+ Post New Thread

Results 1 to 5 of 5

- 18th March 2017, 10:33 #1

- Join Date
- Mar 2017
- Posts
- 2
- Helped
- 0 / 0
- Points
- 21
- Level
- 1

## Matrix Multiplication using DSP48 slice in 7 series xilinx FPGA

Dear All,

I have done matrix multiplication based on the logic provided at the link:

http://www.fpga4student.com/2016/11/...re-design.html.

It uses IO's & logic blocks. I want to try with DSP48 slice in 7 series Xilinx FPGA. Please suggest the solution.

Thanks in advance

- 18th March 2017, 10:33

- 18th March 2017, 12:05 #2

- Join Date
- Jun 2010
- Posts
- 6,016
- Helped
- 1770 / 1770
- Points
- 32,893
- Level
- 44

## Re: Matrix Multiplication using DSP48 slice in 7 series xilinx FPGA

This code uses inferred multipliers so should just work in a 7 series.

- 18th March 2017, 12:05

- 19th March 2017, 09:09 #3

- Join Date
- Mar 2017
- Posts
- 2
- Helped
- 0 / 0
- Points
- 21
- Level
- 1

## Re: Matrix Multiplication using DSP48 slice in 7 series xilinx FPGA

The code mentioned in the lik is working fine, but i want to use DSP48 for processing of 500*500 matrix in less clock cycle. Please suggest suitable basic code with dsp48 slice

- 19th March 2017, 09:09

- 19th March 2017, 19:54 #4

- Join Date
- Sep 2013
- Location
- USA
- Posts
- 5,486
- Helped
- 1375 / 1375
- Points
- 24,537
- Level
- 38

## Re: Matrix Multiplication using DSP48 slice in 7 series xilinx FPGA

The use of DSP48 is already inferred. What you will need to do to improve the performance is to make the design compute the elements in a more parallel fashion. The DSP48s won't increase the parallelism of the design. You have to change the architecture of the design.

1 members found this post helpful.

- 26th March 2017, 05:03 #5

- Join Date
- Mar 2017
- Posts
- 1
- Helped
- 1 / 1
- Points
- 21
- Level
- 1

## Re: Matrix Multiplication using DSP48 slice in 7 series xilinx FPGA

I am the author of the post.

ads-ee is absolutely right. Just want to add this. If you want to use DSP48s for the design, just change the synthesis settings in order to let the synthesizer use DSP48 for synthesis process. If you want to increase the speed of the matrix multiplication, you have to parallelize the architecture or using pipelines.

1 members found this post helpful.

+ Post New Thread

Please login