
clock-text-size=12
clock-color=1;0;0;0
nvbuf-memory-type=0

[streammux]
gpu-id=0
##Boolean property to inform muxer that sources are live
live-source=1
batch-size=1
##time out in usec, to wait after the first buffer is available
##to push the batch even if the complete batch is not formed
batched-push-timeout=10000
## Set muxer output width and height
width=1280
height=960
##Enable to maintain aspect ratio wrt source, and allow black borders, works
##along with width, height properties
enable-padding=0
nvbuf-memory-type=0

# config-file property is mandatory for any gie section.
# Other properties are optional and if set will override the properties set in
# the infer config file.
[primary-gie]
enable=1
gpu-id=0
#model-engine-file=model_b1_gpu0_fp32.engine
labelfile-path=labels2.txt
batch-size=1
#Required by the app for OSD, not a plugin property
bbox-border-color0=1;0;0;1
interval=6
gie-unique-id=1
nvbuf-memory-type=0
config-file=config_infer_primary_yoloV3_tiny_2.txt

[tracker]
enable=1
tracker-width=640
tracker-height=384
ll-lib-file=/opt/nvidia/deepstream/deepstream-5.0/lib/libnvds_mot_klt.so

[tests]
file-loop=0

Hi,
Please also execute sudo jetson_clocks to run in max performance. You can check system status through sudo tegrastats.

Thank you. I already run “sudo jetson_clocks” after boot.

This is statistic:
When sources drop-frame-interval = 2:

RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [37%@1479,37%@1479,41%@1479,43%@1479] EMC_FREQ 47%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@35C PMIC@100C GPU@30.5C AO@43C thermal@33.5C POM_5V_IN 7725/7725 POM_5V_GPU 2254/2254 POM_5V_CPU 1463/1463
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [39%@1479,37%@1479,30%@1479,29%@1479] EMC_FREQ 47%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@29.5C CPU@35.5C PMIC@100C GPU@31C AO@45.5C thermal@32.5C POM_5V_IN 9101/8413 POM_5V_GPU 4386/3320 POM_5V_CPU 665/1064
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [45%@1479,32%@1479,43%@1479,37%@1479] EMC_FREQ 47%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@29.5C CPU@36.5C PMIC@100C GPU@31.5C AO@42C thermal@32.75C POM_5V_IN 7205/8010 POM_5V_GPU 2706/3115 POM_5V_CPU 915/1014
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [38%@1479,44%@1479,46%@1479,42%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@35C PMIC@100C GPU@29.5C AO@43.5C thermal@32.5C POM_5V_IN 7459/7872 POM_5V_GPU 1941/2821 POM_5V_CPU 1545/1147
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [39%@1479,39%@1479,35%@1479,38%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@37C PMIC@100C GPU@31.5C AO@45.5C thermal@33.5C POM_5V_IN 9321/8162 POM_5V_GPU 4262/3109 POM_5V_CPU 977/1113
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [45%@1479,40%@1479,35%@1479,40%@1479] EMC_FREQ 47%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@29.5C CPU@36.5C PMIC@100C GPU@30C AO@42.5C thermal@34.25C POM_5V_IN 7165/7996 POM_5V_GPU 2786/3055 POM_5V_CPU 757/1053
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [41%@1479,40%@1479,40%@1479,40%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@35.5C PMIC@100C GPU@31C AO@43.5C thermal@34.25C POM_5V_IN 8213/8027 POM_5V_GPU 2523/2979 POM_5V_CPU 1655/1139
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [41%@1479,44%@1479,42%@1479,34%@1479] EMC_FREQ 47%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@37C PMIC@100C GPU@31.5C AO@45.5C thermal@33.75C POM_5V_IN 9321/8188 POM_5V_GPU 4340/3149 POM_5V_CPU 899/1109
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [44%@1479,38%@1479,43%@1479,42%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 54%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@38C PMIC@100C GPU@31C AO@43.5C thermal@33.25C POM_5V_IN 8187/8188 POM_5V_GPU 3322/3168 POM_5V_CPU 1067/1104
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [47%@1479,42%@1479,39%@1479,46%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@34.5C PMIC@100C GPU@30.5C AO@42C thermal@33C POM_5V_IN 6129/7982 POM_5V_GPU 1240/2976 POM_5V_CPU 1440/1138
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [42%@1479,45%@1479,39%@1479,51%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30C CPU@35.5C PMIC@100C GPU@31C AO@44C thermal@34.25C POM_5V_IN 7949/7979 POM_5V_GPU 2566/2938 POM_5V_CPU 1421/1164
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [39%@1479,58%@1479,42%@1479,37%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30.5C CPU@37.5C PMIC@100C GPU@31.5C AO@45.5C thermal@33.5C POM_5V_IN 9282/8088 POM_5V_GPU 3753/3006 POM_5V_CPU 1407/1184
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [48%@1479,36%@1479,42%@1479,41%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30.5C CPU@37C PMIC@100C GPU@31.5C AO@46.5C thermal@33C POM_5V_IN 9424/8190 POM_5V_GPU 4372/3111 POM_5V_CPU 938/1165
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [36%@1479,34%@1479,45%@1479,45%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30.5C CPU@37C PMIC@100C GPU@32C AO@45.5C thermal@34C POM_5V_IN 9023/8250 POM_5V_GPU 4465/3208 POM_5V_CPU 588/1124
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [42%@1479,40%@1479,41%@1479,39%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@30.5C CPU@35C PMIC@100C GPU@30.5C AO@42.5C thermal@33.5C POM_5V_IN 6410/8127 POM_5V_GPU 2040/3130 POM_5V_CPU 1001/1115
RAM 3070/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [38%@1479,50%@1479,48%@1479,46%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 60%@921 NVENC 716 NVDEC 716 APE 25 PLL@30.5C CPU@35C PMIC@100C GPU@31C AO@43C thermal@34.25C POM_5V_IN 6778/8043 POM_5V_GPU 1634/3036 POM_5V_CPU 1632/1148
RAM 3073/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [50%@1479,37%@1479,44%@1479,42%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@30.5C CPU@37C PMIC@100C GPU@31C AO@45.5C thermal@34.25C POM_5V_IN 8959/8097 POM_5V_GPU 3373/3056 POM_5V_CPU 1449/1165
RAM 3073/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [45%@1479,31%@1479,38%@1479,41%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@31C CPU@37.5C PMIC@100C GPU@32C AO@46C thermal@35.25C POM_5V_IN 9580/8179 POM_5V_GPU 4411/3131 POM_5V_CPU 1171/1166
RAM 3071/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [45%@1479,39%@1479,33%@1479,34%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@30.5C CPU@36.5C PMIC@100C GPU@31C AO@42.5C thermal@33.75C POM_5V_IN 6600/8096 POM_5V_GPU 1637/3053 POM_5V_CPU 1517/1184
RAM 3072/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [48%@1479,39%@1479,37%@1479,36%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 87%@921 NVENC 716 NVDEC 716 APE 25 PLL@31C CPU@37C PMIC@100C GPU@31C AO@43.5C thermal@33C POM_5V_IN 7233/8053 POM_5V_GPU 1626/2981 POM_5V_CPU 1864/1218
RAM 3072/3956MB (lfb 2x4MB) SWAP 625/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [41%@1479,34%@1479,50%@1479,46%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@31C CPU@37.5C PMIC@100C GPU@31.5C AO@46C thermal@34.75C POM_5V_IN 9580/8125 POM_5V_GPU 3591/3010 POM_5V_CPU 1873/1249

And when drop-frame-interval = 1:

