EmbDev.net

Forum: FPGA, VHDL & Verilog Is this nonesense?


Is this nonesense?
von Julian M. (Company: Relevant Technologies Ltd) (geoffreym)

Rate this post
• ▲ useful
▼ not useful 
Hi

I have multiple clocked processes, all of which must access the same 
resources in my ip block. I often end up with something like this:
1
process(ps_reset_cir_b, reset_cir_cmd_rqst, reset_cir_cmd_out, 

2
        ps_init_cir_b,  init_cir_cmd_rqst,  init_cir_cmd_out,  init_cir_arg_out, 

3
        ps_rx_row_b,    rx_row_cmd_rqst,    rx_row_cmd_out,     

4
        ps_tx_max_b,    tx_max_cmd_rqst,    tx_max_cmd_out,                     tx_max_tready,

5
        ps_tx_cir_b,    tx_cir_cmd_rqst,    tx_cir_cmd_out,    tx_cir_arg_out,  tx_cir_tready,

6
        ps_wr_bram_b,   wr_bram_cmd_rqst,   wr_bram_cmd_out,   wr_bram_arg_out, 

7
        ps_rd_bram_b,   rd_bram_cmd_rqst,   rd_bram_cmd_out,   rd_bram_arg_out, rd_bram_tready,

8
        ps_wr_reg_b,    wr_reg_cmd_rqst,    wr_reg_cmd_out,    wr_reg_arg_out, 

9
        ps_rd_reg_b,    rd_reg_cmd_rqst,    rd_reg_cmd_out,    rd_reg_arg_out,  rd_reg_tready

10
  )

11
begin

12
  if (ps_reset_cir_b = '1') then 

13
    s_axis_cir_tready      <= '0';

14
    cmd_rqst               <= reset_cir_cmd_rqst;   

15
    cmd_out                <= reset_cir_cmd_out;    

16
    arg_out                <= (others=>'0');    

17
  elsif (ps_init_cir_b = '1') then  

18
    s_axis_cir_tready      <= '0';

19
    cmd_rqst               <= init_cir_cmd_rqst;    

20
    cmd_out                <= init_cir_cmd_out;     

21
    arg_out                <= init_cir_arg_out;     

22
  elsif (ps_rx_row_b = '1') then

23
    s_axis_cir_tready      <= '0';

24
    cmd_rqst               <= rx_row_cmd_rqst;      

25
    cmd_out                <= rx_row_cmd_out;       

26
    arg_out                <= (others=>'0');       

27
  elsif (ps_tx_max_b = '1') then

28
    s_axis_cir_tready      <= tx_max_tready;

29
    cmd_rqst               <= tx_max_cmd_rqst;      

30
    cmd_out                <= tx_max_cmd_out;       

31
    arg_out                <= (others=>'0');       

32
  elsif (ps_tx_cir_b = '1') then

33
    s_axis_cir_tready      <= tx_cir_tready;

34
    cmd_rqst               <= tx_cir_cmd_rqst;      

35
    cmd_out                <= tx_cir_cmd_out;       

36
    arg_out                <= tx_cir_arg_out;       

37
  elsif (ps_wr_bram_b = '1') then

38
    s_axis_cir_tready      <= '0';

39
    cmd_rqst               <= wr_bram_cmd_rqst;     

40
    cmd_out                <= wr_bram_cmd_out;      

41
    arg_out                <= wr_bram_arg_out;      

42
  elsif (ps_rd_bram_b = '1') then

43
    s_axis_cir_tready      <= rd_bram_tready;

44
    cmd_rqst               <= rd_bram_cmd_rqst;     

45
    cmd_out                <= rd_bram_cmd_out;      

46
    arg_out                <= rd_bram_arg_out;      

47
  elsif (ps_wr_reg_b = '1') then

48
    s_axis_cir_tready      <= '0';

49
    cmd_rqst               <= wr_reg_cmd_rqst;  

50
    cmd_out                <= wr_reg_cmd_out;   

51
    arg_out                <= wr_reg_arg_out;   

52
  elsif (ps_rd_reg_b = '1') then

53
    s_axis_cir_tready      <= rd_reg_tready;

54
    cmd_rqst               <= rd_reg_cmd_rqst;  

55
    cmd_out                <= rd_reg_cmd_out;   

56
    arg_out                <= rd_reg_arg_out;

57
  else

58
    s_axis_cir_tready      <= '0';

59
    cmd_rqst               <= (others=>'0');  

60
    cmd_out                <= (others=>'0');  

61
    arg_out                <= (others=>'0');

62
  end if;   

63
end process;


Does this look silly? The various bits indicate the process which is 
accessing the mux is active. When all processes are in the idle state, 
the final else is active. The various processes, in this case, are 
servicing requests from the host cpu. The signals on the left are 
interfaces, and each process must use some or all of these.

