Merge remote-tracking branch 'origin/main' into ol_integration

.readthedocs.yaml

@@ -0,0 +1,8 @@
+version: 2
+conda:
+  environment: environment_old.yml
+build:
+  os: "ubuntu-20.04"
+  tools:
+    python: "mambaforge-22.9"
+

acme/acme/requirements.txt

@@ -1,30 +1,30 @@
-absl-py
-jaxlib==0.3.5
-bsuite
-tensorflow_probability==0.15.0
 dm-env
-jax==0.3.6
-optax
-rlds
-dm-control==0.0.364896371
-atari-py
-gym
-pillow
-gym[atari]
-dm-haiku
-dm-sonnet
+tensorflow-estimator==2.8.0
+dm-tree
 dm-launchpad==0.5.2
-pygame==2.1.0
 typing-extensions
-tensorflow-datasets==4.5.2
-chex
-rlax
+gym
+tensorflow_probability==0.15.0
 tensorflow==2.8.0
-tensorflow-estimator==2.8.0
-dm-tree
-flax
+dm-haiku
+bsuite
 keras==2.8.0
+optax
+absl-py
+flax
 numpy==1.22.4
-trfl
+tensorflow-datasets==4.5.2
+rlds
+jaxlib==0.3.5
 dm-reverb==0.7.2
+chex
+trfl
+gym[atari]
+jax==0.3.6
+dm-sonnet
+dm-control==0.0.364896371
+pillow
+rlax
+atari-py
+pygame==2.1.0
 dm-reverb==0.7.2

acme/setup.py

@@ -59,7 +59,7 @@
 jax_requirements = [
     'chex',
     'jax==0.3.6',  # Update when TF2.9 is release.
-    'jaxlib==0.3.5',  # Update when TF2.9 is release.
+    'jaxlib==0.3.15',  # Update when TF2.9 is release.
     'dm-haiku',
     'flax',
     'optax',

arch_gym/envs/AstraSimEnv.py

@@ -115,7 +115,9 @@ def calculate_reward(self, observations):
 
     # give it one action: one set of parameters from json file
     def step(self, action_dict):
-
+        """
+        Step function for the environment
+        """
         # write the three config files
         # with open(self.network_config, "w") as outfile:
         #     outfile.write(json.dumps(action_dict['network'], indent=4))

arch_gym/envs/AstraSimWrapper.py

@@ -64,7 +64,9 @@ def reset(self) -> dm_env.TimeStep:
 
 
   def step(self, action: types.NestedArray) -> dm_env.TimeStep:
-    """Steps the environment."""
+    """
+    Steps the environment.
+    """
     if self._reset_next_step:
       return self.reset()
     if(self.env_wrapper_sel=='macme' or self.env_wrapper_sel=='macme_continuous'):
@@ -125,6 +127,9 @@ def __getattr__(self, name: str):
     return getattr(self._environment, name)
 
   def close(self):
+    """
+    Closes the environment.
+    """
     self._environment.close()
 
 

arch_gym/envs/RandomParameterEnv.py

@@ -32,6 +32,9 @@ def __init__(self, natural=False):
         self.reset()
 
     def step(self, action):
+        """
+        Step function for the environment
+        """
 
         self.stepN = self.stepN + 1
         self.steps = self.steps + 1

arch_gym/envs/custom_env.py

@@ -9,21 +9,26 @@
 class CustomEnv(gym.Env):
     def __init__(self, max_steps=10):
         super(CustomEnv, self).__init__()
-        self.observation_space = spaces.Dict({"energy": spaces.Box(0, 1, (1,)), "area": spaces.Box(0, 1, (1,)), "latency": spaces.Box(0, 1, (1,))})
+        self.observation_space = spaces.Dict({"energy": spaces.Box(0, 1, (1,)),
+                                               "area": spaces.Box(0, 1, (1,)), 
+                                               "latency": spaces.Box(0, 1, (1,))})
 
-        self.action_space = spaces.Dict({"num_cores": spaces.Discrete(15), "freq": spaces.Box(low = 0.5, high = 3, dtype = float),
-        "mem_type": spaces.Discrete(3), "mem_size": spaces.Discrete(65)})
-        # mem_type_list = ['DRAM', 'SRAM', 'Hybrid']
+        self.action_space = spaces.Dict(
+                            {"num_cores": spaces.Discrete(15), 
+                            "freq": spaces.Box(low = 0.5, high = 3, dtype = float),
+                            "mem_type": spaces.Discrete(3), # mem_type is one of 'DRAM', 'SRAM', 'Hybrid'
+                            "mem_size": spaces.Discrete(65)})
+
 
         self.max_steps = max_steps
         self.counter = 0
-        self.energy = 10
-        self.area = 15
-        self.latency = 12
+        self.energy = 0
+        self.area = 0
+        self.latency = 0
         self.initial_state = np.array([self.energy, self.area, self.latency])
         self.observation = None
         self.done = False
-
+        self.ideal = np.array([4, 2.0, 1, 32]) #ideal values for action space [num_cores, freq, mem_type, mem_size]
 
     def reset(self):
         return self.initial_state
@@ -42,31 +47,17 @@ def step(self, action):
             self.reset()
         else:
             self.counter += 1
-
+        # Compute the new state based on the action (random formulae for now)
         self.energy += num_cores*1 + freq*2 + mem_size*3
         self.area += num_cores*2 + freq*3 + mem_size*1
         self.latency += num_cores*3 + freq*3 + mem_size*1
 
         observation = np.array([self.energy, self.area, self.latency])
         ideal_values = np.array([4, 2.0, 1, 32])
-        print("custom env", observation)
-
         self.observation = observation
-        # print(action)
-        # print(ideal_values)
-        reward = -np.linalg.norm(action - ideal_values)
+        reward = -np.linalg.norm(action - self.ideal)
 
-        print("custom env rew", reward)
-
         return observation, reward, self.done, {}
 
     def render(self, mode='human'):
         print (f'Energy: {self.energy}, Area: {self.area}, Latency: {self.latency}')
-    # def main(self):
-
-    #     # observation = self.reset()
-    #     # print("The initial observation is {}".format(observation))
-    #     action = self.action_space.sample()
-    #     print("The taken action is {}".format(action))
-    #     obs, reward, done, info = self.step(action)
-    #     print("The outcome is {}".format(reward))

arch_gym/envs/simpleEnv.py

@@ -46,6 +46,9 @@ def reset(self):
     return self.state
 
   def step(self, action):
+    """
+    Step function for the environment
+    """
     if self.done:
       print("Episode Done!")
     elif (self.count == self.max_steps):

docs/build/html/.buildinfo

@@ -0,0 +1,4 @@
+# Sphinx build info version 1
+# This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
+config: 2204b31f9468ba8dbee3f93a98f6fbf5
+tags: 645f666f9bcd5a90fca523b33c5a78b7