RAM 3078/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [59%@1479,33%@1479,33%@1479,39%@1479] EMC_FREQ 47%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@39.5C PMIC@100C GPU@33.5C AO@48C thermal@34.5C POM_5V_IN 9533/6930 POM_5V_GPU 4418/2516 POM_5V_CPU 979/922
RAM 3079/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [37%@1479,43%@1479,32%@1479,32%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@36.5C PMIC@100C GPU@32.5C AO@44C thermal@34.5C POM_5V_IN 6370/6902 POM_5V_GPU 1802/2481 POM_5V_CPU 1201/936
RAM 3079/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [40%@1479,38%@1479,43%@1479,42%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@37.5C PMIC@100C GPU@32.5C AO@46.5C thermal@34.25C POM_5V_IN 8489/6977 POM_5V_GPU 3193/2515 POM_5V_CPU 1300/953
RAM 3079/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [43%@1479,40%@1479,35%@1479,39%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@38.5C PMIC@100C GPU@33.5C AO@47.5C thermal@34.25C POM_5V_IN 9517/7093 POM_5V_GPU 4621/2610 POM_5V_CPU 823/947
RAM 3080/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [38%@1479,35%@1479,38%@1479,40%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@36.5C PMIC@100C GPU@32C AO@44C thermal@35.5C POM_5V_IN 6099/7049 POM_5V_GPU 1564/2565 POM_5V_CPU 1203/958
RAM 3081/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [53%@1479,50%@1479,39%@1479,34%@1479] EMC_FREQ 44%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@37C PMIC@100C GPU@32C AO@44.5C thermal@36C POM_5V_IN 6179/7013 POM_5V_GPU 1121/2505 POM_5V_CPU 1680/988
RAM 3081/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [56%@1479,46%@1479,45%@1479,37%@1479] EMC_FREQ 44%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@38C PMIC@100C GPU@32.5C AO@46C thermal@35C POM_5V_IN 8147/7058 POM_5V_GPU 2606/2509 POM_5V_CPU 1697/1017
RAM 3082/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [43%@1479,44%@1479,42%@1479,55%@1479] EMC_FREQ 44%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@38.5C PMIC@100C GPU@33.5C AO@47.5C thermal@36.5C POM_5V_IN 9556/7154 POM_5V_GPU 4105/2570 POM_5V_CPU 1329/1029
RAM 3082/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [48%@1479,54%@1479,31%@1479,47%@1479] EMC_FREQ 44%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32.5C CPU@39C PMIC@100C GPU@33.5C AO@48C thermal@35.5C POM_5V_IN 9533/7242 POM_5V_GPU 4340/2636 POM_5V_CPU 1094/1031
RAM 3083/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [51%@1479,59%@1479,42%@1479,37%@1479] EMC_FREQ 44%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32.5C CPU@39C PMIC@100C GPU@34C AO@47C thermal@35C POM_5V_IN 9431/7321 POM_5V_GPU 4276/2694 POM_5V_CPU 982/1029
RAM 3083/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [58%@1479,53%@1479,27%@1479,49%@1479] EMC_FREQ 44%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@38.5C PMIC@100C GPU@34C AO@47.5C thermal@35.75C POM_5V_IN 9635/7400 POM_5V_GPU 4614/2760 POM_5V_CPU 902/1025
RAM 3084/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [53%@1479,40%@1479,52%@1479,48%@1479] EMC_FREQ 44%@1600 GR3D_FREQ 81%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@38C PMIC@100C GPU@33.5C AO@46C thermal@35.75C POM_5V_IN 8568/7439 POM_5V_GPU 3553/2787 POM_5V_CPU 1028/1025
RAM 3084/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [46%@1479,47%@1479,61%@1479,53%@1479] EMC_FREQ 43%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@33C CPU@39C PMIC@100C GPU@34C AO@46.5C thermal@35.75C POM_5V_IN 9131/7494 POM_5V_GPU 3968/2825 POM_5V_CPU 1023/1025
RAM 3085/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [59%@1479,68%@1479,42%@1479,48%@1479] EMC_FREQ 42%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@40C PMIC@100C GPU@33.5C AO@47.5C thermal@35.5C POM_5V_IN 9971/7571 POM_5V_GPU 4294/2871 POM_5V_CPU 1561/1042
RAM 3086/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [31%@1479,45%@1479,36%@1479,44%@1479] EMC_FREQ 43%@1600 GR3D_FREQ 0%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@37C PMIC@100C GPU@32.5C AO@44.5C thermal@37C POM_5V_IN 6530/7540 POM_5V_GPU 2240/2852 POM_5V_CPU 801/1034
RAM 3086/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [26%@1479,23%@1479,23%@1479,22%@1479] EMC_FREQ 45%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@39C PMIC@100C GPU@34C AO@48.5C thermal@36.25C POM_5V_IN 9234/7590 POM_5V_GPU 4550/2901 POM_5V_CPU 627/1022
RAM 3087/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [20%@1479,14%@1479,28%@1479,24%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@38.5C PMIC@100C GPU@33.5C AO@47.5C thermal@34.75C POM_5V_IN 9116/7633 POM_5V_GPU 4636/2951 POM_5V_CPU 471/1006
RAM 3087/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [24%@1479,18%@1479,28%@1479,23%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@38C PMIC@100C GPU@33.5C AO@48C thermal@36C POM_5V_IN 8673/7662 POM_5V_GPU 4054/2982 POM_5V_CPU 590/995
RAM 3088/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [33%@1479,19%@1479,26%@1479,25%@1479] EMC_FREQ 46%@1600 GR3D_FREQ 99%@921 NVENC 716 NVDEC 716 APE 25 PLL@32C CPU@37C PMIC@100C GPU@33.5C AO@46.5C thermal@36C POM_5V_IN 7471/7657 POM_5V_GPU 2861/2978 POM_5V_CPU 715/987
RAM 3088/3956MB (lfb 2x4MB) SWAP 627/1978MB (cached 29MB) IRAM 0/252kB(lfb 252kB) CPU [27%@1479,26%@1479,19%@1479,21%@1479] EMC_FREQ 47%@1600 GR3D_FREQ 49%@921 NVENC 716 NVDEC 716 APE 25 PLL@32.5C CPU@37C PMIC@100C GPU@32.5C AO@45.5C thermal@35C POM_5V_IN 6948/7638 POM_5V_GPU 1993/2952 POM_5V_CPU 996/988

