Spaces:
Runtime error
Runtime error
import gradio as gr | |
import matplotlib.pyplot as plt | |
import numpy as np | |
from descriptions import basic_texts, descriptions | |
def generate_data_parallel_groups(world_size, tensor_model_parallel_size, pipeline_model_parallel_size, context_parallel_size): | |
""" | |
Generate data parallel groups based on the provided parallelism parameters. | |
""" | |
assert world_size % (pipeline_model_parallel_size * tensor_model_parallel_size * context_parallel_size) == 0, "world_size must be divisible by the product of pipeline_model_parallel_size, tensor_model_parallel_size, and context_parallel_size" | |
data_parallel_group_ranks = [] | |
num_pipeline_model_parallel_groups = world_size // pipeline_model_parallel_size | |
for i in range(pipeline_model_parallel_size): | |
start_rank = i * num_pipeline_model_parallel_groups | |
end_rank = (i + 1) * num_pipeline_model_parallel_groups | |
for j in range(context_parallel_size * tensor_model_parallel_size): | |
ranks = range( | |
start_rank + j, end_rank, context_parallel_size * tensor_model_parallel_size | |
) | |
data_parallel_group_ranks.append(list(ranks)) | |
return data_parallel_group_ranks | |
def generate_context_data_parallel_groups(world_size, tensor_model_parallel_size, pipeline_model_parallel_size, context_parallel_size): | |
""" | |
Generate data parallel groups considering context parallelism. | |
""" | |
assert world_size % (pipeline_model_parallel_size * tensor_model_parallel_size * context_parallel_size) == 0, "world_size must be divisible by the product of pipeline_model_parallel_size, tensor_model_parallel_size, and context_parallel_size" | |
all_data_parallel_group_ranks_with_cp = [] | |
num_pipeline_model_parallel_groups = world_size // pipeline_model_parallel_size | |
for i in range(pipeline_model_parallel_size): | |
start_rank = i * num_pipeline_model_parallel_groups | |
end_rank = (i + 1) * num_pipeline_model_parallel_groups | |
for j in range(tensor_model_parallel_size): | |
ranks_with_cp = range(start_rank + j, end_rank, tensor_model_parallel_size) | |
all_data_parallel_group_ranks_with_cp.append(list(ranks_with_cp)) | |
return all_data_parallel_group_ranks_with_cp | |
def generate_tensor_model_parallel_groups(world_size, tensor_model_parallel_size): | |
""" | |
Generate model parallel groups based on tensor model parallel size. | |
""" | |
assert world_size % tensor_model_parallel_size == 0, "world_size must be divisible by tensor_model_parallel_size" | |
num_tensor_model_parallel_groups = world_size // tensor_model_parallel_size | |
tensor_model_parallel_group_ranks = [] | |
for i in range(num_tensor_model_parallel_groups): | |
ranks = range(i * tensor_model_parallel_size, (i + 1) * tensor_model_parallel_size) | |
tensor_model_parallel_group_ranks.append(list(ranks)) | |
return tensor_model_parallel_group_ranks | |
def generate_pipeline_parallel_groups(world_size, pipeline_model_parallel_size): | |
""" | |
Generate pipeline parallel groups based on pipeline model parallel size. | |
""" | |
assert world_size % pipeline_model_parallel_size == 0, "world_size must be divisible by pipeline_model_parallel_size" | |
num_pipeline_model_parallel_groups = world_size // pipeline_model_parallel_size | |
pipline_parallel_group_ranks = [] | |
for i in range(num_pipeline_model_parallel_groups): | |
ranks = range(i, world_size, num_pipeline_model_parallel_groups) | |
pipline_parallel_group_ranks.append(list(ranks)) | |
return pipline_parallel_group_ranks | |
def generate_context_parallel_groups(world_size, context_parallel_size, tensor_model_parallel_size, pipeline_model_parallel_size): | |
""" | |
Generate context parallel groups based on context parallel size, considering tensor and pipeline model parallel sizes. | |
""" | |
assert world_size % (context_parallel_size * tensor_model_parallel_size * pipeline_model_parallel_size) == 0, "world_size must be divisible by the product of context_parallel_size, tensor_model_parallel_size, and pipeline_model_parallel_size" | |
data_parallel_size = world_size // (tensor_model_parallel_size * pipeline_model_parallel_size * context_parallel_size) | |
context_parallel_group_ranks = [] | |
num_pipeline_model_parallel_groups: int = world_size // pipeline_model_parallel_size | |
