Commit a815c9a0 authored by Sakthivel Velumani's avatar Sakthivel Velumani
Browse files

Floating point twiddle factors 

Use floating point twiddle for first 8 butterflies in 16 point DFT.
Later stages like 256 point DFT use floating point twiddle only in first
4 butterflies. Doing so reduces EVM in first 4 frequency bins by 6 times
for 1024 DFT.
parent 15a64e2c

openair1/PHY/TOOLS/dfts_load.c

+0 −1
Changes for openair1/PHY/TOOLS/dfts_load.c: 0 added lines, 1 removed line.
Original line number Diff line number Diff line
@@ -101,7 +101,6 @@ uint32_t IDFT_SCALING_3072[2][4] = {{1, 1, 1, 3}, {1, 1, 1, 3}}; 


uint32_t DFT_SCALING_16[][1] = {{2}};

uint32_t DFT_SCALING_64[5][2] = {{3, 0}, {2, 1}, {1, 2}, {1, 2}, {1, 2}};

uint32_t DFT_SCALING_128[5][3]  = {{4,0,0},{3,1,0},{2,2,0},{1,3,0},{0,4,0}};

uint32_t DFT_SCALING_256[5][3]  = {{4,0,0},{3,1,0},{2,2,0},{1,3,0},{0,4,0}};

int32_t DFT_SCALING_512_THRES[7] = {53, 57, 59, 63, 65, 69, 100};

int32_t DFT_SCALING_1024_THRES[5] = {49,55,63,69,100};

openair1/PHY/TOOLS/oai_dfts.c

+533 −87

File changed.

Preview size limit exceeded, changes collapsed.