I am getting slower that expected performance from the c674x-DSPLIB. I expect a multiply-add (dot-product) to take about 10 ns @ 300MHz, yet it is taking 120 ns. Do I have a setup problem?
This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
I am getting slower that expected performance from the c674x-DSPLIB. I expect a multiply-add (dot-product) to take about 10 ns @ 300MHz, yet it is taking 120 ns. Do I have a setup problem?
Hi William,
Thanks for your post.
It seems that the benchmarking for this DSPLIB kernel was done earlier in the CCSv3.3 environment on a cycle accurate simulator using C6000 CGT version 6.1.9 and I believe, the performance numbers wouldn't vary with other CCS versions too. After loading CCS, select Profile->Clock->Enable and this will allow the kernel demonstration apps to accurately display cycle counts.
May be, you could refer the c674x floating & fixed point benchmark performance numbers for vector dot product and kindly ensure the expected performance and evaluate on the same:
http://www.ti.com/lsds/ti/dsp/c6000_dsp/c674x/benchmarks.page#vector
http://www.ti.com/lsds/ti/dsp/c6000_dsp/c674x/benchmarks.page#vector2
Please compare your performance results with the above benchmark metrics.
Thanks & regards,
Sivaraj K
---------------------------------------------------------------------------------
Please click the Verify Answer button on this post if it answers your question.
---------------------------------------------------------------------------------