Note
Click here to download the full example code
Layer NormalizationΒΆ
Out:
layer-norm-backward:
N Triton Torch Apex
0 1024.0 361.411758 99.497980 307.200008
1 1536.0 405.098894 134.540150 338.201833
2 2048.0 491.520012 158.554837 321.254900
3 2560.0 461.954908 182.314537 328.556154
4 3072.0 522.893629 191.999993 315.076914
5 3584.0 551.384634 208.271186 310.527060
6 4096.0 564.965515 219.919464 301.546004
7 4608.0 502.690905 233.316456 288.751954
8 5120.0 527.381977 240.469672 282.482762
9 5632.0 542.843364 241.371422 287.591490
10 6144.0 546.133354 250.349744 286.322318
11 6656.0 527.207907 255.182111 284.748652
12 7168.0 505.976473 256.000002 280.182402
13 7680.0 483.779539 264.447629 279.272719
14 8192.0 463.698115 265.327937 277.694924
15 8704.0 413.655443 263.757583 282.673891
16 9216.0 427.822068 270.065931 286.879380
17 9728.0 438.033784 279.942444 288.593329
18 10240.0 444.412281 285.435547 286.767793
19 10752.0 430.797982 246.699797 289.616170
20 11264.0 427.746848 244.205952 283.966395
21 11776.0 421.198220 249.227509 288.097854
22 12288.0 417.131525 253.578674 294.323369
23 12800.0 412.072450 253.465340 288.721817
24 13312.0 409.075539 253.763296 291.769875
25 13824.0 404.112047 257.190689 292.313649
26 14336.0 394.568805 252.246330 285.530296
27 14848.0 383.999990 257.479779 289.012175
28 15360.0 378.869469 260.338991 290.496454
29 15872.0 370.552519 263.800544 292.796308
import torch
import triton
import triton.language as tl
try:
# This is https://github.com/NVIDIA/apex, NOT the apex on PyPi, so it
# should not be added to extras_require in setup.py.
import apex
HAS_APEX = True
except ModuleNotFoundError:
HAS_APEX = False
# Forward Pass
@triton.jit
def _layer_norm_fwd_fused(X, Y, W, B, M, V, stride, N, eps,
BLOCK_SIZE: tl.constexpr):
# position of elements processed by this program
row = tl.program_id(0)
cols = tl.arange(0, BLOCK_SIZE)
mask = cols < N
# offset data pointers to start at the row of interest
X += row * stride
Y += row * stride
# load data and cast to float32
x = tl.load(X + cols, mask=mask, other=0).to(tl.float32)
# compute mean
mean = tl.sum(x, axis=0) / N
# compute std
xmean = tl.where(mask, x - mean, 0.)
var = tl.sum(xmean * xmean, axis=0) / N
rstd = 1 / tl.sqrt(var + eps)
xhat = xmean * rstd
# write-back mean/rstd
tl.store(M + row, mean)
tl.store(V + row, rstd)
# multiply by weight and add bias
w = tl.load(W + cols, mask=mask)
b = tl.load(B + cols, mask=mask)
y = xhat * w + b
# write-back
tl.store(Y + cols, y, mask=mask)
# Backward pass (DX + partial DW + partial DB)
@triton.jit
def _layer_norm_bwd_dx_fused(DX, DY, DW, DB, X, W, B, M, V, Lock, stride, N, eps,
GROUP_SIZE_M: tl.constexpr, BLOCK_SIZE_N: tl.constexpr):
# position of elements processed by this program
row = tl.program_id(0)
cols = tl.arange(0, BLOCK_SIZE_N)
mask = cols < N
# offset data pointers to start at the row of interest
X += row * stride
DY += row * stride
DX += row * stride
# offset locks and weight/bias gradient pointer
# each kernel instance accumulates partial sums for
# DW and DB into one of GROUP_SIZE_M independent buffers
# these buffers stay in the L2, which allow this kernel
# to be fast
lock_id = row % GROUP_SIZE_M
Lock += lock_id
Count = Lock + GROUP_SIZE_M
DW = DW + lock_id * N + cols
DB = DB + lock_id * N + cols
# load data to SRAM
x = tl.load(X + cols, mask=mask, other=0).to(tl.float32)
dy = tl.load(DY + cols, mask=mask, other=0).to(tl.float32)
w = tl.load(W + cols, mask=mask).to(tl.float32)
mean = tl.load(M + row)
rstd = tl.load(V + row)
# compute dx
xhat = (x - mean) * rstd
wdy = w * dy
xhat = tl.where(mask, xhat, 0.)