In the pipeline you uploaded I can’t read the names of the elements, do you mind uploading it in a higher resolution or transcribing the elements into text?

My guess is that another element is also limiting the framerate. Try adding a queue also before the first nvvideoconvert and any other element that you suspect is limiting the framerate.

You can also use GstShark to measure latency and framerate at each source pad.

Hi @miguel.taylor,
Do you have any suggestion to convert graph from .dot file to .png file with high resolution? I guess the graphiz dot app to convert .dot file to image, but it is automatically reduce the resolution of the output image :(
Here is command I used:

dot -Tpng -x -Gdpi=1200 -o playing.png 0.01.26.445328829-ds-app-playing.dot

I have tried to convert to SVG image, but the output svg file can’t be opened. GIMP returns error when open the output svg file.
Thank you.

I just use

dot -Tpng 0.00.00.015559436-gst-launch.PAUSED_PLAYING.dot > pipe.png

I never had a problem with the image resolution.

Your command gives same output with mine.

dot -Tpng 0.01.26.445328829-ds-app-playing.dot > playing.png
dot: graph is too large for cairo-renderer bitmaps. Scaling by 0.530923 to fit

Output png has 32766x699 resolution.

Can you read all the elements in your computer? It is possible that compression during upload or download is messing the image.

Hi,
The tegrastats shows GPU is at full load. You may set large inerval, or try to reduce some loading from the YoloV3 tiny model.