adapt launch script for e-core and p-core arch (#1559)

Author: jingxu10

docs/tutorials/performance_tuning/launch_script.md

@@ -34,6 +34,7 @@ Launcher Common Arguments:
 | :-- | :--: | :--: | :-- |
 | `--ncores-per-instance` | int | 0 | Number of cores per instance |
 | `--nodes-list` | str | '' | Specify nodes list for multiple instances to run on, in format of list of single node ids "node_id,node_id,..." or list of node ranges "node_id-node_id,...". By default all nodes will be used. |
+| `--use-e-cores` | - | False | Use Efficient-Cores on the workloads or not. By default, only Performance-Cores are used. |
 | `--memory-allocator` | str | 'auto' | Choose which memory allocator to run the workloads with. Supported choices are ['auto', 'default', 'tcmalloc', 'jemalloc']. |
 | `--omp-runtime` | str | 'auto' | Choose which OpenMP runtime to run the workloads with. Supported choices are ['auto', 'default', 'intel']. |
 

intel_extension_for_pytorch/cpu/launch/cpu_info.py

@@ -8,61 +8,86 @@
 # # The following is the parsable format, which can be fed to other
 # # programs. Each different item in every column has an unique ID
 # # starting from zero.
-# # CPU,Core,Socket,Node
-# 0,0,0,0
-# 1,0,0,0
-# 2,1,0,0
-# 3,1,0,0
-# 4,2,1,1
-# 5,2,1,1
-# 6,3,1,1
-# 7,3,1,1
+# CPU NODE SOCKET CORE L1d:L1i:L2:L3 ONLINE    MAXMHZ   MINMHZ      MHZ
+#   0    0      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   1    0      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   2    0      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   3    0      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   4    1      1    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   5    1      1    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   6    1      1    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   7    1      1    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
 
-# 0,0,0,
-# 1,0,0,
-# 2,1,0,
-# 3,1,0,
-# 4,2,0,
-# 5,2,0,
-# 6,3,0,
-# 7,3,0,
+# CPU SOCKET CORE L1d:L1i:L2:L3 ONLINE    MAXMHZ   MINMHZ      MHZ
+#   0      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   1      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   2      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   3      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   4      0    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   5      0    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   6      0    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   7      0    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
 
-# 0,0,0,0
-# 1,1,0,0
-# 2,2,1,1
-# 3,3,1,1
-# 4,0,0,0
-# 5,1,0,0
-# 6,2,1,1
-# 7,3,1,1
+# CPU NODE SOCKET CORE L1d:L1i:L2:L3 ONLINE    MAXMHZ   MINMHZ      MHZ
+#   0    0      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   1    0      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   2    1      1    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   3    1      1    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   4    0      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   5    0      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   6    1      1    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   7    1      1    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
+
+# CPU NODE SOCKET CORE L1d:L1i:L2:L3 ONLINE    MAXMHZ   MINMHZ      MHZ
+#   0    0      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   1    0      0    0 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   2    0      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   3    0      0    1 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   4    0      0    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   5    0      0    2 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   6    0      0    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   7    0      0    3 0:0:0:0          yes 5000.0000 800.0000 2400.000
+#   8    0      0    4 0:0:0:0          yes 3800.0000 800.0000 2400.000
+#   9    0      0    5 0:0:0:0          yes 3800.0000 800.0000 2400.000
+#  10    0      0    6 0:0:0:0          yes 3800.0000 800.0000 2400.000
+#  11    0      0    7 0:0:0:0          yes 3800.0000 800.0000 2400.000
 
 class CoreInfo():
     '''
     Class to store core-specific information, including:
     - [int] CPU index
     - [int] Core index
     - [int] Numa node index
+    - [int] Socket index
     - [bool] is a physical core or not
+    - [float] maxmhz
+    - [bool] is a performance core
     '''
-    def __init__(self, lscpu_txt=''):
+    def __init__(self, lscpu_txt='', headers={}):
         self.cpu = -1
         self.core = -1
+        self.socket = -1
         self.node = -1
         self.is_physical_core = True
-        if lscpu_txt != '':
-            self.parse_raw(lscpu_txt)
+        self.maxmhz = 0
+        self.is_p_core = True
+        if lscpu_txt != '' and len(headers) > 0:
+            self.parse_raw(lscpu_txt, headers)
 
