[Simulator] Add high-fidelity CPU-based inference simulator (#33824)
Co-authored-by: zhouhaizhu.zhz <zhouhaizhu.zhz@alibaba-inc.com> Co-authored-by: LinSiyuan814 <linsiyuan.lsy@alibaba-inc.com> Co-authored-by: hzh0425 <hzh0425@apache.org>
This commit is contained in:
co-authored by
zhouhaizhu.zhz
LinSiyuan814
hzh0425
parent
a5f07b1241
commit
59799a3687
@@ -0,0 +1,21 @@
|
||||
{
|
||||
"platform": {
|
||||
"accelerator": {"name": "a100_sxm", "hbm_capacity_gb": 80},
|
||||
"disk_read_bandwidth_gb": 8,
|
||||
"disk_write_bandwidth_gb": 8,
|
||||
"memory_read_bandwidth_gb": 64,
|
||||
"memory_write_bandwidth_gb": 64,
|
||||
"num_device_per_node": 8
|
||||
},
|
||||
"predictor": {
|
||||
"name": "aiconfigurator",
|
||||
"database_mode": "SILICON"
|
||||
},
|
||||
"scheduler": {
|
||||
"tp_size": 1,
|
||||
"ep_size": 1,
|
||||
"dp_size": 1,
|
||||
"backend_name": "sglang",
|
||||
"backend_version": "0.5.9"
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,21 @@
|
||||
{
|
||||
"platform": {
|
||||
"accelerator": {"name": "a100_sxm", "hbm_capacity_gb": 80},
|
||||
"disk_read_bandwidth_gb": 8,
|
||||
"disk_write_bandwidth_gb": 8,
|
||||
"memory_read_bandwidth_gb": 64,
|
||||
"memory_write_bandwidth_gb": 64,
|
||||
"num_device_per_node": 8
|
||||
},
|
||||
"predictor": {
|
||||
"name": "aiconfigurator",
|
||||
"database_mode": "SOL"
|
||||
},
|
||||
"scheduler": {
|
||||
"tp_size": 1,
|
||||
"ep_size": 1,
|
||||
"dp_size": 1,
|
||||
"backend_name": "sglang",
|
||||
"backend_version": "0.5.9"
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,22 @@
|
||||
{
|
||||
"platform": {
|
||||
"accelerator": {"name": "a100_sxm", "hbm_capacity_gb": 80},
|
||||
"disk_read_bandwidth_gb": 8,
|
||||
"disk_write_bandwidth_gb": 8,
|
||||
"memory_read_bandwidth_gb": 64,
|
||||
"memory_write_bandwidth_gb": 64,
|
||||
"num_device_per_node": 8
|
||||
},
|
||||
"predictor": {
|
||||
"name": "ml",
|
||||
"database_path": "../assets/model.pkl",
|
||||
"latency_scale": 1.0
|
||||
},
|
||||
"scheduler": {
|
||||
"tp_size": 1,
|
||||
"ep_size": 1,
|
||||
"dp_size": 1,
|
||||
"backend_name": "sglang",
|
||||
"backend_version": "0.5.9"
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,23 @@
|
||||
{
|
||||
"platform": {
|
||||
"accelerator": {"name": "a100_sxm", "hbm_capacity_gb": 80},
|
||||
"disk_read_bandwidth_gb": 8,
|
||||
"disk_write_bandwidth_gb": 8,
|
||||
"memory_read_bandwidth_gb": 64,
|
||||
"memory_write_bandwidth_gb": 64,
|
||||
"num_device_per_node": 8
|
||||
},
|
||||
"predictor": {
|
||||
"name": "replay",
|
||||
"database_path": "../assets/replay_table.json",
|
||||
"miss_strategy": "knn",
|
||||
"miss_knn_k": 1
|
||||
},
|
||||
"scheduler": {
|
||||
"tp_size": 1,
|
||||
"ep_size": 1,
|
||||
"dp_size": 1,
|
||||
"backend_name": "sglang",
|
||||
"backend_version": "0.5.9"
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user