diff options
| author | JRPan <[email protected]> | 2023-05-10 14:57:09 -0400 |
|---|---|---|
| committer | GitHub <[email protected]> | 2023-05-10 14:57:09 -0400 |
| commit | 57aa5ab28bdcb59b32762bd746586824707282eb (patch) | |
| tree | c8ab3b5ed3660283b9f95ec63cb620331aac94a7 /configs/tested-cfgs/SM7_TITANV/gpgpusim.config | |
| parent | da6a16a990a007edb7a760a2eb5b9b48ccc06e4c (diff) | |
| parent | 948c0e1a0e379e37e60c83b9ab622217522aea86 (diff) | |
Merge branch 'dev' into fix_different_latencies_to_same_ex_unit
Diffstat (limited to 'configs/tested-cfgs/SM7_TITANV/gpgpusim.config')
| -rw-r--r-- | configs/tested-cfgs/SM7_TITANV/gpgpusim.config | 22 |
1 files changed, 12 insertions, 10 deletions
diff --git a/configs/tested-cfgs/SM7_TITANV/gpgpusim.config b/configs/tested-cfgs/SM7_TITANV/gpgpusim.config index 3fa51ee..af561de 100644 --- a/configs/tested-cfgs/SM7_TITANV/gpgpusim.config +++ b/configs/tested-cfgs/SM7_TITANV/gpgpusim.config @@ -18,6 +18,7 @@ -gpgpu_heap_size_limit 8388608 -gpgpu_runtime_sync_depth_limit 2 -gpgpu_runtime_pending_launch_count_limit 2048 +-gpgpu_max_concurrent_kernel 128 # Compute Capability -gpgpu_compute_capability_major 7 @@ -100,7 +101,7 @@ # Volta has four schedulers per core -gpgpu_num_sched_per_core 4 # Greedy then oldest scheduler --gpgpu_scheduler gto +-gpgpu_scheduler lrr ## In Volta, a warp scheduler can issue 1 inst per cycle -gpgpu_max_insn_issue_per_warp 1 -gpgpu_dual_issue_diff_exec_units 1 @@ -114,17 +115,21 @@ # For more info, see https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#shared-memory-7-x # disable this mode in case of multi kernels/apps execution -gpgpu_adaptive_cache_config 1 -# Volta unified cache has four banks +-gpgpu_shmem_option 0,8,16,32,64,96 +-gpgpu_unified_l1d_size 128 +# L1 cache configuration -gpgpu_l1_banks 4 --gpgpu_cache:dl1 S:1:128:256,L:L:s:N:L,A:256:8,16:0,32 +-gpgpu_cache:dl1 S:4:128:64,L:T:m:L:L,A:512:8,16:0,32 +-gpgpu_l1_cache_write_ratio 25 +-gpgpu_gmem_skip_L1D 0 +-gpgpu_l1_latency 20 +-gpgpu_flush_l1_cache 1 +-gpgpu_n_cluster_ejection_buffer_size 32 +# shared memory configuration -gpgpu_shmem_size 98304 -gpgpu_shmem_sizeDefault 98304 -gpgpu_shmem_per_block 65536 --gpgpu_gmem_skip_L1D 0 --gpgpu_n_cluster_ejection_buffer_size 32 --gpgpu_l1_latency 20 -gpgpu_smem_latency 20 --gpgpu_flush_l1_cache 1 # 32 sets, each 128 bytes 24-way for each memory sub partition (96 KB per memory sub partition). This gives us 4.5MB L2 cache -gpgpu_cache:dl2 S:32:128:24,L:B:m:L:P,A:192:4,32:0,32 @@ -196,9 +201,6 @@ -enable_ptx_file_line_stats 1 -visualizer_enabled 0 -# power model configs, disable it untill we create a real energy model for Volta --power_simulation_enabled 0 - # tracing functionality #-trace_enabled 1 #-trace_components WARP_SCHEDULER,SCOREBOARD |