AN75779 - Did this sample supports recieving data which row size is bigger than DMA buffer?

Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

cross mob
chte_4406741
Level 1
Level 1

Hello the specialist team,

After go through the AN75779 doc and  sample project. I found the DMA was set to 16K byte and whole UVC solution supports 4 DMA bufs.

And I have a question "What's the biggest sensor the FX3 could support? Did it supports 64M or 48M sensors?"

I'm not sure whether it can support the sensors which row size is bigger that the DMA buffer size. ex.row size=9000 pixel * 16bit DCMI=18000 bytes, this value is bigger than DMA size 16K byte.  Did the AN75779 sample project still works? Will there be some data missing, Cause exchanging DMA chain when row data on transmission?

If it works, please help to explain more about the relationship between DMA settings and actually image data

If not , what can I do to figure it out?

0 Likes
1 Solution
JayakrishnaT_76
Moderator
Moderator
Moderator
First question asked 1000 replies posted 750 replies posted

Hello,

Please find my comments below:

1. Your understanding that AN75779 project make use of DMA buffers of 16K is correct. But the number of DMA buffers allocated is actually 8. This is because for each GPIF socket, 4 buffers are allocated. So, the total DMA buffer size used is around 16K*8 = 128K.

2. It is possible to support sensors having row size greater than Buffer size. Please refer to the section 3.4 of AN75779 document. This shows the thread switching procedure that is implemented in the state machine. From this, when the state machine is sampling data into DMA buffers in a thread and the buffer becomes full, it will switch to the next thread and sample data into the DMA buffers allocated with that thread. The same is implemented in the state machine of AN75779 project. Hence it is possible to use a sensor having line size greater than buffer size.

3. The maximum resolution that can be supported depends on practical USB bandwidth and GPIF II bandwidth. The bandwidth of the video is calculated as

Bandwidth = h-active * v-active * bits per pixel * fps.

Note that h-active is in number of pixels unit and v-active is in number of lines in the above calculation.

If a 32 bit GPIF II interface is used with the image sensor, then a maximum of 3.2Gbps(400MBps) can be supported by FX3. This is because the maximum supported GPIF II clock frequency is 100MHz. So, the bandwidth of the sensor should be less than this. Note that if you are using a 16 bit GPIF II interface with the sensor, then the maximum bandwidth further reduces to 1.6Gbps(200MBps).

Example : For a 64MP sensor if the number of bits per pixel is 16(2 Bytes), and if a 32 bit GPIF II interface is used, then

64MP*2*fps<=400MBps

You can find that from the above calculation, it may support a max of 3fps.

Using the above calculation, you can find the maximum fps for 48MP also depending on the parameters that you wish to use.

Please let me know if you have further queries on this.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna

View solution in original post

0 Likes
2 Replies
JayakrishnaT_76
Moderator
Moderator
Moderator
First question asked 1000 replies posted 750 replies posted

Hello,

Please find my comments below:

1. Your understanding that AN75779 project make use of DMA buffers of 16K is correct. But the number of DMA buffers allocated is actually 8. This is because for each GPIF socket, 4 buffers are allocated. So, the total DMA buffer size used is around 16K*8 = 128K.

2. It is possible to support sensors having row size greater than Buffer size. Please refer to the section 3.4 of AN75779 document. This shows the thread switching procedure that is implemented in the state machine. From this, when the state machine is sampling data into DMA buffers in a thread and the buffer becomes full, it will switch to the next thread and sample data into the DMA buffers allocated with that thread. The same is implemented in the state machine of AN75779 project. Hence it is possible to use a sensor having line size greater than buffer size.

3. The maximum resolution that can be supported depends on practical USB bandwidth and GPIF II bandwidth. The bandwidth of the video is calculated as

Bandwidth = h-active * v-active * bits per pixel * fps.

Note that h-active is in number of pixels unit and v-active is in number of lines in the above calculation.

If a 32 bit GPIF II interface is used with the image sensor, then a maximum of 3.2Gbps(400MBps) can be supported by FX3. This is because the maximum supported GPIF II clock frequency is 100MHz. So, the bandwidth of the sensor should be less than this. Note that if you are using a 16 bit GPIF II interface with the sensor, then the maximum bandwidth further reduces to 1.6Gbps(200MBps).

Example : For a 64MP sensor if the number of bits per pixel is 16(2 Bytes), and if a 32 bit GPIF II interface is used, then

64MP*2*fps<=400MBps

You can find that from the above calculation, it may support a max of 3fps.

Using the above calculation, you can find the maximum fps for 48MP also depending on the parameters that you wish to use.

Please let me know if you have further queries on this.

Best Regards,

Jayakrishna

Best Regards,
Jayakrishna
0 Likes

Hi JayakrishnaT_76 Moderator ,

Thanks very much for your detailed description. It means a lot to my current projects.

0 Likes