diff options
| author | Tor Aamodt <[email protected]> | 2010-10-24 09:56:14 -0800 |
|---|---|---|
| committer | Tor Aamodt <[email protected]> | 2010-10-24 09:56:14 -0800 |
| commit | 9ce235efd70782eba687da0f3163ff0f3f2840ef (patch) | |
| tree | e44be5be27f12a1b631e44d09be5f044501206d2 /src/gpgpu-sim/gpu-sim.cc | |
| parent | 6eee7514ea8b72fbecd761c50ccfd3394edf2307 (diff) | |
add back per shader icount tracking for visualizer
[git-p4: depot-paths = "//depot/gpgpu_sim_research/fermi/distribution/": change = 7910]
Diffstat (limited to 'src/gpgpu-sim/gpu-sim.cc')
| -rw-r--r-- | src/gpgpu-sim/gpu-sim.cc | 7 |
1 files changed, 2 insertions, 5 deletions
diff --git a/src/gpgpu-sim/gpu-sim.cc b/src/gpgpu-sim/gpu-sim.cc index e31ab67..8c1be87 100644 --- a/src/gpgpu-sim/gpu-sim.cc +++ b/src/gpgpu-sim/gpu-sim.cc @@ -336,7 +336,6 @@ gpgpu_sim::gpgpu_sim( const gpgpu_sim_config &config ) gpu_sim_insn = 0; gpu_tot_sim_insn = 0; gpu_tot_issued_cta = 0; - gpu_tot_completed_thread = 0; gpu_deadlock = false; m_cluster = new simt_core_cluster*[m_shader_config->n_simt_clusters]; @@ -489,7 +488,6 @@ unsigned int gpgpu_sim::run_gpu_sim() m_memory_stats->memlatstat_lat_pw(m_config.num_shader(),m_shader_config->n_thread_per_shader,m_shader_config->warp_size); gpu_tot_sim_cycle += gpu_sim_cycle; gpu_tot_sim_insn += gpu_sim_insn; - gpu_tot_completed_thread += m_shader_stats->get_gpu_completed_thread(); ptx_file_line_stats_write_file(); @@ -545,7 +543,6 @@ void gpgpu_sim::gpu_print_stat() const printf("gpu_tot_sim_cycle = %lld\n", gpu_tot_sim_cycle); printf("gpu_tot_sim_insn = %lld\n", gpu_tot_sim_insn); printf("gpu_tot_ipc = %12.4f\n", (float)gpu_tot_sim_insn / gpu_tot_sim_cycle); - printf("gpu_tot_completed_thread = %lld\n", gpu_tot_completed_thread); printf("gpu_tot_issued_cta = %lld\n", gpu_tot_issued_cta); // performance counter for stalls due to congestion. @@ -666,14 +663,14 @@ void shader_core_ctx::issue_block2core( kernel_info_t &kernel ) // now that we know which warps are used in this CTA, we can allocate // resources for use in CTA-wide barrier operations - allocate_barrier( free_cta_hw_id, warps ); + m_barriers.allocate_barrier(free_cta_hw_id,warps); // initialize the SIMT stacks and fetch hardware init_warps( free_cta_hw_id, start_thread, end_thread); m_n_active_cta++; shader_CTA_count_log(m_sid, 1); - printf("GPGPU-Sim uArch: core:%d, cta:%u initialized @(%lld,%lld)\n", m_sid, free_cta_hw_id, gpu_sim_cycle, gpu_tot_sim_cycle ); + printf("GPGPU-Sim uArch: core:%3d, cta:%2u initialized @(%lld,%lld)\n", m_sid, free_cta_hw_id, gpu_sim_cycle, gpu_tot_sim_cycle ); } /////////////////////////////////////////////////////////////////////////////////////////// |