for i in range(pipeline_model_parallel_size): | |
for j in range(data_parallel_size): | |
start_rank = ( | |
i * num_pipeline_model_parallel_groups | |
+ j * tensor_model_parallel_size * context_parallel_size | |
) | |
end_rank = ( | |
i * num_pipeline_model_parallel_groups | |
+ (j + 1) * tensor_model_parallel_size * context_parallel_size | |
) | |
for k in range(tensor_model_parallel_size): | |
ranks = range(start_rank + k, end_rank, tensor_model_parallel_size) | |
context_parallel_group_ranks.append(list(ranks)) | |
return context_parallel_group_ranks | |
def plot_parallel_groups(title="Parallel Groups", dp_groups=None, tp_groups=None, pp_groups=None, cp_groups=None): | |
# Initialize a figure | |
fig, ax = plt.subplots(figsize=(8, 6)) | |
# Define the spacing between blocks and their size | |
block_size = 700 # Size of the blocks in the scatter plot | |
spacing = 1.5 # Spacing multiplier between blocks | |
if cp_groups is None: | |
cp_offset_x = 0 | |
cp_offset_y = 0 | |
tp_offset_x = 0.2 | |
tp_offset_y = -0.2 | |
if tp_groups: | |
pp_offset_x = 0.4 | |
pp_offset_y = -0.4 | |
else: | |
pp_offset_x = 0.2 | |
pp_offset_y = -0.2 | |
else: | |
cp_offset_x = 0.2 | |
cp_offset_y = -0.2 | |
tp_offset_x = 0.4 | |
tp_offset_y = -0.4 | |
if tp_groups: | |
pp_offset_x = 0.6 | |
pp_offset_y = -0.6 | |
else: | |
pp_offset_x = 0.4 | |
pp_offset_y = -0.4 | |
# Adjust the grid layout to map GPU ranks from top-left to bottom-right | |
num_cols = 4 # Number of columns in the grid | |
x_positions = np.tile(np.arange(num_cols), num_cols) * spacing | |
y_positions = np.repeat(np.arange(num_cols), num_cols)[::-1] * spacing # Reverse to start from top | |
dp_colors = plt.cm.tab20(np.linspace(0, 1, len(dp_groups))) | |
# 使用tab20b提高颜色区分度 | |
if tp_groups is not None: | |
tp_colors = plt.cm.tab20b(np.linspace(0, 1, len(tp_groups))) | |
# 如果需要更多颜色,可以考虑结合使用tab20b和tab20c | |
if pp_groups is not None: | |
pp_colors = plt.cm.tab20c(np.linspace(0, 1, len(pp_groups))) | |
if cp_groups is not None: | |
cp_colors = plt.cm.tab20c(np.linspace(0, 1, len(cp_groups))) | |
if cp_groups is not None: | |
for group_idx, group in enumerate(cp_groups): | |
for rank in group: | |
x = x_positions[rank % (num_cols*num_cols)] + cp_offset_x | |
y = y_positions[rank % (num_cols*num_cols)] + cp_offset_y | |
ax.scatter(x, y, s=block_size, color=cp_colors[group_idx], edgecolor='black', zorder=5, marker='s') | |
ax.text(x, y, f'CP{rank}', ha='center', va='center', color='white', fontsize=8, zorder=6, fontweight='bold') | |
for group_idx, group in enumerate(dp_groups): | |
for rank in group: | |
x = x_positions[rank % (num_cols*num_cols)] | |
y = y_positions[rank % (num_cols*num_cols)] | |
ax.scatter(x, y, s=block_size, color=dp_colors[group_idx], edgecolor='black', zorder=5, marker='>') | |
ax.text(x, y, f'DP{rank}', ha='center', va='center', color='white', fontsize=8, zorder=6, fontweight='bold') | |
if tp_groups is not None: | |
for group_idx, group in enumerate(tp_groups): | |
for rank in group: | |
x = x_positions[rank % (num_cols*num_cols)] + tp_offset_x | |
y = y_positions[rank % (num_cols*num_cols)] + tp_offset_y | |
ax.scatter(x, y, s=block_size, color=tp_colors[group_idx], edgecolor='black', zorder=5, marker='p') | |
ax.text(x, y, f'TP{rank}', ha='center', va='center', color='white', fontsize=8, zorder=6, fontweight='bold') | |
if pp_groups is not None: | |
for group_idx, group in enumerate(pp_groups): | |
for rank in group: | |
x = x_positions[rank % (num_cols*num_cols)] + pp_offset_x | |
y = y_positions[rank % (num_cols*num_cols)] + pp_offset_y | |
ax.scatter(x, y, s=block_size, color=pp_colors[group_idx], edgecolor='black', zorder=5, marker='h') | |
ax.text(x, y, f'PP{rank}', ha='center', va='center', color='white', fontsize=8, zorder=6, fontweight='bold') | |
# Draw a separating line between Node0 and Node1 | |
mid_y_position = np.max(y_positions) / 2 | |
ax.axhline(y=mid_y_position, color='black', linestyle='-', linewidth=2, zorder=0) | |
# Add Node labels | |
ax.text(-spacing, max(y_positions)/4, 'Node1', verticalalignment='center', fontsize=12) | |
ax.text(-spacing, 3*max(y_positions)/4, 'Node0', verticalalignment='center', fontsize=12) | |