-    def parse_raw(self, lscpu_txt):
-        cols = lscpu_txt.split(',')
-        idx_col_node = 3
-        if cols[idx_col_node] == '':
-            idx_col_node = 2
-        self.cpu = int(cols[0])
-        self.core = int(cols[1])
-        self.node = int(cols[idx_col_node])
+    def parse_raw(self, cols, headers):
+        self.cpu = int(cols[headers['cpu']])
+        self.core = int(cols[headers['core']])
+        if 'node' in headers:
+            self.node = int(cols[headers['node']])
+            self.socket = int(cols[headers['socket']])
+        else:
+            self.node = int(cols[headers['socket']])
+            self.socket = int(cols[headers['socket']])
+        if 'maxmhz' in headers:
+            self.maxmhz = float(cols[headers['maxmhz']])
 
     def __str__(self):
-        return f'{self.cpu}|{self.core}|{self.node}|{self.is_physical_core}'
+        return f'{self.cpu}\t{self.core}\t{self.socket}\t{self.node}\t{self.is_physical_core}\t{self.maxmhz}\t{self.is_p_core}'
 
 class CPUPool(list):
     '''
@@ -109,8 +134,8 @@ def __init__(self, logger=None, lscpu_txt=''):
             '''
             Retrieve CPU information from lscpu.
             '''
-            if lscpu_txt == '':
-                args = ['lscpu', '--parse=CPU,Core,Socket,Node']
+            if lscpu_txt.strip() == '':
+                args = ['lscpu', '--all', '--extended']
                 env_lang = os.getenv('LANG', 'UNSET')
                 os.environ['LANG'] = 'C'
                 lscpu_info = subprocess.check_output(args, env=os.environ, universal_newlines=True)
@@ -125,27 +150,49 @@ def __init__(self, logger=None, lscpu_txt=''):
             Filter out lines that are really useful.
             '''
             lscpu_info = lscpu_info.strip().split('\n')
+            headers = {}
+            num_cols = 0
             for line in lscpu_info:
-                line = line.strip()
-                if re.match('^([\d]+,[\d]+,[\d]+,[\d]?)', line):
-                    self.pool_all.append(CoreInfo(line))
+                line = re.sub(' +', ' ', line.lower().strip())
+                if 'cpu' in line and 'socket' in line and 'core' in line:
+                    t = line.split(' ')
+                    num_cols = len(t)
+                    for i in range(num_cols):
+                        if t[i] in ['cpu', 'core', 'socket', 'node', 'maxmhz']:
+                            headers[t[i]] = i
+                else:
+                    t = line.split(' ')
+                    if len(t) == num_cols and t[headers['cpu']].isdigit() and t[headers['core']].isdigit() and t[headers['socket']].isdigit():
+                        self.pool_all.append(CoreInfo(t, headers))
             assert len(self.pool_all) > 0, 'cpuinfo is empty'
 
-        '''
-        Loop through all cores and determine is_physical_core for each of them.
-        '''
-        phy_cores = [c.core for c in self.pool_all]
-        phy_cores_unique = set(phy_cores)
-        if len(phy_cores) // len(phy_cores_unique) > 1:
-            core_cur = -1
-            self.pool_all.sort(key=lambda x: (x.core, x.cpu))
-            for c in self.pool_all:
-                if core_cur != c.core:
-                    core_cur = c.core
-                else:
-                    c.is_physical_core = False
+        # Determine logical cores
+        core_cur = -1
+        self.pool_all.sort(key=lambda x: (x.core, x.cpu))
+        for c in self.pool_all:
+            if core_cur != c.core:
+                core_cur = c.core
+            else:
+                c.is_physical_core = False
         self.pool_all.sort(key=lambda x: x.cpu)
 
