Hi Team,
I found access to L1 L2 RAM and to shared RAM were several times different.
In my understanding, according to the document, each memory access operation is
L1 L2 ram = SYSCLK1
Shared RAM = SYSCLK2
Although there was a difference of several times in the processing speed, it was simply assumed that it was not a difference of the clock speed of SYSCLK1/ SYSCLK2.
When measured, the data buffer (arrangement) was arranged in the L2 RAM and the shared RAM was distributed several times to several tens times.
It looks also depending on the processing contents.
It changes so far per data buffer arrangement. What is the reason?
ex.1) Multiplication
Multiplication of the data buffers arranged to the L2 RAM is 4.5 usec
Multiplication of the data buffers arranged to the Shared RAM is 159 USEC
About 35 times
ex.2) DSP_sp_biquad
Calling this function in the data buffers arranged to the L2 RAM is 60 usec
Calling this function in the data buffers arranged to the Shared RAM is 404 usec
About 6.7 times
Beat Regards,
Tom Liu