riscv
/
OpenFPGA
mirror of https://github.com/lnis-uofu/OpenFPGA.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[engine] fixing a bug in fabric bitstream encoding

This commit is contained in:
tangxifan 2022-09-01 16:28:17 -07:00
parent fec4a4a29e
commit ea6f609181
1 changed files with 5 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
openfpga/src/fpga_bitstream/fabric_bitstream.h
View File

@ -224,15 +224,12 @@ class FabricBitstream {
* - bit-x number: which encodes the 'x' bits into a number. For example,
* 101x1 -> 00010 -> 2
*
* TODO: There is a limitation here, when the length of address vector is more than 64,
* A size_t number overflows (cannot represent any binary number > 64 bit).
* Such thing can entirely happen even in a medium sized FPGA.
* A solution can be use multiple size_t to fit. But clearly, we should not use vector in vector, which causes large memory overhead!
* Note that when the length of address vector is more than 64, we use multiple 64-bit data to store the encoded values
*/
vtr::vector<FabricBitId, size_t> bit_address_1bits_;
vtr::vector<FabricBitId, size_t> bit_address_xbits_;
vtr::vector<FabricBitId, size_t> bit_wl_address_1bits_;
vtr::vector<FabricBitId, size_t> bit_wl_address_xbits_;
vtr::vector<FabricBitId, std::vector<uint64_t>> bit_address_1bits_;
vtr::vector<FabricBitId, std::vector<uint64_t>> bit_address_xbits_;
vtr::vector<FabricBitId, std::vector<uint64_t>> bit_wl_address_1bits_;
vtr::vector<FabricBitId, std::vector<uint64_t>> bit_wl_address_xbits_;
/* Data input (Din) bits: this is designed for memory decoders */
vtr::vector<FabricBitId, char> bit_dins_;