当前位置:网站首页>C6748 software simulation and hardware test - with detailed FFT hardware measurement time
C6748 software simulation and hardware test - with detailed FFT hardware measurement time
2022-04-23 19:29:00 【ToneChip】
Recently C6748 DSP Do on FFT Time measurement , notice 8192 spot FFT The measurement time is 467us, And the result of my hardware test is 882us, I wonder what's wrong ????
The results look good , Find out CCS There's one " Software emulation " and " Hardware simulation "
Software simulation mode ( No memory access latency ) The difference between and hardware simulation lies in the use of ccxml Different files
Now sort out the comparison results
The debugging condition is that the optimization level is opened to 3
Actual detailed running time of hardware (6748 Dominant frequency run 456M Hz)
--------------------------- Cache Disabled --------------------------- | ||||
DSPF_sp_cfftr2_dit | DSPF_sp_fftSPxSP | DSPF_sp_icfftr2_dif | DSPF_sp_ifftSPxSP | |
32 | 47.51us | 29.46us | 46.11us | 29.32us |
64 | 106.21us | 57.21us | 105.58us | 56.39us |
128 | 239.95us | 154.52us | 238.69us | 152.84us |
256 | 537.04us | 308.41us | 534.76us | 302.72us |
512 | 1184.55us | 756.07us | 1182.24us | 748.86us |
1024 | 2601.47us | 1485.80us | 2598.27us | 1477.94us |
2048 | 5674.80us | 3578.33us | 5666.62us | 3559.37us |
4096 | 12527.40us | 7147.63us | 12470.40us | 7125.05us |
8192 | 27472.50us | 17031.80us | 27218.28us | 16981.55us |
16384 | 59800.18us | 34004.56us | 59975.67us | 34046.73us |
32768 | 129329.05us | 78609.17us | 128633.07us | 78713.64us |
65536 | ||||
131072 | ||||
--------------------------- Cache Enabled --------------------------- | ||||
DSPF_sp_cfftr2_dit | DSPF_sp_fftSPxSP | DSPF_sp_icfftr2_dif | DSPF_sp_ifftSPxSP | |
32 | 0.94us | 1.07us | 1.24us | 1.38us |
64 | 2.06us | 1.92us | 2.51us | 2.59us |
128 | 4.57us | 4.26us | 5.41us | 5.67us |
256 | 10.15us | 8.18us | 11.99us | 11.01us |
512 | 22.43us | 18.39us | 26.37us | 24.08us |
1024 | 49.23us | 36.16us | 58.08us | 46.76us |
2048 | 107.34us | 83.42us | 127.30us | 101.94us |
4096 | 265.84us | 199.11us | 312.63us | 245.76us |
8192 | 880.92us | 769.32us | 1091.82us | 880.77us |
16384 | 1910.95us | 2456.24us | 2699.05us | 2271.15us |
32768 | 6882.59us | 11756.84us | 8638.64us | 11724.79us |
65536 | ||||
131072 |
The picture version is a little clearer
版权声明
本文为[ToneChip]所创,转载请带上原文链接,感谢
https://yzsam.com/2022/04/202204231923488270.html
边栏推荐
- Prefer composition to inheritance
- MySQL数据库 - 连接查询
- 什么是消息队列
- A brief explanation of golang's keyword "competence"
- Go modules daily use
- RuntimeError: Providing a bool or integral fill value without setting the optional `dtype` or `out`
- Virtual machine performance monitoring and fault handling tools
- C语言的十六进制printf为何输出有时候输出带0xFF有时没有
- What is a message queue
- uIP1.0 主动发送的问题理解
猜你喜欢
Zero base to build profit taking away CPS platform official account
深度分析数据恢复原理——那些数据可以恢复那些不可以数据恢复软件
Kubernetes入门到精通-KtConnect(全称Kubernetes Toolkit Connect)是一款基于Kubernetes环境用于提高本地测试联调效率的小工具。
Deep learning -- Summary of Feature Engineering
Garbage collector and memory allocation strategy
【webrtc】Add x264 encoder for CEF/Chromium
山大网安靶场实验平台项目-个人记录(五)
Scrum Patterns之理解各种团队模式
Reflection on the performance of some OpenGL operations in the past
C6748 软件仿真和硬件测试 ---附详细FFT硬件测量时间
随机推荐
Openlayers draw rectangle
arcgis js api dojoConfig配置
视频理解-Video Understanding
Steps to build a deep learning environment GPU
Command - sudo
Modify the font size of hint in editext
An example of using JNI to directly access surface data
First experience of using fluent canvas
Intuitive understanding of the essence of two-dimensional rotation
什么是消息队列
Application of DCT transform
Virtual machine performance monitoring and fault handling tools
Zero base to build profit taking away CPS platform official account
OpenHarmony开源开发者成长计划,寻找改变世界的开源新生力!
Inject Autowired fields into ordinary beans
Speex维纳滤波与超几何分布的改写
Openlayers 5.0 reload the map when the map container size changes
An algorithm problem was encountered during the interview_ Find the mirrored word pairs in the dictionary
Openlayers 5.0 thermal diagram
[report] Microsoft: application of deep learning methods in speech enhancement