File size: 12,975 Bytes
b88c26d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 | #!/usr/bin/env python3
"""
BURT-IMMA Benchmark Script
License: BSL-1.1
Contact: jessica@collectivekitty.com
Benchmarks forward pass, backward pass, memory update, and routing
for different batch sizes and hidden dimensions.
"""
import argparse
import time
import sys
from typing import Dict, List, Tuple
try:
import torch
import torch.nn as nn
HAS_TORCH = True
except ImportError:
HAS_TORCH = False
print("WARNING: PyTorch not available. Benchmarks require PyTorch.")
import numpy as np
# ---------------------------------------------------------------------------
# Minimal PyTorch models for benchmarking
# ---------------------------------------------------------------------------
if HAS_TORCH:
class BenchSmoothLeaky(nn.Module):
def __init__(self, alpha: float = 0.01):
super().__init__()
self.alpha = alpha
def forward(self, x: torch.Tensor) -> torch.Tensor:
return self.alpha * x + (1.0 - self.alpha) * x * torch.sigmoid(x)
class BenchCIFGMemory(nn.Module):
def __init__(self, d_mem: int):
super().__init__()
self.d_mem = d_mem
self.C = nn.Parameter(torch.zeros(d_mem, d_mem), requires_grad=False)
def update(self, key: torch.Tensor, value: torch.Tensor,
forget_gate: float = 0.9) -> torch.Tensor:
candidate = torch.outer(key, value)
candidate = candidate / (candidate.norm() + 1e-8)
new_C = forget_gate * self.C + (1.0 - forget_gate) * candidate
self.C.data.copy_(new_C)
return new_C
def read(self, query: torch.Tensor) -> torch.Tensor:
return self.C @ query
class BenchGatesRouter(nn.Module):
def __init__(self, input_dim: int, num_experts: int, top_k: int = 1):
super().__init__()
self.num_experts = num_experts
self.top_k = top_k
self.gate = nn.Linear(input_dim, num_experts)
self.experts = nn.ModuleList([
nn.Linear(input_dim, input_dim) for _ in range(num_experts)
])
def forward(self, x: torch.Tensor) -> torch.Tensor:
# x: (batch, dim)
scores = self.gate(x) # (batch, num_experts)
probs = torch.softmax(scores, dim=-1)
top_k_vals, top_k_idx = torch.topk(probs, self.top_k, dim=-1)
top_k_vals = top_k_vals / (top_k_vals.sum(dim=-1, keepdim=True) + 1e-8)
output = torch.zeros_like(x)
for k in range(self.top_k):
for e in range(self.num_experts):
mask = (top_k_idx[:, k] == e)
if mask.any():
expert_out = self.experts[e](x[mask])
output[mask] += top_k_vals[mask, k:k+1] * expert_out
return output
class BenchBURTIMMA(nn.Module):
def __init__(self, hidden_dim: int, num_experts: int = 4, top_k: int = 1):
super().__init__()
self.activation = BenchSmoothLeaky()
self.memory = BenchCIFGMemory(hidden_dim)
self.router = BenchGatesRouter(hidden_dim, num_experts, top_k)
self.proj = nn.Linear(hidden_dim, hidden_dim)
self.norm = nn.LayerNorm(hidden_dim)
def forward(self, x: torch.Tensor) -> torch.Tensor:
# x: (batch, seq_len, hidden_dim)
batch, seq_len, dim = x.shape
h = self.activation(x)
# Flatten for routing
h_flat = h.reshape(batch * seq_len, dim)
routed = self.router(h_flat)
routed = routed.reshape(batch, seq_len, dim)
out = self.norm(self.proj(routed) + x)
return out
# ---------------------------------------------------------------------------
# Benchmark utilities
# ---------------------------------------------------------------------------
class BenchmarkResult:
def __init__(self, name: str, batch_size: int, hidden_dim: int,
mean_time_ms: float, std_time_ms: float,
throughput: float, peak_memory_mb: float):
self.name = name
self.batch_size = batch_size
self.hidden_dim = hidden_dim
self.mean_time_ms = mean_time_ms
self.std_time_ms = std_time_ms
self.throughput = throughput
self.peak_memory_mb = peak_memory_mb
def benchmark_forward(model: "nn.Module", x: "torch.Tensor",
num_warmup: int, num_runs: int,
device: str) -> Tuple[float, float, float]:
"""Benchmark forward pass. Returns (mean_ms, std_ms, peak_memory_mb)."""