+        # Determine e cores
+        maxmhzs = list(set([c.maxmhz for c in self.pool_all]))
+        maxmhzs.sort()
+        mmaxmhzs = max(maxmhzs)
+        if mmaxmhzs > 0:
+            maxmhzs_norm = [f/mmaxmhzs for f in maxmhzs]
+            separator_idx = -1
+            for i in range(1, len(maxmhzs_norm)):
+                if maxmhzs_norm[i] - maxmhzs_norm[i-1] >= 0.15:
+                    separator_idx = i
+                    break
+            if separator_idx > -1:
+                e_core_mhzs = maxmhzs[:separator_idx]
+                for c in self.pool_all:
+                    if c.maxmhz in e_core_mhzs:
+                        c.is_p_core = False
+
     def verbose(self, level, msg):
         if self.logger:
             logging_fn = {
@@ -162,6 +209,7 @@ def verbose(self, level, msg):
     - ninstances [int]: Number of instances. Should be a non negative integer, 0 by default. When it is 0, it will be set according to usage scenarios automatically in the function.
     - ncores_per_instance [int]: Number of cores per instance. Should be a non negative integer, 0 by default. When it is 0, it will be set according to usage scenarios automatically in the function.
     - use_logical_cores [bool]: Use logical cores on the workloads or not, False by default. When set to False, only physical cores are used.
+    - use_e_cores [bool]: Use Efficient-Cores, False by default. When set to False, only Performance-Cores are used.
     - skip_cross_node_cores [bool]: Allow instances to be executed on cores across NUMA nodes, False by default.
     - nodes_list [list]: A list containing all node ids that the execution is expected to be running on.
     - cores_list [list]: A list containing all cpu ids that the execution is expected to be running on.
@@ -172,16 +220,19 @@ def gen_pools_ondemand(
             ninstances=0,
             ncores_per_instance=0,
             use_logical_cores=False,
+            use_e_cores=False,
             skip_cross_node_cores=False,
             nodes_list=[],
             cores_list=[],
             return_mode='auto'):
         # Generate an aggregated CPU pool
         if len(cores_list) > 0:
             cores_available = [c.cpu for c in self.pool_all]
-            assert set(cores_list).issubset(set(cores_available)), f'Designated cores list contains invalid cores.'
+            assert set(cores_list).issubset(set(cores_available)), f'Designated cores list {cores_list} contains invalid cores.'
             if use_logical_cores:
                 self.verbose('warning', 'Argument --use-logical-cores won\'t take effect when --cores-list is set.')
+            if use_e_cores:
+                self.verbose('warning', 'Argument --use-e-cores won\'t take effect when --cores-list is set.')
             pool = [c for c in self.pool_all if c.cpu in cores_list]
             nodes = list(set([c.node for c in pool]))
             ncores_per_node = -1
@@ -197,12 +248,17 @@ def gen_pools_ondemand(
         else:
             if len(nodes_list) > 0:
                 nodes_available = set([c.node for c in self.pool_all])
-                assert set(nodes_list).issubset(nodes_available), f'Designated nodes list contains invalid nodes.'
+                assert set(nodes_list).issubset(nodes_available), f'Designated nodes list {nodes_list} contains invalid nodes out from {nodes_available}.'
                 pool = [c for c in self.pool_all if c.node in nodes_list]
             else:
                 pool = self.pool_all
             if not use_logical_cores:
                 pool = [c for c in pool if c.is_physical_core]
+            if not use_e_cores:
+                pool = [c for c in pool if c.is_p_core]
+                e_cores = [c.cpu for c in pool if not c.is_p_core]
+                if len(e_cores) > 0:
+                    self.verbose('warning', f'Efficient-Cores are detected ({e_cores}). Disabled for performance consideration. You can enable them with argument --use-e-cores.')
 
         # Determine ninstances and ncores_per_instance for grouping
         assert ncores_per_instance >= 0, 'Argument --ncores-per-instance cannot be a negative value.'
@@ -256,8 +312,10 @@ def gen_pools_ondemand(
             self.pools_ondemand.append(pool_local)
 
 if __name__ == "__main__":
-    pools = CPUPoolList()
-    pools.gen_pools_ondemand(use_logical_cores=False, nodes_list=[0,1], return_mode='auto', ninstances=3, ncores_per_instance=0, skip_cross_node_cores=False)
+    lscpu_txt = '''
+'''
+    pools = CPUPoolList(lscpu_txt = lscpu_txt)
+    pools.gen_pools_ondemand(use_logical_cores=False, return_mode='auto', ninstances=3, ncores_per_instance=0, use_e_cores=True, skip_cross_node_cores=False)
     print(f'capacity pool_auto:  {pools.pool_all.get_pool_txt(return_mode="auto")}')
     print(f'capacity pool_list:  {pools.pool_all.get_pool_txt(return_mode="list")}')
     print(f'capacity pool_range: {pools.pool_all.get_pool_txt(return_mode="range")}')

intel_extension_for_pytorch/cpu/launch/launcher_base.py

@@ -39,6 +39,13 @@ def add_common_params(self, parser):
             type=str,
             help='Specify nodes list for multiple instances to run on, in format of list of single node ids "node_id,node_id,..." or list of node ranges "node_id-node_id,...". By default all nodes will be used.',
         )
+        group.add_argument(
+            '--use-e-cores',
+            '--use_e_cores',
+            action='store_true',
+            default=False,
+            help='Use Efficient-Cores on the workloads or not. By default, only Performance-Cores are used.',
+        )
         group.add_argument(
             '--memory-allocator',
             '--memory_allocator',

intel_extension_for_pytorch/cpu/launch/launcher_distributed.py

@@ -162,6 +162,7 @@ def launch(self, args):
                 ninstances=args.nprocs_per_node,
                 ncores_per_instance=ncores_per_instance,
                 use_logical_cores=True,
+                use_e_cores=args.use_e_cores,
                 nodes_list=nodes_list
                 )
 

intel_extension_for_pytorch/cpu/launch/launcher_multi_instances.py

@@ -159,6 +159,7 @@ def launch(self, args):
                 ninstances = args.ninstances,
                 ncores_per_instance = args.ncores_per_instance,
                 use_logical_cores = args.use_logical_cores,
+                use_e_cores = args.use_e_cores,
                 skip_cross_node_cores = args.skip_cross_node_cores,
                 nodes_list = nodes_list,
                 cores_list = cores_list