Skip to content

Simplify offset and address calculation #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
yongwonshin wants to merge 1 commit into
base: develop
Choose a base branch
from
Open

Simplify offset and address calculation #3

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
29 changes: 29 additions & 0 deletions runtime/include/executor/ocl/kernel_utils.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -340,6 +340,20 @@ void _change_hab_habpim(__global uint8_t* __restrict__ pim_ctr, uint64_t offset
R_CMD(&pim_ctr[addr + offset]);
}

void _change_gemv_hab_habpim(__global uint8_t* __restrict__ pim_ctr, uint64_t offset
#ifdef EMULATOR
,
__global PimMemTracer* emulator_trace
#endif
)
{
uint64_t addr;
addr = addr_gen_(get_group_id(0), 0, 0, 0, 0x3fff, 0x0);
W_CMD_R_C(&pim_ctr[addr + offset], gemv_hab_to_hab_pim + offset);
R_CMD(&pim_ctr[addr + offset]);
B_CMD(1);
}

void _change_habpim_hab(__global uint8_t* __restrict__ pim_ctr, uint64_t offset
#ifdef EMULATOR
,
Expand Down Expand Up @@ -380,13 +394,27 @@ void _park_out(__global uint8_t* __restrict__ pim_ctr, int gidx, int num_ba, uin
W_CMD(&pim_ctr[addr + offset]);
}

uint64_t addr_gen_s(uint32_t chan, uint32_t rank, uint32_t bankgroup, uint32_t bank, uint32_t row, uint32_t col, uint32_t offset)
{
uint32_t offset_size = 1 << vega20_pbi.num_offset_bit;
uint32_t col_size = vega20_pbi.num_col / vega20_pbi.bl;

uint32_t offset_s = offset % offset_size;
uint32_t new_col = col + offset / offset_size;
uint32_t col_s = new_col % col_size;
uint32_t row_s = row + new_col / col_size;

return addr_gen_(chan, rank, bankgroup, bank, row_s, col_s) + offset_s;
}

#ifdef EMULATOR
#define park_in(a, b, c, d) _park_in(a, b, c, d, emulator_trace)
#define change_sb_hab(a, b) _change_sb_hab(a, b, emulator_trace)
#define program_crf(a, b, c, d) _program_crf(a, b, c, d, emulator_trace)
#define program_crf_mod(a, b, c, d) _program_crf_mod(a, b, c, d, emulator_trace)
#define program_srf(a, b, c) _program_srf(a, b, c, emulator_trace)
#define change_hab_habpim(a, b) _change_hab_habpim(a, b, emulator_trace)
#define change_gemv_hab_habpim(a, b) _change_gemv_hab_habpim(a, b, emulator_trace)
#define change_habpim_hab(a, b) _change_habpim_hab(a, b, emulator_trace)
#define change_hab_sb(a, b, c) _change_hab_sb(a, b, c, emulator_trace)
#define park_out(a, b, c, d) _park_out(a, b, c, d, emulator_trace)
Expand All @@ -398,6 +426,7 @@ void _park_out(__global uint8_t* __restrict__ pim_ctr, int gidx, int num_ba, uin
#define program_crf_mod(a, b, c, d) _program_crf_mod(a, b, c, d)
#define program_srf(a, b, c) _program_srf(a, b, c)
#define change_hab_habpim(a, b) _change_hab_habpim(a, b)
#define change_gemv_hab_habpim(a, b) _change_gemv_hab_habpim(a, b)
#define change_habpim_hab(a, b) _change_habpim_hab(a, b)
#define change_hab_sb(a, b, c) _change_hab_sb(a, b, c)
#define park_out(a, b, c, d) _park_out(a, b, c, d)
Expand Down
Loading