diff options
| -rw-r--r-- | src/abstract_hardware_model.h | 16 | ||||
| -rw-r--r-- | src/gpgpu-sim/shader.cc | 8 |
2 files changed, 19 insertions, 5 deletions
diff --git a/src/abstract_hardware_model.h b/src/abstract_hardware_model.h index 4d2bb4c..ba32358 100644 --- a/src/abstract_hardware_model.h +++ b/src/abstract_hardware_model.h @@ -1320,12 +1320,24 @@ class register_set { // for sub core model we need to figure which reg_id has the ready warp // this function should only be called if has_ready() was true assert(has_ready()); + warp_inst_t **ready; + ready = NULL; + unsigned reg_id; for (unsigned i = 0; i < regs.size(); i++) { if (not regs[i]->empty()) { - return i; + if (ready and (*ready)->get_uid() < regs[i]->get_uid()) { + // ready is oldest + } else { + ready = ®s[i]; + reg_id = i; + } } } - abort(); + return reg_id; + } + unsigned get_schd_id(unsigned reg_id) { + assert(not regs[reg_id]->empty()); + return regs[reg_id]->get_schd_id(); } void move_in(warp_inst_t *&src) { warp_inst_t **free = get_free(); diff --git a/src/gpgpu-sim/shader.cc b/src/gpgpu-sim/shader.cc index 895a2ef..5c27b9b 100644 --- a/src/gpgpu-sim/shader.cc +++ b/src/gpgpu-sim/shader.cc @@ -3976,19 +3976,21 @@ void opndcoll_rfu_t::allocate_cu(unsigned port_num) { bool allocated = false; unsigned cuLowerBound = 0; unsigned cuUpperBound = cu_set.size(); + unsigned schd_id; if(sub_core_model) { // Sub core model only allocates on the subset of CUs assigned to the scheduler that issued unsigned reg_id = (*inp.m_in[i]).get_ready_reg_id(); + schd_id = (*inp.m_in[i]).get_schd_id(reg_id); assert(cu_set.size() % m_num_warp_scheds == 0); unsigned cusPerSched = cu_set.size() / m_num_warp_scheds; - cuLowerBound = reg_id * cusPerSched; + cuLowerBound = schd_id * cusPerSched; cuUpperBound = cuLowerBound + cusPerSched; - std::cout << "reg_id: " << reg_id << " cusPerSched: " << cusPerSched << " lowerBound: " << cuLowerBound << std::endl; + std::cout << "reg_id: " << reg_id << " schd_id: " << schd_id << " cusPerSched: " << cusPerSched << " lowerBound: " << cuLowerBound << std::endl; assert(0 <= cuLowerBound && cuUpperBound <= cu_set.size()); } for (unsigned k = cuLowerBound; k < cuUpperBound; k++) { if (cu_set[k].is_free()) { - std::cout << "Allocated on cu: " << k << std::endl; + std::cout << "Allocated schd_id: " << schd_id << " on cu: " << k << std::endl; collector_unit_t *cu = &cu_set[k]; allocated = cu->allocate(inp.m_in[i], inp.m_out[i]); m_arbiter.add_read_requests(cu); |