# Warmup
with torch.no_grad():
for _ in range(num_warmup):
_ = model(x)
if device.startswith("cuda"):
torch.cuda.synchronize()
torch.cuda.reset_peak_memory_stats()
times = []
with torch.no_grad():
for _ in range(num_runs):
if device.startswith("cuda"):
torch.cuda.synchronize()
start = time.perf_counter()
_ = model(x)
if device.startswith("cuda"):
torch.cuda.synchronize()
end = time.perf_counter()
times.append((end - start) * 1000) # ms
peak_mem = 0.0
if device.startswith("cuda"):
peak_mem = torch.cuda.max_memory_allocated() / (1024 ** 2) # MB
return np.mean(times), np.std(times), peak_mem
def benchmark_backward(model: "nn.Module", x: "torch.Tensor",
num_warmup: int, num_runs: int,
device: str) -> Tuple[float, float, float]:
"""Benchmark backward pass. Returns (mean_ms, std_ms, peak_memory_mb)."""
# Warmup
for _ in range(num_warmup):
out = model(x)
loss = out.sum()
loss.backward()
model.zero_grad()
if device.startswith("cuda"):
torch.cuda.synchronize()
torch.cuda.reset_peak_memory_stats()
times = []
for _ in range(num_runs):
if device.startswith("cuda"):
torch.cuda.synchronize()
start = time.perf_counter()
out = model(x)
loss = out.sum()
loss.backward()
if device.startswith("cuda"):
torch.cuda.synchronize()
end = time.perf_counter()
model.zero_grad()
times.append((end - start) * 1000)
peak_mem = 0.0
if device.startswith("cuda"):
peak_mem = torch.cuda.max_memory_allocated() / (1024 ** 2)
return np.mean(times), np.std(times), peak_mem
def benchmark_memory_update(hidden_dim: int, num_warmup: int, num_runs: int,
device: str) -> Tuple[float, float]:
"""Benchmark CIFG memory update."""
mem = BenchCIFGMemory(hidden_dim).to(device)
key = torch.randn(hidden_dim, device=device)
value = torch.randn(hidden_dim, device=device)
for _ in range(num_warmup):
mem.update(key, value)
if device.startswith("cuda"):
torch.cuda.synchronize()
times = []
for _ in range(num_runs):
if device.startswith("cuda"):
torch.cuda.synchronize()
start = time.perf_counter()
mem.update(key, value)
if device.startswith("cuda"):
torch.cuda.synchronize()
end = time.perf_counter()
times.append((end - start) * 1000)
return np.mean(times), np.std(times)
def benchmark_routing(hidden_dim: int, batch_size: int, num_experts: int,
top_k: int, num_warmup: int, num_runs: int,
device: str) -> Tuple[float, float]:
"""Benchmark routing dispatch."""
router = BenchGatesRouter(hidden_dim, num_experts, top_k).to(device)
x = torch.randn(batch_size, hidden_dim, device=device)
with torch.no_grad():
for _ in range(num_warmup):
_ = router(x)
if device.startswith("cuda"):
torch.cuda.synchronize()
times = []
with torch.no_grad():
for _ in range(num_runs):
if device.startswith("cuda"):
torch.cuda.synchronize()
start = time.perf_counter()
_ = router(x)
if device.startswith("cuda"):
torch.cuda.synchronize()
end = time.perf_counter()
times.append((end - start) * 1000)
return np.mean(times), np.std(times)
# ---------------------------------------------------------------------------
# Main
# ---------------------------------------------------------------------------
def print_table(results: List[BenchmarkResult]):
"""Print benchmark results as formatted table."""