Best regards
Geoff

Report post Edit Move Thread sperren Anmeldepflicht aktivieren Anpinnen Delete topic Thread mit anderem zusammenführen Quote selected text Reply Reply with quote
Re: Is this nonesense?
von Lothar M. (Company: Titel) (lkmiller) (Moderator)

Rate this post
• ▲ useful
▼ not useful 
Julian M. wrote:
> I have multiple clocked processes
I can't see absolutely no clock there...

> Does this look silly?
No, ist just the way to write the multiplexer in a process. You could do 
it as a concurrent descritpion also.

But the most easy way would be: use VHDL-2008 and write process(all).

Then you could shorten it further on by using default values instead of 
the last "else":
1
process(all)

2
begin

3
  s_axis_cir_tready      <= '0';

4
  cmd_rqst               <= (others=>'0');  

5
  cmd_out                <= (others=>'0');  

6
  arg_out                <= (others=>'0');

7
  if (ps_reset_cir_b = '1') then 

8
    cmd_rqst               <= reset_cir_cmd_rqst;   

9
    cmd_out                <= reset_cir_cmd_out;    

10
  elsif (ps_init_cir_b = '1') then  

11
    cmd_rqst               <= init_cir_cmd_rqst;    

12
    cmd_out                <= init_cir_cmd_out;     

13
    arg_out                <= init_cir_arg_out;     

14
  elsif (ps_rx_row_b = '1') then

15
    cmd_rqst               <= rx_row_cmd_rqst;      

16
    cmd_out                <= rx_row_cmd_out;             

17
  elsif (ps_tx_max_b = '1') then

18
    s_axis_cir_tready      <= tx_max_tready;

19
    cmd_rqst               <= tx_max_cmd_rqst;      

20
    cmd_out                <= tx_max_cmd_out;       

21
  elsif (ps_tx_cir_b = '1') then

22
    s_axis_cir_tready      <= tx_cir_tready;

23
    cmd_rqst               <= tx_cir_cmd_rqst;      

24
    cmd_out                <= tx_cir_cmd_out;       

25
    arg_out                <= tx_cir_arg_out;       

26
  elsif (ps_wr_bram_b = '1') then

27
    cmd_rqst               <= wr_bram_cmd_rqst;     

28
    cmd_out                <= wr_bram_cmd_out;      

29
    arg_out                <= wr_bram_arg_out;      

30
  elsif (ps_rd_bram_b = '1') then

31
    s_axis_cir_tready      <= rd_bram_tready;

32
    cmd_rqst               <= rd_bram_cmd_rqst;     

33
    cmd_out                <= rd_bram_cmd_out;      

34
    arg_out                <= rd_bram_arg_out;      

35
  elsif (ps_wr_reg_b = '1') then

36
    cmd_rqst               <= wr_reg_cmd_rqst;  

37
    cmd_out                <= wr_reg_cmd_out;   

38
    arg_out                <= wr_reg_arg_out;   

39
  elsif (ps_rd_reg_b = '1') then

40
    s_axis_cir_tready      <= rd_reg_tready;

41
    cmd_rqst               <= rd_reg_cmd_rqst;  

42
    cmd_out                <= rd_reg_cmd_out;   

43
    arg_out                <= rd_reg_arg_out;

44
  end if;   

45
end process;

: Edited by Moderator
Report post Edit Delete Quote selected text Reply Reply with quote
Re: Is this nonesense?
von Julian M. (Company: Relevant Technologies Ltd) (geoffreym)

Rate this post
• ▲ useful
▼ not useful 
Many thanks, I work alone, and still make some godawful blunders, having 
been doing this for about two years. Great fun, and a big paradigm shift 
from software, a background many will no doubt spot, due to my 
parentheses around conditions in if- statements.

Report post Edit Delete Quote selected text Reply Reply with quote
Re: Is this nonesense?
von Julian M. (Company: Relevant Technologies Ltd) (geoffreym)

Rate this post
• ▲ useful
▼ not useful 
The clocked processes are using these resources, and of the usual sort. 
They are all state machines, which have an idle state in which they do 
nothing at all. So this is combinatorial logic in support of them. They 
are assumed to execute exclusively, and, as soon as they enter a state 
other than "???_IDLE" the bit is set.

Report post Edit Delete Quote selected text Reply Reply with quote
Watch this topic | Disable multi-page view
Please log in before posting. Registration is free and takes only a minute.
Existing account
Do you have a Google/GoogleMail account? No registration required!
Log in with Google account
No account? Register here.