The good alternative was to use the AXI Data Mover. – The transfer commands are delivered by AXI4 Stream. – The status of transfers are delivered back by. The AXI Datamover is a key Interconnect Infrastructure IP which enables high throughput transfer of data between AXI4 memory mapped domain to AXI4- Stream. For you, you are probably looking at AXI Datamover or AXI Central DMA. ” Xilinx provides the AXI Virtual FIFO Controller core to use external.
|Published (Last):||2 February 2007|
|PDF File Size:||9.2 Mb|
|ePub File Size:||17.81 Mb|
|Price:||Free* [*Free Regsitration Required]|
When you state that you are using SDK to read memory location 0x, you do not say how. I did tried the Validation, and even it could synthesize, just with the warning about the different width.
It also seems like the rest of the signals are correct. I will need to do that for a maximum ofclock cycles ms. Please upgrade to a Xilinx. As a result, I created a bit value that is a concatenation of the bit along with a bit count which is the value I am sending to the datamover DMA over AXI-Stream. However, I have that wait state going up to 70 dqtamover cycles before data is sent and still same behaviour.
AXI Datamover Design Problem – Community Forums
Actually I do disable cache in my code before reading the memory location simply by including the following:. Afterwards, and since I am sending bit word at a time, I will include the logic to keep on incrementing the SADDR every time I receive a new data word to send.
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type. It’s a bit strange that the second transfer cannot be executed, since a the FSM goes through through the same steps in the second iteration as it did in the first one, so the protocol is being followed, and b as far as I understand, there’s no need to do any kind of inter-transfer [re]initialization of the DataMover block or is there?
As I am connecting a normal fifo to this input, which is not master axi, then I created an AXI fifo, with the correct width, validate, erase and finally with the correct width connect my normal fifo in the place I had it before. I have a state machine running for the data that would send a bit data word every time a new value becomes available.
I datanover just curious about your experience. These commands will bypass cache and give you the actual values located in memory instead of the cached values. I am trying to create a design using the AXI datamover in a Zynq design using a zedboard yet I am really struggling. You have the same problem?
In addition, although I did try different addresses I am starting to wonder if the addresses I am choosing are being overwritten by something else although I highly doubt. From what it seems, the datamover is not accepting anymore data over the AXIS bus after a few clock cycles.
Embedded Processor System Design: Still working on it though. Thanks a lot for your timely and useful assistance. Its been almost two weeks since I have been trying to get this to work. Please upgrade to a Xilinx. I suspected that there might be something wrong with the command word I am sending but the logic analyzer data tells me otherwise from what I could tell.
Keep in mind that L1 and L2 cache is probably enabled when the cpu reads or writes 0x so you may only be interacting with cache. It’s the mechanism to propagate various parameters like data width. However, when a second write command is issued, the tready signal of the s2mm bus is deasserted, and never asserted again. It sounded like it would be sufficient for my purpose.
Currently I have the command word set for fixed address which I am doing until I get the design to work. I changed my HDL code for testing purposes.
ChromeFirefoxInternet Explorer 11Safari. I found out that I was also dwtamover the same configuration, but haven’t been able to test it because the Datamover Steam Data Width Auto is stuck at 32 even though I have a bus connected to it. Seems like the reading is getting up to a count of 5 and then not reading anymore data. All other trademarks are the property of their respective owners.
I’m not quite sure why that is happening. To the maximum extent permitted by applicable law: Xilinx products are not designed or intended to be fail-safe or for use in any application requiring fail-safe performance; you assume sole risk and liability for use of Xilinx products in Critical Applications: Revision History The following table shows the revision history for this document.
The is still a problem though.