header = f"{'Benchmark':<25} {'Batch':<7} {'Hidden':<7} {'Mean(ms)':<10} {'Std(ms)':<9} {'Throughput':<14} {'Memory(MB)':<10}"
print("\n" + "=" * len(header))
print(header)
print("-" * len(header))
for r in results:
throughput_str = f"{r.throughput:.1f} samp/s"
mem_str = f"{r.peak_memory_mb:.1f}" if r.peak_memory_mb > 0 else "N/A"
print(f"{r.name:<25} {r.batch_size:<7} {r.hidden_dim:<7} "
f"{r.mean_time_ms:<10.3f} {r.std_time_ms:<9.3f} "
f"{throughput_str:<14} {mem_str:<10}")
print("=" * len(header))
def main():
parser = argparse.ArgumentParser(description="BURT-IMMA Benchmark Suite")
parser.add_argument("--device", type=str, default="cpu",
help="Device to benchmark on (cpu or cuda)")
parser.add_argument("--num-warmup", type=int, default=10,
help="Number of warmup iterations (default: 10)")
parser.add_argument("--num-runs", type=int, default=100,
help="Number of benchmark iterations (default: 100)")
args = parser.parse_args()
if not HAS_TORCH:
print("ERROR: PyTorch is required for benchmarks.")
sys.exit(1)
device = args.device
if device.startswith("cuda") and not torch.cuda.is_available():
print("WARNING: CUDA not available, falling back to CPU")
device = "cpu"
print(f"BURT-IMMA Benchmark Suite")
print(f"Device: {device}")
print(f"Warmup: {args.num_warmup}, Runs: {args.num_runs}")
if device.startswith("cuda"):
print(f"GPU: {torch.cuda.get_device_name(0)}")
batch_sizes = [1, 8, 32, 64]
hidden_dims = [64, 128, 256, 512]
seq_len = 16
num_experts = 4
top_k = 1
results: List[BenchmarkResult] = []
# Forward pass benchmarks
print("\n--- Forward Pass ---")
for bs in batch_sizes:
for hd in hidden_dims:
model = BenchBURTIMMA(hd, num_experts, top_k).to(device)
model.eval()
x = torch.randn(bs, seq_len, hd, device=device)
mean_ms, std_ms, peak_mem = benchmark_forward(
model, x, args.num_warmup, args.num_runs, device
)
throughput = (bs * seq_len) / (mean_ms / 1000.0)
results.append(BenchmarkResult(
"forward", bs, hd, mean_ms, std_ms, throughput, peak_mem
))
# Backward pass benchmarks
print("--- Backward Pass ---")
for bs in batch_sizes[:2]: # Fewer configs for backward
for hd in hidden_dims[:2]:
model = BenchBURTIMMA(hd, num_experts, top_k).to(device)
model.train()
x = torch.randn(bs, seq_len, hd, device=device, requires_grad=True)
mean_ms, std_ms, peak_mem = benchmark_backward(
model, x, args.num_warmup, args.num_runs, device
)
throughput = (bs * seq_len) / (mean_ms / 1000.0)
results.append(BenchmarkResult(
"backward", bs, hd, mean_ms, std_ms, throughput, peak_mem
))
# Memory update benchmarks
print("--- Memory Update ---")
for hd in hidden_dims:
mean_ms, std_ms = benchmark_memory_update(
hd, args.num_warmup, args.num_runs, device
)
throughput = 1.0 / (mean_ms / 1000.0)
results.append(BenchmarkResult(
"memory_update", 1, hd, mean_ms, std_ms, throughput, 0.0
))
# Routing benchmarks
print("--- Routing Dispatch ---")
for bs in batch_sizes:
for hd in hidden_dims[:2]:
mean_ms, std_ms = benchmark_routing(
hd, bs, num_experts, top_k,
args.num_warmup, args.num_runs, device
)
throughput = bs / (mean_ms / 1000.0)
results.append(BenchmarkResult(
"routing", bs, hd, mean_ms, std_ms, throughput, 0.0
))
# Print results
print_table(results)
if __name__ == "__main__":
main()
|