wdy = tl.where(mask, wdy, 0.)
mean1 = tl.sum(xhat * wdy, axis=0) / N
mean2 = tl.sum(wdy, axis=0) / N
dx = (wdy - (xhat * mean1 + mean2)) * rstd
# write-back dx
tl.store(DX + cols, dx, mask=mask)
# accumulate partial sums for dw/db
partial_dw = (dy * xhat).to(w.dtype)
partial_db = (dy).to(w.dtype)
while tl.atomic_cas(Lock, 0, 1) == 1:
pass
count = tl.load(Count)
# first store doesn't accumulate
if count == 0:
tl.atomic_xchg(Count, 1)
else:
partial_dw += tl.load(DW, mask=mask)
partial_db += tl.load(DB, mask=mask)
tl.store(DW, partial_dw, mask=mask)
tl.store(DB, partial_db, mask=mask)
# release lock
tl.atomic_xchg(Lock, 0)
# Backward pass (total DW + total DB)
@triton.jit
def _layer_norm_bwd_dwdb(DW, DB, FINAL_DW, FINAL_DB, M, N,
BLOCK_SIZE_M: tl.constexpr, BLOCK_SIZE_N: tl.constexpr):
pid = tl.program_id(0)
cols = pid * BLOCK_SIZE_N + tl.arange(0, BLOCK_SIZE_N)
dw = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)
db = tl.zeros((BLOCK_SIZE_M, BLOCK_SIZE_N), dtype=tl.float32)
for i in range(0, M, BLOCK_SIZE_M):
rows = i + tl.arange(0, BLOCK_SIZE_M)
mask = (rows[:, None] < M) & (cols[None, :] < N)
offs = rows[:, None] * N + cols[None, :]
dw += tl.load(DW + offs, mask=mask, other=0.)
db += tl.load(DB + offs, mask=mask, other=0.)
sum_dw = tl.sum(dw, axis=0)
sum_db = tl.sum(db, axis=0)
tl.store(FINAL_DW + cols, sum_dw, mask=cols < N)
tl.store(FINAL_DB + cols, sum_db, mask=cols < N)
class LayerNorm(torch.autograd.Function):
@staticmethod
def forward(ctx, x, normalized_shape, weight, bias, eps):
# allocate output
y = torch.empty_like(x)
# reshape input data into 2D tensor
x_arg = x.reshape(-1, x.shape[-1])
M, N = x_arg.shape
mean = torch.empty((M, ), dtype=torch.float32, device='cuda')
rstd = torch.empty((M, ), dtype=torch.float32, device='cuda')
# Less than 64KB per feature: enqueue fused kernel
MAX_FUSED_SIZE = 65536 // x.element_size()
BLOCK_SIZE = min(MAX_FUSED_SIZE, triton.next_power_of_2(N))
if N > BLOCK_SIZE:
raise RuntimeError("This layer norm doesn't support feature dim >= 64KB.")
# heuristics for number of warps
num_warps = min(max(BLOCK_SIZE // 256, 1), 8)
# enqueue kernel
_layer_norm_fwd_fused[(M,)](x_arg, y, weight, bias, mean, rstd,
x_arg.stride(0), N, eps,
BLOCK_SIZE=BLOCK_SIZE, num_warps=num_warps)
ctx.save_for_backward(x, weight, bias, mean, rstd)
ctx.BLOCK_SIZE = BLOCK_SIZE
ctx.num_warps = num_warps
ctx.eps = eps
return y
@staticmethod
def backward(ctx, dy):
x, w, b, m, v = ctx.saved_tensors
# heuristics for amount of parallel reduction stream for DG/DB
N = w.shape[0]
GROUP_SIZE_M = 64
if N <= 8192: GROUP_SIZE_M = 96
if N <= 4096: GROUP_SIZE_M = 128
if N <= 1024: GROUP_SIZE_M = 256
# allocate output
locks = torch.zeros(2 * GROUP_SIZE_M, dtype=torch.int32, device='cuda')
_dw = torch.empty((GROUP_SIZE_M, w.shape[0]), dtype=x.dtype, device=w.device)
_db = torch.empty((GROUP_SIZE_M, w.shape[0]), dtype=x.dtype, device=w.device)
dw = torch.empty((w.shape[0],), dtype=w.dtype, device=w.device)
db = torch.empty((w.shape[0],), dtype=w.dtype, device=w.device)
dx = torch.empty_like(dy)
# enqueue kernel using forward pass heuristics
# also compute partial sums for DW and DB
x_arg = x.reshape(-1, x.shape[-1])
M, N = x_arg.shape
_layer_norm_bwd_dx_fused[(M,)](dx, dy, _dw, _db, x, w, b, m, v, locks,
