Intel® Quartus® Prime Software
Intel® Quartus® Prime Design Software, Design Entry, Synthesis, Simulation, Verification, Timing Analysis, System Design (Platform Designer, formerly Qsys)
16556 Discussions

Parameterizing ALTERA_FP_FUNCTIONS

Altera_Forum
Honored Contributor II
1,647 Views

Greetings 

I would like to use " Multiply Add " operation of ALTERA_FP_FUNCTIONS IP for an arria10 based project. The "Arria10 Native Floating-Point DSP Intel FPGA IP User Guide" that I referred, says;  

this mode performs single precision multiplication, followed by addition or subtraction operations and applies the following equations 

  1. out = (ay * az) + chainin 

  2. Out = (Ay * Az) - chainin
  3. out = (ay * az) - ax 

     

  4. out = (ay * az) + ay 

 

 

And I wish to perform the third operation for the implementation of an algorithm. But when I try to Parameterize the above said IP, didn't find any option to choose the specified operation. Please let me know how to parameterize the Multiply Add operation of ALTERA_FP_FUNCTIONS IP for executing equation out = (ay * az) - ax. 

thanking you 

anish 

0 Kudos
6 Replies
Altera_Forum
Honored Contributor II
457 Views

Hi Anish, 

 

 

--- Quote Start ---  

I wish to perform the third operation for the implementation of an algorithm. But when I try to Parameterize the above said IP, didn't find any option to choose the specified operation. Please let me know how to parameterize the Multiply Add operation of ALTERA_FP_FUNCTIONS IP for executing equation out = (ay * az) - ax. 

--- Quote End ---  

 

 

Ensure In “Parameters” window, 

1. DSP Template : Multiply Add 

2. View: Register Enables 

3. Click on Mux in front of input Ax & ensure Ax is connected as lower input to Mux. 

4. Click on Adder near to “Out” & ensure it is in Subtraction mode 

5. If you want Chain Out then click on “Chain Out mux”. 

 

Let me know if this has helped resolve the issue you are facing or if you need any further assistance. 

 

Best Regards 

Vikas Jathar  

Intel Customer Support – Engineering 

(Under Contract to Intel)
0 Kudos
Altera_Forum
Honored Contributor II
457 Views

Thank you for reply. 

After Your reply, I realized that I had confused with 'Multiply Add' operation of DSP template with the same of Arithmetic templaBte of floating point IPcores and referred wrong user guide. Heartfelt sorry for that. I attached the screen shot of the IPcore with its parameter editor window, which I intended about. I tried to know its operation through "floating point IP user guide", but unfortunately not much information available in it. So please let me know the mode of operation of the particular IP. In other words I just want to know whether it is possible to implement the operation Y = [A * B] - C with the " Multiply Add " operation of ALTERA_FP_FUNCTIONS IP available in the arithmetic template of basic function library. 

Thanking you 

Anish
0 Kudos
Altera_Forum
Honored Contributor II
457 Views

Hi Anish, 

 

 

--- Quote Start ---  

In other words I just want to know whether it is possible to implement the operation Y = [A * B] - C with the " Multiply Add " operation of ALTERA_FP_FUNCTIONS IP available in the arithmetic template of basic function library. 

--- Quote End ---  

 

 

No, it is not available. I set Goal to “Manually Specify DSP Registers” in Performance tab but it is not allowing to use “addition/subtraction operator” for subtraction, Check screenshot & refer floating point IP user guide for Performance Tab. 

 

Let me know if this has helped resolve the issue you are facing or if you need any further assistance. 

 

Best Regards 

Vikas Jathar  

Intel Customer Support – Engineering 

(Under Contract to Intel) 

 

https://www.alteraforum.com/forum/attachment.php?attachmentid=15103
0 Kudos
Altera_Forum
Honored Contributor II
457 Views

Thank you for your reply 

 

The screenshot what I got was entirely different from what you shared. I here by attach the same. Also I didn't have a performance tab for this IP option. I am using Quartus Prime 17 standard edition. 

 

Sir, I need one more clarification. Under the library "basic functions" in IP catalog, if we choose arithmetic option, ALTERA_MULT_ADD IP is available. But its operation is not detailed Floating point IP user guide. Whether is it possible to achieve Y = [A * B] - C operation using that IP? 

Thanking you 

Anish
0 Kudos
Altera_Forum
Honored Contributor II
457 Views

Hi Anish, 

 

 

--- Quote Start ---  

The screenshot what I got was entirely different from what you shared. I here by attach the same. Also I didn't have a performance tab for this IP option. I am using Quartus Prime 17 standard edition. 

--- Quote End ---  

 

 

Screenshot provided you seems to be from DSP Templates. 

 

 

--- Quote Start ---  

In other words I just want to know whether it is possible to implement the operation Y = [A * B] - C with the " Multiply Add " operation of ALTERA_FP_FUNCTIONS IP available in the arithmetic template of basic function library. 

--- Quote End ---  

 

 

Screenshot provided by me is from Arithmetic Templates, check again my previous post. 

 

 

--- Quote Start ---  

Under the library "basic functions" in IP catalog, if we choose arithmetic option, ALTERA_MULT_ADD IP is available. But its operation is not detailed Floating point IP user guide. 

--- Quote End ---  

 

 

Refer following link, 

https://www.altera.com/en_us/pdfs/literature/ug/ug_lpm_alt_mfug.pdf 

 

Let me know if this has helped resolve the issue. 

 

Best Regards 

Vikas Jathar  

 

Intel Customer Support – Engineering 

(Under Contract to Intel)
0 Kudos
Altera_Forum
Honored Contributor II
457 Views

Thank you for the reply  

Sir, may i know that both integer multiply add IP and floating point multiply add IP performs the same operation? because in integer IP it has more options like preadder, coefficients, pipeline configuration etc., which are not available in floating point IP editor.
0 Kudos
Reply