# Adjusting the appearance | |
ax.set_aspect('equal') # Keep the aspect ratio square | |
ax.axis('off') # Turn off the axis | |
plt.title(title, pad=30) | |
return fig | |
# Gradio interface setup | |
def create_interface(): | |
def update_plot(parallel_group_type, tensor_model_parallel_size, pipeline_model_parallel_size, context_parallel_size, unused_text): | |
world_size = 16 # Fixed world size for 2 machines with 8 GPUs each | |
description = descriptions.get(parallel_group_type, "Invalid parallel group type") | |
# Initialize groups to None | |
data_groups = tp_groups = pp_groups = cp_groups = None | |
if "CP" in parallel_group_type or parallel_group_type == 'Context Parallel': | |
cp_groups = generate_context_parallel_groups(world_size, context_parallel_size, tensor_model_parallel_size, pipeline_model_parallel_size) | |
if "DP" in parallel_group_type: | |
data_groups = generate_context_data_parallel_groups(world_size, tensor_model_parallel_size, pipeline_model_parallel_size, context_parallel_size) | |
else: | |
if "DP" in parallel_group_type or parallel_group_type == 'Data Parallel': | |
data_groups = generate_data_parallel_groups(world_size, tensor_model_parallel_size, pipeline_model_parallel_size, context_parallel_size) | |
if parallel_group_type in ['Tensor Model Parallel', 'DP+TP', 'DP+TP+PP', 'CP+DP+TP', 'CP+DP+TP+PP']: | |
tp_groups = generate_tensor_model_parallel_groups(world_size, tensor_model_parallel_size) | |
if parallel_group_type in ['Pipeline Parallel', 'DP+PP', 'DP+TP+PP', 'CP+DP+PP', 'CP+DP+TP+PP']: | |
pp_groups = generate_pipeline_parallel_groups(world_size, pipeline_model_parallel_size) | |
# Prepare text description for display | |
groups_list_str = "" | |
if data_groups: | |
groups_list_str += "Data Parallel Groups:\n" | |
groups_list_str += "\n".join([f"Data Group {idx + 1}: {group}" for idx, group in enumerate(data_groups)]) | |
groups_list_str += "\n--------------------------------------\n" | |
if tp_groups: | |
groups_list_str += "Tensor Model Parallel Groups:\n" | |
groups_list_str += "\n".join([f"Tensor Group {idx + 1}: {group}" for idx, group in enumerate(tp_groups)]) | |
groups_list_str += "\n--------------------------------------\n" | |
if pp_groups: | |
groups_list_str += "Pipeline Model Parallel Groups:\n" | |
groups_list_str += "\n".join([f"Pipeline Group {idx + 1}: {group}" for idx, group in enumerate(pp_groups)]) | |
groups_list_str += "\n--------------------------------------\n" | |
if cp_groups: | |
groups_list_str += "Context Parallel Groups:\n" | |
groups_list_str += "\n".join([f"Context Group {idx + 1}: {group}" for idx, group in enumerate(cp_groups)]) | |
groups_list_str += "\n--------------------------------------\n" | |
text_to_display = f"==========Parallel Groups Display==========\n\n{groups_list_str}\n\n{description}" | |
# Generate the figure with the parallel groups | |
fig = plot_parallel_groups(f"{parallel_group_type} Groups", data_groups if data_groups else [], tp_groups=tp_groups, pp_groups=pp_groups, cp_groups=cp_groups) | |
return fig, text_to_display | |
iface = gr.Interface( | |
fn=update_plot, | |
inputs=[ | |
gr.Dropdown(['Data Parallel', 'Tensor Model Parallel', 'Pipeline Parallel', 'Context Parallel', | |
'DP+TP', 'DP+PP', 'DP+TP+PP', | |
'CP+DP', 'CP+DP+TP', 'CP+DP+PP', 'CP+DP+TP+PP'], label="Parallel Group Type"), | |
gr.Slider(1, 8, step=1, label="Tensor Model Parallel Size"), | |
gr.Slider(1, 8, step=1, label="Pipeline Model Parallel Size"), | |
gr.Slider(1, 8, step=1, label="Context Parallel Size"), | |
gr.Textbox(basic_texts, interactive=False) | |
], | |
outputs=[ | |
"plot", | |
"text" | |
], | |
title="Megatron-LM Parallel Group Visualization", | |
description="Select parallel sizes and types to visualize different parallel groups with distinct colors. This includes combinations of Data Parallel (DP), Tensor Model Parallel (TP), Pipeline Parallel (PP), and Context Parallel (CP). Note that the size of data parallelism is automatically calculated based on world_size (which is stable at 16 here) as well as tensor_model_parallel_size, pipeline_model_parallel_size, and context_parallel_size.", | |
live=True | |
) | |
return iface | |
# Create and launch the interface | |
iface = create_interface() | |
iface.launch(share=True) | |