x_arg.stride(0), N, ctx.eps,
BLOCK_SIZE_N=ctx.BLOCK_SIZE,
GROUP_SIZE_M=GROUP_SIZE_M,
num_warps=ctx.num_warps)
grid = lambda meta: [triton.cdiv(N, meta['BLOCK_SIZE_N'])]
# accumulate partial sums in separate kernel
_layer_norm_bwd_dwdb[grid](_dw, _db, dw, db, GROUP_SIZE_M, N,
BLOCK_SIZE_M=32,
BLOCK_SIZE_N=128)
return dx, None, dw, db, None
layer_norm = LayerNorm.apply
def test_layer_norm(M, N, dtype, eps=1e-5, device='cuda'):
# create data
x_shape = (M, N)
w_shape = (x_shape[-1], )
weight = torch.rand(w_shape, dtype=dtype, device='cuda', requires_grad=True)
bias = torch.rand(w_shape, dtype=dtype, device='cuda', requires_grad=True)
x = -2.3 + 0.5 * torch.randn(x_shape, dtype=dtype, device='cuda')
dy = .1 * torch.randn_like(x)
x.requires_grad_(True)
# forward pass
y_tri = layer_norm(x, w_shape, weight, bias, eps)
y_ref = torch.nn.functional.layer_norm(x, w_shape, weight, bias, eps).to(dtype)
# backward pass (triton)
y_tri.backward(dy, retain_graph=True)
dx_tri, dw_tri, db_tri = [_.grad.clone() for _ in [x, weight, bias]]
x.grad, weight.grad, bias.grad = None, None, None
# backward pass (torch)
y_ref.backward(dy, retain_graph=True)
dx_ref, dw_ref, db_ref = [_.grad.clone() for _ in [x, weight, bias]]
# compare
triton.testing.assert_almost_equal(y_tri, y_ref)
triton.testing.assert_almost_equal(dx_tri, dx_ref)
triton.testing.assert_almost_equal(db_tri, db_ref, decimal=1)
triton.testing.assert_almost_equal(dw_tri, dw_ref, decimal=1)
@triton.testing.perf_report(
triton.testing.Benchmark(
x_names=['N'],
x_vals=[512 * i for i in range(2, 32)],
line_arg='provider',
line_vals=['triton', 'torch'] + (['apex'] if HAS_APEX else []),
line_names=['Triton', 'Torch'] + (['Apex'] if HAS_APEX else []),
styles=[('blue', '-'), ('green', '-'), ('orange', '-')],
ylabel='GB/s',
plot_name='layer-norm-backward',
args={'M': 4096, 'dtype': torch.float16, 'mode': 'backward'}
)
)
def bench_layer_norm(M, N, dtype, provider, mode='backward', eps=1e-5, device='cuda'):
# create data
x_shape = (M, N)
w_shape = (x_shape[-1], )
weight = torch.rand(w_shape, dtype=dtype, device='cuda', requires_grad=True)
bias = torch.rand(w_shape, dtype=dtype, device='cuda', requires_grad=True)
x = -2.3 + 0.5 * torch.randn(x_shape, dtype=dtype, device='cuda')
dy = .1 * torch.randn_like(x)
x.requires_grad_(True)
# utility functions
if provider == 'triton':
y_fwd = lambda: layer_norm(x, w_shape, weight, bias, eps)
if provider == 'torch':
y_fwd = lambda: torch.nn.functional.layer_norm(x, w_shape, weight, bias, eps)
if provider == 'apex':
apex_layer_norm = apex.normalization.FusedLayerNorm(w_shape).to(x.device).to(x.dtype)
y_fwd = lambda: apex_layer_norm(x)
# forward pass
if mode == 'forward':
gbps = lambda ms: 2 * x.numel() * x.element_size() / ms * 1e-6
ms, min_ms, max_ms = triton.testing.do_bench(y_fwd, rep=500)
# backward pass
if mode == 'backward':
gbps = lambda ms: 3 * x.numel() * x.element_size() / ms * 1e-6
y = y_fwd()
ms, min_ms, max_ms = triton.testing.do_bench(lambda: y.backward(dy, retain_graph=True),
grad_to_none=[x], rep=500)
return gbps(ms), gbps(max_ms), gbps(min_ms)
bench_layer_norm.run(save_path='.', print_data=True)
Total running time of the script: ( 2 minutes 13.338 seconds)