Comments (2)
some details.
qnnpack
conv2d/1, time_ct = 0.011000 // qnnp_create_convolution2d_nhwc_q8
conv2d/1, time_st = 5.091000 // qnnp_setup_convolution2d_nhwc_q8
conv2d/1, time_op = 12.962000 // qnnp_run_operator
depthwise_conv2d/4, time_ct = 0.021000
depthwise_conv2d/4, time_st = 88.609000
depthwise_conv2d/4, time_op = 6.246000
'''
tf-lite
conv2d/1, time_op = 52.826000
depthwise_conv2d/4, time_op = 13.568000
'''
it should be noted that I have listed the time required for each part of qnnpack convolution(create\ setup\run), while for tf-lite,i package the three ops, so i just print out the time required for the whole convolution calculation。
from qnnpack.
Typically, you call create
and setup
only once, and then call qnnp_run_operator
for each inference run. As you can see, qnnp_run_operator
is much cheaper in QNNPACK than in TFLite.
from qnnpack.
Related Issues (20)
- Per channel quantization for the weights HOT 1
- Implementation detail things... HOT 9
- Deconvolutional Layer is slow for large kernel size HOT 2
- No performance difference between ARMv7 (32-bit) Android and ARM64 (64-bit) Android HOT 1
- Request to add Conv1d related operation HOT 1
- What's the relationship between the ASM and C version micro kernel? HOT 5
- Request to add a Concat layer HOT 1
- Crashed when running conv3x3 test
- Typo in the announcement blog? HOT 1
- 3d convolution support HOT 1
- Is the code multithreaded for arm64? HOT 7
- build error
- is there any example to use fp16 gemm?
- COMPAIR ERROR HOT 2
- How to build dependencies separately
- Is QNNPACK still being developed? HOT 1
- Understanding XZP ukernel HOT 1
- Build failed on rasp pi 4 (Error: selected processor does not support `yield' in ARM mode) HOT 1
- Support for nn.ConvTranspose2d
- Qnnpack accuracy very poor on unet model HOT 3
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from qnnpack.