pulp-platform / tech_cells_generic

Technology dependent cells instantiated in the design for generic process (simulation, FPGA)
Other
32 stars 30 forks source link

Add generic SRAM #4

Closed WRoenninger closed 4 years ago

WRoenninger commented 4 years ago
zarubaf commented 4 years ago

Ping @accuminium

WRoenninger commented 4 years ago

Moved the latching of the read output to the same process as the write of the internal memory array. From the Xilinx XPM documentation:

A read operation is implicitly performed to address addr[a|b] combinatorially. The data output
is registered each clk[a|b] cycle that en[a|b] is asserted.

The latch of the read pipeline now takes place before a value gets written to the internal memory array.

andreaskurth commented 4 years ago

I think this makes sense, because it eliminates the combinatorial path from written to read data also in the special case that both addresses are equal.

zarubaf commented 4 years ago

Thanks for following up on this! LGTM

zarubaf commented 4 years ago

Can you also change the README and CHANGELOG, please? Then I am going to merge and release.

andreaskurth commented 4 years ago

@zarubaf Do not forget to rebase before merging to resolve all the fixups @WRoenninger has nicely committed.

zarubaf commented 4 years ago

Merged! Thanks