第一次压缩正确,参数如下:
[m3video] videnc2DynamicParams -> inputHeight : 1080
[m3video] videnc2DynamicParams -> inputWidth : 1920
[m3video] videnc2DynamicParams -> refFrameRate : 60000
[m3video] videnc2DynamicParams -> targetFrameRate : 60000
[m3video] videnc2DynamicParams -> targetBitRate : 8000000
[m3video] videnc2DynamicParams -> intraFrameInterval : 30
[m3video] videnc2DynamicParams -> generateHeader : 0
[m3video] videnc2DynamicParams -> captureWidth : 1984
[m3video] videnc2DynamicParams -> forceFrame : -1
[m3video] videnc2DynamicParams -> interFrameInterval : 1
[m3video] videnc2DynamicParams -> mvAccuracy : 2
[m3video] videnc2DynamicParams -> sampleAspectRatioHeight : 1080
[m3video] videnc2DynamicParams -> sampleAspectRatioWidth : 1920
[m3video] videnc2DynamicParams -> ignoreOutbufSizeFlag : 1
[m3video] videnc2DynamicParams -> lateAcquireArg : 0
但第二次以后,压缩错误,打印发现,参数是乱的,不知道怎么引起的
[m3video] ——— CH 0 : H264 ENC : Dynamic Params ——-
[m3video]
[m3video] videnc2DynamicParams -> inputHeight : 538976288
[m3video] videnc2DynamicParams -> inputWidth : 538976288
[m3video] videnc2DynamicParams -> refFrameRate : 538976288
[m3video] videnc2DynamicParams -> targetFrameRate : 538976288
[m3video] videnc2DynamicParams -> targetBitRate : 538976288
[m3video] videnc2DynamicParams -> intraFrameInterval : 538976288
[m3video] videnc2DynamicParams -> generateHeader : 538976288
[m3video] videnc2DynamicParams -> captureWidth : 538976288
[m3video] videnc2DynamicParams -> forceFrame : 538976288
[m3video] videnc2DynamicParams -> interFrameInterval : 538976288
[m3video] videnc2DynamicParams -> mvAccuracy : 538976288
[m3video] videnc2DynamicParams -> sampleAspectRatioHeight : 538976288
[m3video] videnc2DynamicParams -> sampleAspectRatioWidth : 538976288
[m3video] videnc2DynamicParams -> ignoreOutbufSizeFlag : 538976288
[m3video] videnc2DynamicParams -> lateAcquireArg : 538976288
Chris Meng:
Xin Jin,
可能的原因有:
1. 有代码修改了该参数,有可能是vidoe m3,也有可能是其他核越界篡改。
2. DDR不稳定,DDR上数据错误
jin xin:
回复 Chris Meng:
DDR的频率是333MHZ,使用multich_tristream_lowPower,只是三个stream,改成1个stream,去掉D1和JPEG。
而且都是有规律的,都是第二次开始。
Chris Meng:
回复 jin xin:
Xin Jin,
不修改use case,是否就没有该问题了?
jin xin:
回复 Chris Meng:
我只能通过串口烧写程序,时间较长,只有明天才能验证不修改use case 是否正常。还有一个奇怪的现象:
FPGA发送0xff,dm385收到0xef
FPGA发送0xef,dm385收到0xdf
FPGA发送0xdf,dm385收到0xcf
FPGA发送0xcf,dm385收到0xbf
FPGA发送0xbf,dm385收到0xaf
以此类推,高4位总比接收到的数据少1.